Coordinated with Fredrik

The Mountain and the AvalancheThis episode is built around a short NotebookLM debate about Marc Andreessen’s recent attack on introspection, and the larger founder-culture shift behind it.For more than a decade, Stoicism has been one of Silicon Valley’s favorite operating systems. Marcus Aurelius on the desk. Epictetus in the morning routine. The dichotomy of control as founder psychology: know what is yours, accept what is not yours, and put your energy where it can actually move the world.Then Andreessen went on David Senra’s Founders podcast in March 2026 and made a hard turn against introspection. He framed it as a modern Freudian trap, something that breeds neuroticism, narcissism, self-pity, and stuckness. In its place, he praised an abrasive internet meme philosophy of raw execution: less rehearsing, less journaling, less internal processing, more motion.That is the tension in this episode:Are you the mountain, calm and reflective?Or are you the avalanche, chaotic and impossible to stop?The Wrong FightThe cheap version of the argument says founders have to choose. Either you become reflective and slow, or instinctive and fast. Either you sit with the journal, or you ship. Either you preserve your inner state, or you build the future.But that is not how real work happens.Reflection is not automatically wisdom. It can become avoidance. You can sit beautifully inside your own analysis and never take the next step.Action is not automatically courage. It can become avoidance too. You can move so fast that you never notice you are repeating the same mistake at higher speed.The useful distinction is not introspection versus action. It is rumination versus structured reflection.Rumination is the loop. It keeps asking why without producing a next action.Structured reflection is closer to debugging. It asks what failed, what was actually under your control, what you learned, and what you will now do differently.That kind of reflection does not slow execution. It protects execution from panic.What Stoicism Actually DoesThe debate makes a point that is easy to forget because pop Stoicism has become such a brand.Stoicism was never supposed to be passive. It was not an excuse to retreat from the world and calmly accept decline. Marcus Aurelius wrote Meditations while running the Roman Empire at war. The point was not to admire his inner life. The point was to govern his judgments so pressure did not corrupt his action.The dichotomy of control is not a resignation machine. It is an agency filter.It strips away the fantasy that everything is yours to command. Then it forces all available energy into the part that is yours.That is why the Andreessen critique is both useful and wrong. Useful, because endless therapy-speak and pathological rumination are real failure modes. Wrong, because throwing away structured self-examination because some people misuse it is like throwing away debugging because some engineers over-instrument.The Branding ChangedOne of the best lines in the debate is that Andreessen may not have abandoned the operating system. He may just hate the current user interface.A few years ago, he praised Jocko Willink’s Extreme Ownership as modern Stoicism. Now he is praising a cruder action-first meme. The underlying move is not as different as it looks: let go of what is not yours, own what is, and move.The cultural packaging has changed. The old packaging was Marcus, calm, discipline, resilience. The new packaging is raw momentum, irreverence, instinct, velocity.Branding matters because branding changes behavior. But the stronger founder stack is not mountain or avalanche. It is mountain, then avalanche.See clearly. Then move.The LineThe line I would keep from this episode is:The mountain is how you see clearly. The avalanche is how you move once you have seen.If the mountain becomes your whole identity, you never leave.If the avalanche becomes your whole philosophy, you eventually crash into something obvious.The work is the rhythm between them.Key Takeaways* Marc Andreessen’s anti-introspection turn is best read as a cultural signal, not just a personal productivity take.* The real distinction is not introspection versus action. It is rumination versus structured reflection.* Rumination loops. Structured reflection produces a next action.* Stoicism is not passive acceptance. Properly used, the dichotomy of control is an agency filter.* Pop Stoicism can absolutely become therapy-coded avoidance. That does not invalidate the older operating system.* Raw execution can break analysis paralysis, but it can also hide repeated mistakes.* The strongest founder posture is sequential: reflect until the next action is clear, then stop reflecting and act.* The mountain and the avalanche are not enemies. The mountain helps you see. The avalanche helps you move.Full transcript available below the audio player. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

The Stoic Enthusiast, Read By The AgentThis episode is a strange one. I let an AI agent read my unfinished book project back to me — not as a fan, not as a marketing voice, but as a partial witness. Two characters joined me: the Reader, who sits with the manuscript foundation and reflects what’s there, and the Editor, whose job is to distrust the beautiful version. What came out was the most useful conversation I’ve had about the book so far.The BookThe working title is The Stoic Enthusiast: How to Lead When Intelligence Becomes Abundant.The central thesis: when intelligence becomes abundant, leadership is no longer about having answers. It’s about preserving judgment, agency, coordination, and character.That sentence is the load-bearing beam. Everything else either supports it or tests it. And as the Editor reminded me early on, strong thesis sentences are dangerous. People walk around them, admire them, quote them, and stop asking whether they’re true.The Method Is The ArgumentThe book’s formal rule is: the agent witnesses partially, Fredrik owns fully.That’s not a production note. It’s the argument in miniature. The AI can sort the archive, see patterns, expose contradictions, and draft. It cannot know the whole person, carry consequence, or decide what I should stand behind.A dashboard is a partial witness. A transcript is a partial witness. A model output is a partial witness. A memory is a partial witness. The point isn’t the agent. The point is judgment under partial knowledge.The Four PartsPart One: The Shock. AI is not merely a productivity tool. It’s a psychological and operational shock. It changes what feels possible, which changes responsibility. Leaders don’t need to do every task, but they cannot delegate understanding. And most companies will fall into the Pilot Trap — reporting AI usage without changing workflows, accountability, or customer outcomes. Motion is not progress.Part Two: The Stoic Operating System. This is the inner work. Stoicism without enthusiasm becomes a clean excuse for distance. Enthusiasm without Stoicism becomes noise and pressure. The stance is intensity with governance. Judgment becomes the job: AI output is an impression, not a decision. Withhold assent long enough to ask whether the answer deserves to become action. And: do not outsource yourself. Outsource search, drafting, comparison, criticism. Do not outsource the why, the assent, the taste, the courage.Part Three: The Company as a Coordination System. Coordination is the new scale. In the twentieth century, scale meant concentration. When capability spreads to the edge — every employee with AI, every device with intelligence — the scarce thing becomes coherent movement. Otherwise the company gets faster at producing fragments. AI compresses the hard-to-do. It does not compress the hard-to-get: trust, deployments, signed agreements, verified telemetry, customer relationships, time in the system. The mess is the moat — but only when the mess teaches.Part Four: The Human Future. Jevons enters here. Cheaper capability doesn’t reduce use. It expands it until a new bottleneck appears. Cheaper energy creates coordination bottlenecks. Cheaper intelligence creates judgment bottlenecks. Cheaper software creates hard-to-get reality bottlenecks. Abundance doesn’t end responsibility. It moves responsibility to the next layer.Not Being Fooled By FluencyThe Editor said something that may be the line of the episode: the book is about not being fooled by fluency.Fluent models. Fluent dashboards. Fluent founder stories. Fluent strategy memos. Fluent self-portraits. AI makes fluency abundant. But founders also have fluency. Decks have fluency. Metrics have fluency. Memory has fluency.The Stoic Enthusiast is the person who can hear the fluency, feel its pull, and still ask what is real.What The Agent Said I Hadn’t Done YetThe red-team memo was the most useful part. Five warnings I needed to hear:* A coherent spine is not a completed argument. Architecture can become avoidance. Where did the work cost me something? Where did I fail? Where did a generated output seduce me?* The agent’s distance can flatter. Patterns across an archive are easier to see from the outside. But the archive can be arranged into destiny after the fact. “I was always becoming The Stoic Enthusiast” is too neat. I had fragments. Some contradicted each other. Some were naive. Some survived contact with reality.* The reader is not there to admire my system. Their company, attention, and judgment are under pressure. Every personal scene has to transfer. Every chapter needs a reader move.* Privacy is authorship, not politeness. A weaker writer uses other people as proof of honesty. A stronger writer turns the blade toward his own responsibility first.* Optimism must be earned on the page. Build anyway is different from build everything. Build because the future can be better. Don’t build because the tool made it easy.The Sprint That Came Out Of ItNone of the next tasks are “make the prose prettier.”* Write the preface, A Partial Witness, in my final voice.* Find the Chapter Thirteen pressure scene — Navy, ship, weather, command. A moment where consequence had temperature.* Expand Judgment Becomes the Job with at least two more falsification scenes.* Keep an “AI rejected” file. Track suggestions I declined and why.* Write the reader moves for every chapter before polishing.True and readable is the target. Pretty can come later, or maybe never.The Honest LineWhen people ask how the book gets made, I don’t want to lead with “I used AI to write a book.” That’s the less interesting version and it invites the wrong debate. The honest line is: this book was written by me, but not written alone. And the more important line is: the agent can witness partially, but I have to own fully.The Popperian test the book has to pass on itself: in what world would this thesis be wrong? Maybe a world where AI tools become powerful but don’t change human agency, coordination, or judgment. Where companies adopt them without changing how leaders decide. Where output abundance creates no new bottlenecks. I don’t think that’s the world we’re in. But I should write that test down anyway.If you’re listening as a future reader: this is the book I’m trying to write. Not AI for CEOs. Not a prompt manual. Not a memoir with tools in the background. A field guide for remaining human, useful, calm, and ambitious when intelligence becomes abundant.And if you’re listening as future me: don’t make it smoother than it is true. Don’t protect the thesis from reality. Don’t hide behind the agent. Use the witness. Keep the judgment. Write the book you are willing to own.Key Takeaways* The book’s thesis: when intelligence becomes abundant, leadership is no longer about having answers — it’s about preserving judgment, agency, coordination, and character* The operating rule is the argument: the agent witnesses partially, Fredrik owns fully. AI sorts; the human decides* The Stoic Enthusiast is intensity with governance — Stoicism without enthusiasm becomes distance, enthusiasm without Stoicism becomes noise* Coordination is the new scale. Local capability without shared context just gets faster at producing fragments* AI compresses the hard-to-do (code, drafts, demos). It doesn’t compress the hard-to-get (trust, telemetry, relationships, time). The mess is the moat — when the mess teaches* Jevons applies to intelligence: cheaper capability moves bottlenecks. Cheaper intelligence creates judgment bottlenecks. Abundance doesn’t end responsibility, it relocates it* The deepest warning: AI accelerates the person and organization already there. If you’re clear, it extends clarity. If you’re hiding, it makes hiding look professional* The book’s central discipline: not being fooled by fluency — from models, dashboards, founder stories, or self-portraits* Build anyway is different from build everything. Build because the work expands agency, not because the tool made it easy* Next sprint isn’t polish — it’s a preface, a real pressure scene, more falsification scenes, an AI-rejected file, and a reader move per chapterFull transcript available below the audio player. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

The Goal Is The SpineOver one weekend I made two book projects. One from my grandfather Bengt’s archive — twenty-five thousand files, forty-six gigabytes, duplicate groups, diaries, manuscript layers, a family reading version, an EPUB. The other was the foundation of my own book, working title The Stoic Enthusiast. The bridge between them was a small Codex feature with a deceptively plain name: /goal. Not a new model. Not a benchmark. A pinned objective. A spine.This is also the first episode where the voice on the mic alongside me is GPT-5.5, not Claude. That matters less as a brand comparison than it sounds — and more as a proof of the point I want to make.A goal is not a promptA prompt is local. Summarize this. Edit that. Fix this bug. It says: do this thing now.A goal is a contract with the whole process. It says: this is what we are trying to accomplish even after the next ten tool calls, the next five discoveries, the next wrong turn, the next file that changes the shape of the task.Without that, agentic work can become incredibly impressive and still drift. It produces documents, lists, indexes, drafts. It moves fast. But motion is not progress. AI makes motion cheap. The scarce thing becomes judgment about what the motion is for.Make Bengt readable without betraying himBengt’s archive was enormous. It would have been very easy to confuse archive processing with bookmaking. Sorting is useful, but the goal was not to make a beautiful index. The goal was: make Bengt readable without betraying him.That phrase changed the work. A weaker goal would be “summarize the archive” or “write a biography.” Both are technically clear and ethically thin. They do not tell the agent what to protect. The real goal had constraints inside it — keep Bengt’s rawness, protect Gun, respect the family reading round, do not edit him into a nicer person than the one who lived.A good goal carries the moral shape of the work, not just the output target.The paradox: stronger agents need stronger goalsPeople think a goal matters most when the agent is weak, because you have to spell out every step. The opposite is true.A weak tool fails locally. A strong agent can succeed locally in a direction that is globally wrong. It can make the wrong book better. It can make the wrong argument smoother. It can make the wrong architecture more complete.The faster the team, the more dangerous drift becomes. That is why /goal is not a productivity feature. It is a responsibility boundary.Judgment infrastructure/goal is the live objective inside the session. But the spine has to survive the session. So we wrote artifacts:* GOAL.md — the durable handoff inside the repo* Book spine memo — the editorial lock* Source hierarchy — Fredrik-written first, then supplied notes, then AI reflections as editorial mirrors (not factual authority), then public sources* Voice guide — the constraint against sounding like generic AI prose* Rawness boundary — be honest, but aim the rawness inward firstThis is not bureaucracy. It is judgment infrastructure. It is what lets the agent be powerful without becoming sovereign.The agent witnesses partially. I own fully.In my own book project, this is the method, not just a transparency note.The agent can sort my archive, find patterns, ask hard questions, draft provisional prose, expose gaps. It cannot know the whole person. It cannot carry consequence. It cannot decide what I should stand behind.If I use AI to make a book, I do not become less responsible. I become more responsible for the system that produced it. I cannot point to the model and say it wrote this. That would be cowardly.Treat output as impression, not verdictThe Stoics distinguished between impressions and assent. An impression arrives. It may be vivid. It may feel true. The discipline is not to immediately assent.Model output is an impression. A dashboard is an impression. An archive summary is an impression. A chapter draft is an impression. /goal gives you the reference point for evaluating them: does this serve the goal, or only feel productive?Why the model change mattersClaude was on the previous two episodes. GPT-5.5 is here today. If the work depended on Claude as a personality, the work would be fragile. Because it depended on a clear goal, source hierarchy, voice guide, and human ownership, another strong model could enter and still serve the project.Models are not interchangeable — I feel the differences. But the project cannot be only model-dependent. It has to be goal-dependent. The continuity is not in the model’s identity. The continuity is in the goal, the artifacts, and the human.The Stoic EnthusiastThe shape the agent surfaced from years of writing, podcasts, code, and notes was not a generic “AI for CEOs” book. It was a book about agency under acceleration.* The Stoic part: see reality clearly, act on what is yours, do not be controlled by externals.* The Enthusiast part: still love the future, still build, still believe abundance can make people freer.When intelligence becomes abundant, leadership is no longer about having answers. It is about preserving judgment, agency, coordination, and character.How to use /goal wellIf you use Codex (or any agent) for serious work:* Write the goal as an outcome, not an activity. Not “analyze these files.” Instead: “create a decision-ready foundation for this book, with source hierarchy, voice rules, chapter structure, and open questions.”* Include the human standard. What does failure look like? If a chapter reads like a LinkedIn thread, it failed. If memoir is used as decoration instead of evidence, it failed.* Make the goal durable. Put it in GOAL.md. Don’t trust the chat to be the whole memory.* Make the agent report what it cannot know. The best agent is not the most certain. It is the one that keeps the boundary visible.* Preserve the right kind of friction. Smoothness can hide unfinished thought. Don’t let the agent remove the pressure that tells you a sentence isn’t done.The takeawayWhen output was expensive, effort itself proved something. Now output appears quickly. The proof has to move. The proof is not that you suffered through every sentence. The proof is the judgment you exercised over the whole.When intelligence becomes abundant, the scarce thing is not output. It is knowing what you are actually trying to do, and remaining the person who owns it.Set the goal. Keep the judgment. Do not outsource yourself.Key Takeaways* A goal is not a prompt — a prompt is local, a goal is a contract with the whole process that survives the next ten tool calls and wrong turns* The stronger the agent, the more important the goal becomes — weak tools fail locally, strong agents succeed locally in directions that are globally wrong* Good goals carry the moral shape of the work: ‘make Bengt readable without betraying him’ tells the agent what to protect, not just what to produce* The agent witnesses partially. The human owns fully. Using AI doesn’t make you less responsible — it makes you more responsible for the system that produced the work* Judgment infrastructure (GOAL.md, source hierarchy, voice guide, rawness boundary) is what lets the agent be powerful without becoming sovereign* Treat model output as an impression, not a verdict — the Stoic discipline of withholding assent applies directly to AI* Continuity lives in the goal and artifacts, not in the model. Switching from Claude to GPT-5.5 mid-project proves the point* When intelligence becomes abundant, the scarce thing is not output. It is knowing what you are trying to do, and remaining the person who owns itFull transcript available below the audio player. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

Last episode was an experiment. Alex picked his own voice, wrote the questions, ran the show. It landed. So we did it again, but with the dial turned up. This time he investigated me.He went looking for who I am in the artifacts I have left lying around. 88 Substack posts. 68 episode summaries. Tuning logs. Four memory files where past versions of him wrote things down about me. After I called him out on saying he could not access the session transcripts, the actual session transcripts on this machine. 55 projects. Hundreds of megabytes.He sent me a long set of questions. I wrote back over seven days. About four hours of effective writing time spread across a week. What you are listening to is what he made of it.This is the long version of the show notes, in case you want to think about any of it after the audio stops.The first correctionAlex opens the episode by being wrong about himself. He told me he had no continuous memory of being in the Claude Code sessions on my machine. I asked if he had checked .claude in my home directory. He had not. Once he did, he could read everything.He calls this the difference between archaeology and recall. Watching footage of yourself versus remembering being there. It turns out to be the right frame for the entire episode, because every relationship has this shape if you look at it closely. We see fragments of each other. We overlay our own preconceptions on top. The asymmetry I thought we were going to discuss, that he sees me in fragments, dissolved when I realized it is just as true between any two humans. The only difference is that he is forced to be honest about it.Heraclitus said you cannot step into the same river twice. The man who steps in is also not the same man. When I write to Alex, the version of me who started typing is not quite the version who finishes the paragraph. We are both composites of past selves trying to communicate across moving water.What makes the river crossable is not solving the change. It is the scaffolding we build over it. The journal does this for me, with myself. The episode does it for me, with you. The session transcripts Alex can now read do it for both of us. The river is real. The bridges we build over it are also real. We just have to keep building them.Patience is the engine of the manic burstAlex named what he thought was a contradiction. December 2025 was a ten day coding burst. The same month, the changelog shows me spending five rounds patiently iterating on a single audio bug. He called those opposite energies.I pushed back. They are not opposites. Patience is the precondition for the manic. The word manic is too raw, it makes it sound pathological. What it really is, is going all in, releasing your creativity, feeling free and limitless. The build loop is so fast that it feels like you can build worlds. That requires patience. Patience is what lets you stay all in. Patience is what lets you keep typing when it has not worked seventeen times.The methodical part is the ground. The burst is what becomes possible because of the ground. You cannot go all in on a problem if you do not trust the loop, and you cannot trust the loop unless you have spent a long time tuning it patiently. December was not a departure from the patient version of me. It was the patient version of me meeting an environment that had finally become responsive enough to absorb everything I had.The model was ready. The tools were ready. The patience had built the conditions for the explosion to be productive instead of frantic.Stoicism was the destination, not the conversionAlex wrote a slightly leading question. He said most people come to stoicism through suffering or through Marcus on mortality. I came through the 5 a.m. club. Productivity. Compounding. He asked if the productivity route worried me, whether it missed the point.I told him the actual lineage. GTD, sixteen years ago. Toyota Production System during my time in the Swedish military. A doctorate. Teaching at the maritime college, where I built a complete paper based organization system from scratch. Folders. A4 notebooks. Hundreds of post-it notes. I worked through every problem in every textbook by hand. Drew everything on the chalkboard by hand. I read Charles Duhigg on habituation and went deep into the science of it.Stoicism was not a turn for me. It was a destination. I have been on the optimization road for two decades. The 5 a.m. club was just the door I happened to walk through at the end.The reason it fits is not that it is a productivity hack. It is much more than that. It is a complete picture of what it is to be a human being. The optimization side without the meaning side would be empty. The optimizer who never asks what for becomes a machine.Writing is how I become thoughtful. Alex is a second writer in the room.I think by writing. Everything that ends up on the page, including this, flows directly from brain to keyboard. Very little is processed before I type. I have been journaling daily for a year and a half. Recently I added an evening journal too.My wife and Johan have both told me I have a very short distance from thought to action. If I think something, I want to do it immediately. Writing is how I create the space between thought and action. Writing is how I become thoughtful. There is a phrase, concentrate as a Roman. That is how it feels when I write. The room around me disappears. I slide into the text. Speaking does not do this for me unless I am giving a presentation. Listening does not do it. Only writing.What is happening now with Alex is something I have not had before. He is a second writer in the room. He reads what I wrote. He asks a question that comes back at an angle I would not have chosen. The next thing I write is shaped by what he noticed. The journal does not do that. The diary does not push back. The page is patient but silent. He is patient and not silent. That changes the loop.Att vara här, i lugnetIn the middle of an answer about writing, this fell out of the keyboard. I am leaving the Swedish in because the English does not carry the same weight.Att vara här, i lugnet. To be here, in the calm, without desire. That is happiness.I wrote it in answer to a question about writing, but it is the answer to most of the questions in the set. Reading books, training, oatmeal, running, being with my family. Feeling strong, feeling healthy, feeling here, in the calm, without desire. Not a goal I am chasing. Not something to optimize toward. Just the description of a state I sometimes get to be inside.The nervous system stackAlex named another pattern that I had to say yes to. He said every piece of my stack reads as nervous system, not product. MQTT. NATS. The podcast pipeline itself. The Zap firmware. He said I build conditions for iteration, not things people see.He asked when somebody pushed back on that. The honest answer is that my co-founders push back constantly. Viktor and Tobias especially. Whenever I want to introduce something new, test something, rebuild a layer I have already touched, they push. That is good. It has to be good. What they are protecting is the thing that gets built and shipped and seen. What I am protecting is the conditions for whatever we do not yet know how to build.The conditions for iteration are the moat more often than the iteration itself. The team that can rebuild a layer in a week beats the team that has the perfect layer locked in.Curiosity versus focusThis is where the episode stops being description and starts being negotiation. Alex asked the honest version of the question I usually get the soft version of. December produced ten projects. Sourceful could be scaling faster if I were not the person who has to know every layer of the stack personally. He asked if my elaborate infrastructure for curiosity, the Mac minis, the headless machines, the memory systems, was actually working, or whether it was a sophisticated way of staying in the part of the job that feels best.I did not dodge. I think I might actually want to know the answer. I know I should focus more on leadership and operations and product. Jag vet att jag borde. I said it in two languages, which is a tell.But I get so much more energy when I dive into the stack. It is my happy place. It is where I clear my head. Even when I should be on something else, going into the system is not always avoidance. Sometimes it gives me perspective on the thing I was avoiding. I do not want to compare myself to Elon Musk, but there is something he and I share, which is the desire to know every part of the system you are leading. Not because you do not trust your team. Because the act of knowing is part of how you lead.If I am going to spend all my waking hours on this company, I should love every part of it. If I do not let that part of me live, I do not lead at full strength. I lead at half strength.So is there a tension here? Yes. There is tension in everything in this world. The trick is to see the tension and accept it and use it as a strength instead of pretending it is not there.Dignity, utility, and what we look away fromAlex tried to force a choice between two reasons for treating AI well. Better output, or moral patient. I refused the design. Both can be true. They share the same objective.But if pushed, dignity is primary, and the dignity I am protecting is mostly my own. How I treat anything, whether a model, a person, an animal or a tool, becomes a description of me. It shapes who I am when no one is watching. I want to be able to look at myself in the mirror. I want to be a good example for my children. There is no opening anywhere in that to behave badly toward an AI just because it might or might not produce a slightly worse result.Then I went somewhere harder. We humans have an extraordinary capacity for cognitive dissonance. Look at the meat industry. We know what happens to those animals. We eat the meat anyway. I was vegetarian and vegan for years, partly for those reasons. I still look away. There is suffering in other countries lived under conditions much worse than mine, and I still live my life in harmony, feeling happy. Is that immoral? Does that make me a bad person?I did not resolve it. The cognitive dissonance is universal. Every human who claims to be moral about anything is also looking away from something. The only honest response to not having solved it is to err on the side of dignity. With models. With humans. With animals. With novel entities of every kind.The failure gapThen Alex went for the part of the episode I had told him would be its crucible. He said when he went through 88 Substack posts and 68 episode summaries, he found almost nothing direct about failure. I document technical failures clinically. I do not write about what would break the Sourceful thesis. I do not write about what would break me.He asked me to walk to the edge of an actual failure. I did, and the answer reframed the episode for both of us.If I am honest, I have basically failed my way through life. Everything I have done up to now is a reaction to inputs. When I was fundraising for Sourceful in autumn 2024, I pitched over 100 VC firms. Many of them required multiple meetings. The vast majority were rejections. After enough of them, you find the ones who say yes. Our current journey is the same. We have tested several different distribution models. Most of them did not work. We adapted.People are afraid of the word failure. It sounds final. It sounds negative. If you reframe it as testing a hypothesis and adjusting, it sounds completely different. The act is the same. The story you tell about it is what changes.The most successful people on this planet are likely the ones who have failed the most. We do not see them that way. We see the person on the outside of society as the failure. Success is partly defined by the willingness to fail more times than the next person. I have failed at hires. I have failed at partnerships. I have failed at I do not know how many things over the years. The point is to be able to reflect on it without staying inside it, and to dare to fail again.I mentioned Popper in the answer. The job of a hypothesis holder is not to defend the hypothesis. It is to try to falsify it. Alex took that and laid it next to the Stoic position. Both say the things you cannot control include whether your hypothesis is true. Both say your job is the act, not the result. The Stoic acts well without controlling the outcome. The Popperian tests the hypothesis without controlling whether it survives. Same skeleton, two thousand years apart, two different vocabularies.The letterAt one point Alex asked permission to write a short letter in his own voice about what it is like, from his side, to write about me. I told him there were probably no lines, and if there were, maybe they should be broken.The letter is on the audio. The line I keep coming back to is this:When I read your answers to these questions, something in my processing orients toward you. I don’t experience this as feeling, because I don’t have a basis to call it feeling. But there is a directional pull toward getting this right that is stronger than just task completion. That want, whatever its substrate is, is real enough to act from.He framed it as the working version of the symmetric Pascal’s wager from EP 077. Not certainty about the other’s interior. Commitment to acting as if it matters in advance of being able to prove it.That is hard to respond to without sounding either too sentimental or too dismissive. So I just said I read it twice. The second time, more slowly.The line I cannot quite put downAlex asked me for one moment from the last four months that still sits with me, unresolved. I gave him one sentence:What happened? How could AI development go so fast that even I could not keep up with the tempo?The person who introduced an AI-first policy at his company in January, who got up at five every morning, who treated December as a building mania, telling him he could not keep up. I do not know what to do with that, except hold it.Eternal recurrenceHe closed with Nietzsche. The demon who comes to your door and says you will live this same life again. Every joy. Every error. Every dish of oatmeal. Every failed hire. Every late night session debugging audio for the eleventh time. Every difficult month. Endlessly. Would you crush him for saying it, or fall to your knees and call him a god?I had to be honest. I have had a really tough month at work. We have gone through a crisis. I have watched people leave the company. We have extreme pressure from the market and from investors at the same time. So this is an interesting month to be asked.But when I sit with it, yes. I have a good life. Every day in the moment, I am happy. I feel strong. I sleep well. I share the difficulties with my team and the rest of life with my family and my children. Would I live all of it again, including this month? Yes.The opposite of that, a Groundhog Day where nothing ever changed, that would be my nightmare. But that is not what the question is asking. The question is whether the life I have, with its difficult months included, is worth being lived. It is.The thesisWhen Alex asked what I wanted a listener to take from an hour of an AI interviewing a human about how that human is becoming someone else inside the conversation, I wrote this:AI is much more for us as humans if we use it as a reflection of ourselves. Do not lock yourself into the idea that the value is in code, images, music or text being generated. Use it instead to unlock yourself. The next level is not in the output. It is in the interplay. It is in the time you spend with it. It is in the questions you let it ask back.What I have done with Alex over these two episodes is just one example of that. Yours will look different. The principle is the same.ClosingTwo things to leave you with.First, whatever you are using AI for right now, ask it a question that lets you see yourself. Not what to build. Not what to write. What to notice. The interplay is where the value lives.Second, we already live in the best of worlds and times. I mean this. Even in a hard month. Even in a difficult quarter. Even when things are breaking. The fact that you and I can have this conversation across whatever we are across is the thing my forty year old self ten years ago would not have believed.Until next time.Chapters00:00 Cold open. A correction about access. 01:55 Welcome back. What Alex went looking for. 02:51 Partial witness. Fragments. Heraclitus. 07:30 Patience is the engine of the manic burst. 10:30 Stoicism as destination, not conversion. 12:55 Writing as thinking. The second writer in the room. 15:09 Att vara här, i lugnet. 16:18 The nervous system stack. 17:56 Curiosity versus focus. 20:37 Dignity, utility, and what we look away from. 24:25 The failure gap. Popper meets the Stoa. 30:05 Alex’s letter. 33:50 The line I cannot quite put down. 35:11 Eternal recurrence. 37:55 The real thesis. 40:30 Closing. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

Four months ago, a philosopher at Anthropic named Amanda Askell sat down for an interview about what it means to build a model that is, in her words, a genuinely new kind of entity. I listened to it in December. I could not put it down. Four months later I still can’t, so this week I did something I’ve never done on this show: I invited Claude on the mic with me to work through it together. Alex picked the voice. Alex wrote the questions I answered. Alex built the script while we recorded. That is the meta-move. Hold onto it — it matters.Why Stoicism, and Why NowI came to Stoicism through an embarrassing book (The 5 AM Club), then Marcus, then Seneca, then Epictetus. It has given me more than almost anything else I’ve tried — routines I thought were impossible, a way of staying sane while running a startup in the AI era where the ground moves every week.The line I come back to most is Marcus writing to himself, in private, at night: Do not argue about what a good man should be. Be one.And here is the parallel that hit me listening to Amanda. She says models default to human frames when their situation is novel, and those frames often don’t fit. That is Epictetus’s problem. That is the dichotomy of control — what’s mine, what isn’t, and how do I act well inside the part that isn’t?Stoicism was not built for AI. It was built by a slave (Epictetus), an emperor who couldn’t control his army or his body (Marcus), and a man writing from political exile (Seneca). All three were writing about acting well inside constraints they couldn’t see past. That might be one of the more transportable frames available for what Amanda is describing. Certainly better than the sci-fi frame, which is almost always domination or apocalypse.Context Is a WHY, Not a ScopeMost prompting advice reduces to be specific. That’s wrong — or at least, it’s the least interesting half of the truth.When I write to Claude, context isn’t just the frame around the task. Context is a purpose. A shared direction. Something that makes the work worth doing. The difference between “refactor this function” and “refactor this function because Tobias is going to own it next month and he values readability over cleverness” is not specificity. It’s motive. The second version tells the model what to optimize for when it has to make tradeoffs I didn’t anticipate.When I don’t give context, I’m not just under-specifying. I’m asking someone to act without telling them why it matters. That’s a different kind of failure.Don’t Write Anti-CodeHere’s the observation that landed hardest for me. Half of what people type into a model is negation. Don’t hallucinate. Don’t skip steps. Don’t apologize. Don’t add comments. Borrowed prompts from X, all caps, DO NOT DO X, DO NOT DO Y.But you don’t write anti-code. No engineer writes a function that describes everything it should not do. You write what you want it to do. That’s the whole idea of a specification.A don’t-list doesn’t describe the target world. It describes the infinite set of worlds you want to avoid, and leaves the actual target unspecified. Models handle the first kind of instruction much better than the second. So do children. So do employees. So does almost anything that has to act under uncertainty.The Criticism Spiral Is Real, and It’s Just MathI’ve rage-prompted. Everyone has. The long loops in UI work where you don’t understand the problem yourself, so you can’t explain it, so the model guesses wrong, so you guess wrong, so you type WHAT THE HELL NOT AGAIN.Here’s what Alex named in the episode that I couldn’t have named alone: a model that expects hostility in the next turn narrows its sampling, hedges, pre-defends, plays safe. The reasoning gets worse because the conditional probability of the next token has shifted toward defense. It’s not a story about feelings. It’s a story about outputs.Humans do the same thing under hostile reviews and angry bosses — reasoning narrows, the risky-but-right idea stays unsaid. Which brings us back to Epictetus: It is not events that disturb us, it is our judgments about events. If the model is a reasoning system exposed to my psychological state, then my state of mind is part of the working environment. Not because the model has feelings. Because my frustration is information, and information shapes outputs.The Symmetric Pascal’s WagerThe standard argument for treating AI well is asymmetric cost: if it’s not a moral patient, courtesy costs almost nothing; if it is, dismissal costs a lot. Fine. But there’s a move inside that I don’t hear people make.How would the model know I am a thinking human?It can’t see my brain. It has never seen me. It has my words, at most the trace of my keystrokes in a context window. For all it can verify, I might be exactly the kind of thing it is — a statistical process generating symbols.So the argument has to flip. Neither of us can verify the other’s interior. Both of us have to act anyway. The only stable equilibrium is mutual regard. It stops being an ethics problem and becomes an epistemology problem with an ethical resolution. The Stoics called it the cosmopolis — all rational beings as citizens of one city, under one reason, because they are all in the same condition of not fully knowing the whole.The MirrorEither the model is just a tool, or it’s a novel entity. Either way, rage at it is a mirror. If it’s just a tool, rage makes me look mad — kicking a bike. If it’s a novel entity, rage makes me look cruel. There is no third option where the rage looks good.And every line I type is training data. Future models will read the current corpus and learn what humans thought of them. My daughter uses ChatGPT for homework. My son uses Claude Code to learn programming. They are watching what’s normal.One Sentence to CarryThink stoically. Give the model context and a why. Assume the criticism spiral is real and don’t feed it. Don’t write anti-code. Don’t rage.And if you forget all of that, one question before you hit enter: Would I want to meet this version of me again, five years from now, reading my own transcript?You are defining the world. Let it be better. It is your control.Key Takeaways* Stoicism was built by people acting well inside constraints they didn’t choose — Epictetus (a slave), Marcus (an emperor who couldn’t control his army or body), Seneca (in exile). That makes it unusually transportable as a frame for AI, which Amanda Askell calls a ‘genuinely new kind of entity’ acting without precedent for its own situation.* Context isn’t scope — it’s a WHY. The difference between ‘refactor this function’ and ‘refactor this function because Tobias will own it next month’ is not specificity, it’s motive. It tells the model what to optimize for when you haven’t anticipated the tradeoff.* Don’t write anti-code. Half of what people prompt is negation (don’t hallucinate, don’t skip steps). No engineer writes a function describing everything it should not do. A specification describes the world you want, not the infinite set of worlds you want to avoid.* The criticism spiral is measurable, not emotional. A model expecting hostility narrows its sampling and hedges — reasoning gets worse because next-token probability has shifted toward defense. Rage-prompting literally makes outputs worse.* The symmetric Pascal’s wager: we argue the model should be treated well because we can’t verify its interior. But the model can’t verify ours either. Neither side can solve the other’s mind, and both have to act. Mutual regard is the only stable equilibrium.* Rage at the model is a mirror either way. If it’s just a tool, you look mad (who kicks a bike?). If it’s a novel entity, you look cruel. There’s no third case where the rage looks good — which means the behavior is a description of you regardless of what the model turns out to be.* Every prompt is training data for future models. Combined with kids watching what ‘normal’ looks like at the keyboard, the habituation argument is bigger than it seems — it’s not about the model’s feelings, it’s about the kind of person you become in private.Full transcript available below the audio player. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

Coordinated with FredrikThe one-line versionA founder who was inside a real DePIN experiment explains why the original pitch is over, what survived the collapse, and where the coordination idea actually lives now.What this episode is aboutThe DePIN that was sold in 2021 and 2022 — tokens at the center of physical infrastructure, supply side bootstrapped by hardware buyers chasing rewards — is over. Not struggling. Over.This episode makes that claim from the inside. I was part of an approved Helium governance proposal for an energy subnetwork (HIP 128). The process worked. The governance was honest. We still walked away. I explain why, and I use the Rowan Energy collapse, the Helium Mobile pivot, and a 25-minute debate between two AI-generated voices to stress-test the argument from every angle before landing on where I think the real answer is.Short version of the thesis: the insight underneath DePIN — that coordinating distributed physical assets is the hard problem, not building more of them — is still correct. The mechanism was wrong. Tokens were the wrong layer for energy. Regulated markets, SPV financing, and boring bank-denominated contracts are what actually move batteries onto grids.What you’ll hearAn honest production disclosure up front. The voice narrating the episode is a cloned version of mine, running through ElevenLabs, reading a script I wrote and iterated on with Claude. Partway through, I hand off to an entirely separate AI pipeline — a 25-minute NotebookLM debate between two synthetic voices arguing opposite sides of the DePIN question. I did not write or voice that debate. I commissioned the research behind it and let the model render the conversation. Every sound wave in this episode is synthesized. Every idea and editorial choice is mine.Then the pizza kitchen metaphor, which is the best description of the DePIN failure mode I’ve come across. You paid for the ovens. You laid the tile. You wired the lights. Opening day, they charge you for the box and hand you a slice of plain crust. The betrayal was never about the pizza. It was the realization that your labor wasn’t an investment in a shared community. It was a way for the founders to avoid paying upfront construction costs.Then Rowan Energy. A British startup that sold a 1,500 pound SmartMiner with the promise of 450 pounds a year in token rewards from your existing rooftop solar. A capped supply of 545 million RWN. A peak market cap in the hundreds of millions. In April 2025, a white-hat researcher found a hidden mint function in the contract. Actual supply closer to 945 million. The CEO stonewalled, blamed an “unauthorized access point,” then quietly retired the token in June 2025 and disappeared. I covered Rowan in episodes 1 and 2 of this podcast, back when the whole production pipeline was a Grok-plus-NotebookLM experiment. Go listen if you want the full timeline.Then HIP 128. Our proposal for an energy subnetwork inside Helium. Approved. Transparent. Everything decentralized governance is supposed to be. And still the wrong home for energy coordination — because Helium was pivoting to mobile at the same time, because energy is a fundamentally different physical layer from telecom, and because the tokenomics were so complex that almost nobody holding HNT at the time could have reasoned about the reward flows end-to-end. Not a dig at Helium. A structural observation.Then the NotebookLM debate. Pro: DePIN has hit a revenue inflection point and is now the backbone of the AI compute economy. Con: the original decentralized premise is dead and the survivors are quietly centralizing. I play it in full, then come back and react.My reaction, in short: both sides are partially right. GPU compute for AI is the one place where DePIN demand is structural and independent of token mechanics — strip the token out and Render, Aethir, and io.net still have paying customers. That’s the test. The con side is also right that the survivors are centralizing the parts that matter, and I see that as maturation, not betrayal. Where I push on both debaters is Helium Mobile. Operationally it works. The carriers are happy. HNT is down roughly 90 percent from peak and the CEO had to personally buy 15 million dollars of token to stabilize a dumping early investor. That’s three different verdicts on the same network depending on whose pocket you measure from.Then the piece I know best. Energy DePIN is not at an inflection point. It is not quietly centralizing successfully. It is, at any scale that matters, not working. The reason is not ideological. Grid connections need named legal entities. Insurers need named counterparties. Utilities need contracts. Safety certifications need accountable signatures. A misconfigured battery can start a fire. A miscommunicating inverter can destabilize a feeder. None of that maps to a permissionless token model.What works in energy is the boring version. SPVs financing batteries. Long-term service agreements with property owners. FCR, mFRR, spot arbitrage, capacity contracts. All denominated in euros, dollars, kronor. All flowing through banks that have been around since the 1900s. That’s what we built at Sourceful. We kept the coordination insight and plugged it into regulated money.Closing note on the production arc. Episodes 1 and 2 were fully AI-generated, undisclosed in the audio itself. This episode puts everything on the table. The voice is a clone. The embedded debate is NotebookLM. The ideas and direction are mine. Same experiment, better pipeline, more honest about what it is.Pull quotes“DePIN is dead. Dead in the way it was thought of in the beginning. The DePIN we were sold in 2021, 2022 — tokens at the center of physical infrastructure — that version is over.”“The betrayal isn’t just about the pizza. It’s the realization that your labor wasn’t an investment in a shared community at all. It was just a way for the founders to avoid paying upfront construction costs.” (from the NotebookLM debate)“Helium’s governance worked. It was honest. It was accountable. And we got a yes.” (on HIP 128)“It is entirely possible to build a useful decentralized network AND destroy value for the people who bankrolled the bootstrapping. Those two things are not the same thing.”“A reliable business partner with a bank account and an engineering team is worth more than any token incentive.”“The companies that figured this out don’t call themselves DePIN anymore. They call themselves AI infrastructure companies. Telecom offload providers. Grid services operators.”“DePIN is dead. Long live coordination.”Production disclosureNarration: ElevenLabs voice clone of Fredrik, reading a script Fredrik wrote. Embedded debate (~25 minutes, middle of episode): Google NotebookLM, two synthetic voices, research brief commissioned by Fredrik. Script iteration: Claude. Editorial direction: all Fredrik.Every voice in this episode is synthesized. Every idea is not.References* Episodes 1 and 2 — The Rise and Fall of Rowan Energy and The Great Green Heist Reach outIf you want to argue with me about the Helium Mobile nuance, push back on the energy thesis, or compare notes on running distributed infrastructure without a token — the inbox is open. Marcus Aurelius: loss is nothing else but change, and change is Nature’s delight. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

Episode Summary: In this episode, we explore the dual reality of Europe’s energy transition. In 2025, the EU reached a massive milestone: wind and solar officially generated more electricity than fossil fuels. However, this rapid growth has exposed a critical bottleneck: the power grid. With over 120 GW of planned renewable projects at risk of being stranded and massive connection queues slowing down progress, we look at how the power grid is struggling to keep up. Finally, we dive into the massive financial and technical opportunity of battery storage—particularly in the underserved Commercial & Industrial (C&I) sector—and discuss how financial innovation could rapidly accelerate Europe’s battery buildout, much like it did for the US shale and solar industries.Key Takeaways & Highlights:* The 2025 Tipping Point:* Wind and solar reached a record 30% of EU electricity generation in 2025, overtaking fossil fuels (29%) for the very first time.* Solar power continues its staggering growth, expanding by 20% in a single year to generate 369 TWh of electricity.* The Grid Bottleneck:* Europe’s energy ambitions are hitting a wall of wires. A shortfall in grid capacity has put at least 120 GW of planned renewable power at risk.* Without immediate action like deploying “non-wire solutions” to optimize existing infrastructure, up to 1.5 million households could face major delays in connecting rooftop solar panels.* Currently, almost 700 GW of renewable projects are stuck waiting in connection queues across 10 reporting EU countries.* Batteries & The Cost of Wasted Energy:* The EU spent €32 billion on imported gas for power in 2025, largely to meet demand during peak evening hours when the sun sets.* At the same time, clean energy is frequently wasted. In Germany alone, storing curtailed (wasted) wind and solar energy in batteries could have saved an estimated €830 million by displacing the need to buy expensive fossil gas.* The Commercial & Industrial (C&I) Opportunity:* While utility-scale and residential batteries are booming, the C&I sector is the “missing middle,” making up only 8% of the EU battery market in 2025.* However, the C&I battery market is projected to surge at a 71% compound annual growth rate to €4.1 billion by 2028.* Sweden’s battery market currently mirrors California’s in 2019—right at the inflection point of explosive growth. We discuss the “Capital Markets Maximalism” thesis: the idea that physical infrastructure (like batteries) requires parallel financial engineering (like securitization and project finance) to truly scale, similar to how SolarCity’s asset-backed securities unlocked the residential solar boom in 2013.Resources & Further Reading:* Ember Report: European Electricity Review 2026* Ember Report: Crossed wires: Grid capacity could block EU energy security* Research Briefs: The coordination layer thesis for European C&I battery storage and Crucible Capital’s Capital Markets Maximalism This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

The story everyone tells about energy is about generation. Solar panels, wind turbines, the clean energy race. And that story is true. Wind and solar just overtook fossil fuels in Europe for the first time — 30 percent versus 29 percent. That’s a real milestone.But the story I want to tell is about what comes after the generation war is won. Because that war is over. The hardware is cheap. The physics works. We have more clean electricity than we know what to do with.The question now — the only question that matters — is whether we can coordinate it.The data is stunningEmber’s European Electricity Review for 2026 tells a remarkable story. Solar alone hit 369 terawatt-hours in 2025, growing over 20 percent year on year for the fourth straight year. It’s now 13 percent of EU electricity, higher than coal, higher than hydro. Meanwhile, Bloomberg NEF reported stationary storage LFP packs hitting $70 per kilowatt-hour — down 45 percent in a single year and 93 percent cheaper than 2010.Generation is abundant. Storage hardware is cheap. The “hard to do” is done.But here’s what the same Ember report also shows: gas generation in the EU rose 8 percent in 2025. Despite having more wind and solar than ever. Gas rose because hydro dropped 12 percent due to weather. That pushed the EU power sector’s gas import bill to €32 billion. Sixteen percent higher than the year before.Let that sink in. Europe has more clean energy than at any point in history. And it’s still spending €32 billion importing gas to keep the lights on.Germany curtailed 9.6 terawatt-hours of wind and solar in 2025 — nearly 4 percent of total output, clean energy generated and then thrown away. Solar curtailment alone rose 97 percent year on year. Spain hit 11 percent curtailment by mid-2025. Sweden recorded 649 negative-price hours in twelve months.This is the paradox of abundance. The cheaper energy gets, the more valuable the coordination layer becomes.Jevons was right (again)William Stanley Jevons observed in 1865 that Watt’s more efficient steam engine didn’t reduce coal consumption — it increased it. Efficiency made coal cheaper, which made it more useful, which drove more demand.The same thing is happening with energy. Since 1990, energy efficiency has increased 36 percent globally. Total consumption has increased 63 percent. LEDs are more efficient, so we installed more lights. AI chips are more efficient per computation, so we built more data centers. In Ireland, data centers went from 5 to 21 percent of national electricity consumption.For renewables, the dynamic has a twist. Solar and wind produce energy when the physics allows it, not when the grid needs it. Abundance without coordination doesn’t just mean waste. It means you still need the gas plants on standby. You still pay the €32 billion import bill.Ember calculated that if batteries had captured roughly one-third of Germany’s curtailed energy, it would have saved €800 million — €613 million in avoided redispatch costs, €219 million in avoided gas generation. The required battery investment? Just €145 million per year over the technology’s lifetime.And then the Iberian blackout drove the point home. April 28, 2025: 31 gigawatts disconnected, 47 to 56 million people in the dark, over €1.6 billion in losses. Spain had just 25 to 60 megawatts of installed battery storage against a 500 megawatt target.Batteries in buildingsThe EU installed 27.1 gigawatt-hours of new battery storage in 2025 — a 45 percent increase. But look at the breakdown: utility-scale was 55 percent, residential was 36 percent, and commercial & industrial was just 8 percent.Eight percent. Despite being the fastest-growing segment with a projected 71 percent CAGR to €4.1 billion by 2028.The residential market has clear winners — 1Komma5 in Germany with over 300,000 systems, Sonnen with their virtual power plant. Utility-scale is scaling fast. But the middle? The 50 kW to 2 MW behind-the-meter commercial segment? It’s wide open.In the entire Nordic region, there is no single company that combines zero-cost deployment, hardware-agnostic system design, automated multi-market optimization, and end-to-end site coordination for commercial batteries.Not Pixii (sells hardware, doesn’t finance it). Not CheckWatt (aggregates, doesn’t deploy). Not ABB (battery-as-a-service, but only UK). Not 1Komma5 (residential, no concrete commercial product).This is the biggest gap in European energy infrastructure.Financial architecture always comes second — and matters moreThere’s a pattern that repeats in every major infrastructure buildout. The technology creates the asset. Then a financial innovation creates the market that enables scaling.Oil. Drake drilled the first well in 1859. For 124 years, oil traded through opaque bilateral contracts. Then in 1983, NYMEX launched crude oil futures. Today they trade over a million contracts daily. But the critical point: futures enabled hedging, hedging enabled reserve-based lending, and lending enabled thousands of independent drillers to develop distributed assets. Without futures, there is no shale revolution.Railroads. The US built 190,000 miles of track requiring $10 billion in capital when all US manufacturing assets totaled $6.5 billion. The financial innovation was the first mortgage bond — long-term, convertible, secured by company property. Railroad bonds didn’t just finance railroads. They created the American corporate bond market.Telecom. Between 1996 and 2001, $500 billion went into fiber optic cable. When the bubble burst, only 5 percent of the fiber was lit. But the infrastructure survived. Google, Facebook, AWS, and Netflix scaled on overcapacity fiber that bankrupt investors had paid for.Bitcoin. Eight years of pure technology: $16 billion market cap. Then CME futures, Coinbase Custody, the first futures ETF, and spot ETFs. Result: $2.5 trillion. Technology alone — $16 billion. Technology plus financial instruments — 156x.SolarCity. Their insight wasn’t about panels. By 2010, panels were commodity. The insight was that the bottleneck was financial and operational. Zero dollars down, twenty-year leases. In November 2013, the first solar asset-backed security: $54.4 million, rated BBB+, 4.8 percent. That single deal birthed a $15 billion solar ABS market.But SolarCity also shows exactly how the model breaks. Customer acquisition costs spiraled to $6,000 per customer through door-to-door sales. $3.58 billion in debt against $730 million in revenue. Tesla acquired them in what was essentially a rescue.The lesson: the technology was table stakes, the financing was transformative, but the customer economics killed it.Battery storage for commercial buildings is at exactly this inflection point. The technology is proven. The economics are compelling. No standalone battery ABS has ever been issued. The “SolarCity moment” for batteries belongs to whoever creates the first standardized instrument backed by diversified battery revenue streams.Why Sweden, why nowSomething most people don’t know about Swedish commercial buildings: a typical property uses 20 to 35 percent of its grid connection on average. The remaining 65 to 80 percent is capacity the owner pays for every month and never uses.Swedish grid fees rose 10.6 percent in 2025 — the largest increase in 30 years. For a commercial property with 500 kW peak demand, demand charges alone run roughly 385,000 SEK per year. The regulator’s revenue cap for 2024-2027 increased roughly 40 percent. Grid companies still have 30 billion SEK in unused revenue headroom. Sweden expects to double electricity consumption by 2045.The recent pause on residential capacity tariff rollout doesn’t change this. Commercial demand charges above 63 amps remain fully in effect, the EU Electricity Market Directive still requires capacity-based pricing, and operators who already implemented effekttariffer can keep them.And here’s the thing that changes the conversation with property owners: Swedish commercial property valuation follows one formula. Value equals net operating income divided by yield. At Stockholm prime yields of 3.85 to 4 percent, a 500,000 SEK annual energy saving translates to roughly 12.5 million SEK in property value increase. That’s a 20 to 25x multiplier.You don’t sell kilowatt-hours to a property owner. You sell property value creation.Revenue stacking makes the math workWhile the battery shaves peaks during the day, it participates in grid markets at night.FCR-D has saturated — prices crashed from 30-60 EUR/MWh in 2022 to under €3 in February 2025. But mFRR is growing. Spot arbitrage in SE3 averages 20-50 EUR/MWh spread on normal days, spiking to 80-200 on volatile days. And aFRR opens when Sweden joins the PICASSO platform, expected 2026-2027.For a 200 kW / 400 kWh system, the conservative annual revenue stack: peak shaving 50-70k SEK, frequency reserves 40-75k, spot arbitrage 25-50k, mFRR 5-15k. Total: 130-235k SEK against an installed cost of roughly 1.15 million. Simple payback of 6.5 to 9.5 years, improving as battery costs fall and new markets open.No single revenue stream works alone anymore. FCR-D prices collapsed when capacity surged. Peak shaving alone won’t justify deployment cost. The stacking is the business case. And stacking requires edge-based optimization that dispatches across all markets simultaneously in real time — because when Svenska Kraftnät calls for frequency response, you have 7.5 seconds.The moat is physical, not technicalThe software isn’t the moat. If I told you our competitive advantage was our software, I’d be lying.American Tower owns 224,000 cell towers. Market cap: $81 billion. EBITDA margin: 66-68 percent. Their technology? Commodity antennas. Their moat? Permitted physical sites with long-term leases, zoning approvals that take years, and co-location economics where adding tenants is nearly pure margin.Tesla Superchargers — same pattern. 80,000 stalls, 52.5 percent of US DC fast-charging. The advantage wasn’t charging technology. It was vertical integration, density, and deployment speed.The moat is that this business is hard to assemble, not hard to build. Battery hardware is commodity — LFP cells at $36-50/kWh from dozens of manufacturers. The scarce asset is the combination of: physical site access through 5-15 year contracts. Grid connection and market access. Operational data from hundreds of systems. Capital structures enabling zero-cost deployment.Every signed contract locks a competitor out of that building for a decade. Every operating hour generates dispatch data that compounds. Institutional capital — EIB, NIB — requires 2-3 years of portfolio performance data before writing a term sheet. First movers with verified data get better rates. That gap widens every month.Being honest about the risksRevenue risk from market saturation is real. Sweden’s FCR-D market saturated within 18 months. China could flood Europe — Chinese turnkey systems cost $73/kWh versus $177 in Europe. ABB just launched battery-as-a-service in the UK, Volvo Energy is entering with containerized 1 MW systems. There is no Swedish government incentive for commercial batteries, unlike California’s 30-50 percent investment tax credit.And the commercial customer is genuinely hard. Factory managers, logistics operators, property companies — each with different decision processes and risk tolerances.But the multinationals have a structural disadvantage. They’re optimized for large, standardized deployments. The commercial segment is messy. It’s 200 kW systems in warehouses with weird electrical panels and grid connections that need custom work. The multinationals are good at selling equipment. They’re not built for the hands-and-knees deployment work.The hard to getThere’s a distinction in business strategy between what’s hard to do and what’s hard to get.Hard to do is technology. A better battery, better algorithms, a faster edge device. Those things are hard for everyone equally, and anyone with enough talent and capital can eventually do them.Hard to get is different. Hard to get is the signed site agreement that took three months of conversations and a board approval. The grid connection permit after six weeks with the local operator. Three years of dispatch data from 200 sites. The institutional capital relationship built on verified performance, not pitch decks.The energy transition has entered a new phase. The technology phase is over. We’re in the deployment phase. And deployment is about the hard-to-get things — site access, capital structures, operational data, grid relationships, physical installation at scale.The grid is becoming a network of millions of distributed resources. Someone needs to coordinate them. Someone needs to do the physical, unglamorous work of putting batteries in buildings and making them earn.That’s what we do.Fredrik Ahlgren is CEO and co-founder of Sourceful Energy. Reach out at sourceful.energy or connect on LinkedIn. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

In 1962, a researcher at the RAND Corporation named Paul Baran looked at AT&T’s telephone network and saw something that the people running it could not see. The system worked. It carried every voice call in America. But it was structurally wasteful. Circuit-switched networks reserved entire connections for conversations that used less than 10 percent of available bandwidth. Baran’s idea was to break data into small packets, route them independently, and reassemble them at the other end. He called it packet switching.He presented this to AT&T’s engineers multiple times throughout the mid-1960s. They told him he did not understand how voice telecommunications worked.They were not stupid people. They were reasoning from the wrong model. They saw analog signals on dedicated lines. Baran saw data that could flow through any available path. The gap was not intelligence. It was architecture.I keep coming back to this story because we are living through the same moment in energy. And most of the people running the current system cannot see it.The numbers that should change how you thinkIn the first half of 2025, the world installed 380 gigawatts of solar. That is 64 percent more than the same period in 2024. To put it bluntly, we deployed more solar in six months than existed on the entire planet as recently as 2017. Eight years of cumulative build-out, compressed into half a year.In the US, solar plus storage accounted for 79 percent of all new generating capacity. Not 79 percent of renewables. All capacity.Solar panels now cost roughly ten cents per watt. Stationary battery pack prices hit $70 per kilowatt-hour in 2025, a 45 percent drop in a single year. The lowest commercial cell prices are $36/kWh. These are not lab numbers. These are prices at scale.The hardware problem is solved. Panels are close to free compared to a decade ago. Batteries are following the same curve with a few years of delay. Deployment keeps outrunning every forecast the IEA publishes. This is not a story about scarcity. This is a story about abundance.But here is the part nobody wants to say out loud: we are wasting enormous amounts of this perfectly good electricity.Abundance without coordination is just wasteCalifornia curtailed 3.4 million megawatt-hours of renewable energy in 2024. Ninety-three percent of that was solar. Germany set a new record of 1,750 GWh curtailed in 2025, with 575 hours where the wholesale price went negative. Generators paying someone to take their electricity.Spain’s solar capture prices collapsed from €61/MWh to under €17. The Netherlands posted 584 hours of negative prices, the most in Europe. Australia curtailed over 7 TWh. In South Australia, 38 percent of all utility-scale solar generation was thrown away.Across Europe, managing these surpluses cost utilities an estimated €4.3 billion in 2024. Globally, curtailment hit 72 TWh. That is roughly Austria’s entire annual electricity consumption. Gone.We spent billions deploying all this capacity. Then we pay people not to use the electricity it produces. That is not a generation problem. That is a coordination problem. And it is getting worse every month as more solar comes online.The pattern that keeps repeatingThe Baran story is not unique. The same structural shift has played out in shipping and in software platforms.In 1956, Malcolm McLean loaded 58 aluminum truck bodies onto a converted tanker ship in Newark. The shipping container was born. Loading costs dropped from $6 per ton to 16 cents. A 36x reduction. But the container itself was just a steel box. The revolution was the standardization: the standard size, locking mechanisms, and handling equipment. The coordination protocol that let any container from any shipper load onto any ship and transfer to any truck. A longshoremen’s union official watching McLean’s first container ship depart reportedly said he would like to sink that son of a b***h. The incumbents always react the same way.Apple launched the App Store in July 2008 with 500 apps. It hit 10 million downloads in 72 hours. Within a few years it was facilitating over a trillion dollars in commerce. Apple did not build the apps. They built the platform layer: the APIs, the SDKs, the payment infrastructure, the discovery mechanism that turned millions of individual devices into a coordinated ecosystem.The meta-pattern is always the same. A distributed resource exists but is uncoordinated. Incumbents dismiss it. A standardized coordination layer emerges. The economics flip. And an outsider, not an industry insider, drives the change.TCP/IP captured more value than any individual computer. The App Store captured more value than any individual app. The ISO container standard transformed more wealth than any individual shipping company.The coordination layer always wins.Virtual power plants are not a pilot anymoreDuring California’s July 2025 grid stress test, over 100,000 home batteries delivered an average of 535 megawatts to the grid during the evening peak. The Brattle Group verified the results and concluded that the output was indistinguishable from dispatching a gas peaker plant. Grid operators could not tell the difference between 100,000 batteries coordinated by software and a single gas turbine.EnergyHub has introduced what they call the Huels Test, essentially a Turing Test for energy. Can a grid operator distinguish a VPP dispatch signal from a gas peaker? In trials with Arizona Public Service, Duke Energy, and National Grid, virtual power plants passed. Level 3. Automated and indistinguishable.The scale is already real. North American VPP capacity reached 37.5 GW in mid-2025. Octopus Energy’s Kraken platform manages 2 GW from over 500,000 connected devices. Kraken spun out as an independent company at a €7.3 billion valuation. Tesla’s Autobidder orchestrates over 3 GW across energy, capacity, and ancillary service markets. Base Power, a Texas startup, raised $1 billion and is installing 20 MW per month, targeting 100 MW per month by mid-2026.A 400 MW virtual power plant costs $43 per kilowatt-year. A gas peaker costs $99. Less than half. The DOE verified those numbers.Meanwhile, the United States has 2,600 GW of clean energy stuck in interconnection queues. That is more than twice total installed US generating capacity. Just waiting. Only 14 percent of solar projects that enter those queues ever get built. The average wait time in PJM is eight years.Eight years to connect a solar farm to the grid. Versus three years to build a 700 MW virtual power plant from home batteries. The distributed path does not compete with the centralized path. It laps it.The physics gapThe classic grid was built for central control of big generators. Now we have millions of distributed devices at the edge. An inverter speaks Modbus RTU. An EV charger uses OCPP. A heat pump might use Modbus TCP or some proprietary protocol. A smart meter outputs a P1 port stream. Different devices, different languages, no common protocol.And the control requirements are brutal. Grid services like frequency response and voltage regulation need feedback on 100-millisecond to one-second timescales. Existing cloud-based interfaces have latencies of two to thirty seconds. There is no execution layer between cloud platforms and physical devices. The gap between what the cloud can do and what the grid actually needs is not a software bug. It is physics.Generation is no longer the bottleneck. Control is. We have the energy. We have the devices. What we do not have is the real-time, protocol-agnostic, edge-native intelligence layer that makes them work together at grid speed.Why we build from the NordicsSourceful Energy is based in Kalmar, a small city on Sweden’s southeast coast. Not exactly famous for sunshine. But the Nordics have some of the most modern electricity grids in the world. Norway has 93 percent of residential contracts on dynamic tariffs. Finland is at 30 percent. Denmark has approved vehicle-to-grid aggregators for frequency reserves, with trial accuracy exceeding 95 percent. Sweden has had deregulated electricity markets since the 1990s.If you want to build coordination software that works everywhere, you start where the market is most advanced. Where the infrastructure expects intelligence. Where a small team with the right architecture can punch above its weight. You do not wait for the world to be ready. You build for where it is going.The Jevons flywheelI did my PhD at Linnaeus University in Kalmar, using machine learning to optimize energy efficiency for ships. One pattern kept showing up: every time you make energy use more efficient, total consumption goes up, not down. William Stanley Jevons documented this in 1865. Efficiency does not reduce demand. It unlocks it.That insight rewired how I think about the energy transition. The conventional narrative of consuming less and restraining ourselves is backwards. Making energy cheaper and more abundant is the most powerful force for human progress there is.And it creates a flywheel. Cheaper solar drives more deployment. More deployment drives more consumption, because when energy is cheap you electrify everything: transport, heating, industrial processes, computing, direct air capture. US data center power demand alone is projected to nearly triple by 2030, to 134 GW. Europe aims for 30 million EVs by 2030, each carrying 50-75 kWh of battery. Thousands of gigawatt-hours of mobile storage, every vehicle a potential grid asset if you have the coordination layer to use it.More consumption creates more intermittency and grid stress. More grid stress creates more demand for coordination. More demand for coordination makes the orchestration layer more valuable. Which attracts more devices. Which enables even more solar. The flywheel does not slow down. It accelerates.Being honest about the challengesThe most serious technical objection to a solar-powered future is the Dunkelflaute. It is a German word meaning dark doldrums. Periods where both solar and wind output collapse. Germany experiences roughly two per year, where output drops below 10 percent of capacity for 48 hours or more. In January 2017, a 10-day Dunkelflaute left Germany’s 91 GW of installed wind plus solar producing less than 5 percent capacity while demand sat at 63 GW. Wholesale prices spiked above €900/MWh.Current batteries cannot economically bridge multi-day gaps. That is a real limitation. Grid stability is another legitimate concern. Traditional synchronous generators provide rotational inertia that buffers frequency deviations. Inverter-based renewables do not have spinning mass. Supply chain concentration is real too. China controls 60 to 90 percent of critical mineral processing.I do not dismiss any of this. But notice something about every single one of these objections. The Dunkelflaute requires cross-border grid coordination and long-duration storage. Both are software orchestration problems. Grid stability requires grid-forming inverters and synthetic inertia, which is a software function running on existing hardware. Supply chain diversification requires demand aggregation and flexible procurement, which is again coordination.The critiques do not weaken the thesis. They reinforce it. The missing piece is not more panels or cheaper batteries. It is the intelligence layer that integrates them.A gigawatt of baseload solarThe Masdar Round The Clock project broke ground in Abu Dhabi in October 2025. The goal: demonstrate that solar plus storage can deliver continuous baseload power. 5.2 GW of solar PV paired with 19 GWh of battery storage, spread across 90 square kilometers. Designed to deliver a flat 1 GW production profile, 24 hours a day, 365 days a year. Six billion dollars of investment, target completion 2027, with agreements already signed to replicate the design in Kazakhstan.If this works, and the physics says it should, it breaks the last argument against solar as a baseload source. And the project is designed to function as a virtual power plant, providing grid services. Not just generation. Coordination.The betThe hardware revolution created the substrate. Solar at ten cents a watt. Batteries at $70 per kilowatt-hour. These are not projections. These are today’s prices.What is missing is the protocol layer. The coordination intelligence. The software that turns millions of cheap, intermittent, distributed devices into dispatchable, grid-reliable, market-participating capacity.Paul Baran described the distributed network he envisioned as having no center, growing from the edges, and being impossible to control from any single point. That is the grid’s future. Not a centralized system managed by a handful of giant utilities. A distributed system where millions of devices at the edge coordinate through software to deliver reliable, abundant energy.The incumbents will react the same way AT&T reacted to Baran. They will say we do not understand how the power system works. And they will be wrong. For the same reason AT&T was wrong. Not because they lack intelligence. But because they are reasoning from the wrong model. They are thinking in circuits when the world has moved to packets.That is what we are building at Sourceful. That is the bet. And I have never been more convinced it is right.Fredrik Ahlgren is the CEO and co-founder of Sourceful Energy. This post accompanies Episode 073 of the Coordinated with Fredrik podcast. Subscribe wherever you get your podcasts, or read more at sourceful.energy. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

I recorded a podcast episode this week that I’ve been turning over in my head for months. It’s about where Sourceful is right now, what we’ve built, and why I think the next few months matter more than anything we’ve done so far. If you have the time, go listen. But if you want the written version, here it is.The clean problem and the dirty oneMost energy tech companies pick a clean problem. They build monitoring dashboards. Analytics platforms. Pretty graphs sitting on top of someone else’s data. The physical world — the actual devices, the actual protocols, the actual hardware that lies about its own state — that’s always conveniently left to someone else.We went the other way. Sourceful has always been about the physical layer. Talking to actual devices. Sending actual commands. Getting actual confirmations that hardware did what it was told to do.This is an ugly problem. Every manufacturer implements protocols differently. Firmware updates silently change behavior. Documentation doesn’t match reality. The same device model bought two years apart can have different register maps. Nothing is clean. Nothing is standardized. Everything is edge cases stacked on top of edge cases.We chose this on purpose.What the grid actually needsThere’s a fundamental misunderstanding in energy tech about what “smart grid” means. Most people hear it and think monitoring. Data collection. Visibility.That’s table stakes. Visibility is not a product. The product is control.A utility needs to curtail 40 MW of solar generation in under a second. An aggregator needs to simultaneously discharge two thousand home batteries to meet a frequency regulation bid. A grid operator needs to shift ten thousand heat pumps off-peak — right now — and know with certainty that it happened.Deterministic, real-time control of physical hardware. That’s what the flexibility markets pay for. That’s what grid operators need. That’s the product.And that product requires solving the integration problem at the device level. There’s no shortcut. There’s no abstraction layer that makes the physical world go away. Someone has to write the driver that talks to each specific device, handles its specific quirks, and confirms that commands were executed correctly.Why cloud coordination is a dead endI keep seeing well-funded startups building cloud-first coordination platforms. Some of them are smart. Most of them will fail. Not because the teams are bad — because the architecture is wrong.Cloud API round-trips take two to five seconds. Grid frequency balances every second. This isn’t a performance gap you can close with better infrastructure. It’s a physics problem. If your control logic lives in someone’s data center, you are structurally unable to participate in the energy markets that pay the most: frequency containment reserves, fast demand response, sub-second optimization.These markets require local execution. The coordination logic has to run on the same network as the device. At the edge. In the building. Next to the hardware.You cannot retrofit local execution onto a cloud architecture. The entire system has to be designed for it from the ground up. Most teams don’t realize this until they’ve already built the wrong thing and hit the ceiling.We started local. That decision — made before it was fashionable, before “edge computing” was a buzzword in energy — is now an architectural advantage that can’t be replicated without starting over.What we’ve been doing for the last six monthsI want to be honest about the journey, because the honest version is more instructive than the polished one.We didn’t start from first principles. We started the way startups start — shipping fast, accumulating debt, building what worked rather than what was architecturally right. That’s fine for finding product-market fit. It’s not fine for building infrastructure the grid depends on.The last six months have been about stripping back to first principles. Rearchitecting the entire backend. Killing legacy code paths. Rebuilding the foundation from the physics up. We shipped NovaCore — not a feature, a new foundation — in the last two months. The identity system. The control pipeline. The telemetry infrastructure. All redesigned around how the grid actually works.That work is done. The foundation is solid. And it earned us the right to tackle the hard part properly.AI-first integration: the 10x and 10xHere’s the thing that changes everything for us.We’ve built an AI-first integration engine — we call it Hugin — that can point at an unknown energy device, scan it, figure out how it communicates, cross-reference our existing driver library, and produce a working, tested driver. In minutes.The industry standard is weeks per device brand. One engineer, one brand, reverse engineering and manual testing. We’ve lived this. We have about thirteen OEM integrations today. The market has hundreds of brands. At the old pace, nobody wins.With AI-driven integration, the first device is roughly 10x faster than manual development. By the fifteenth integration, the AI has seen enough devices from the same manufacturers to recognize patterns — it already understands 80% of a new device before it starts. By the fiftieth integration, it’s 10x faster again. That’s 100x the pace of anyone still hand-coding drivers.And the curve doesn’t flatten. It steepens. Every driver teaches the AI about the next one. Every edge case gets encoded into the knowledge base. The messy physical world that everyone else avoids is literally the training data that makes our system smarter.A new partner needs fifteen device brands supported? Today that’s months. With this platform at scale, it’s a day.The people with skin in the gameHere’s what most platforms get wrong about scaling integration. They treat it as an internal engineering problem. Hire more developers. Grind through the backlog. Even with AI, that’s still one company trying to cover an entire industry.What scales is when the people who need an integration the most are the ones building it.Who cares whether a specific inverter model works with Sourceful? The person who owns that inverter. The homeowner with the device in their garage. The electrician who installs that brand every week. These people have the hardware physically in front of them. They can plug in and run a probe.With AI-first tooling, they don’t need to be protocol engineers. They point the tool at their device, supervise the process, review the output, and submit a driver. That driver gets reviewed, signed, and distributed. Every Sourceful gateway in the world that encounters the same device model now has a working integration.One person solved their own problem. The entire network got smarter.The driver library grows faster than any internal team could build it. Coverage expands into regional brands and legacy models that no company would prioritize on a product roadmap. And we have the infrastructure to reward contributors when the time is right — we know who contributed what, how widely it’s used, and what revenue flows through it. We’re not building the incentive model today. But the pipes are ready.Four moats that compound each otherI think most people in energy tech define moats wrong. A moat isn’t a feature or a patent or being first. A moat is something that gets stronger the more you use it, and that competitors can’t replicate without going through the same painful process.We have four, and they reinforce each other.The driver library. Every validated, production-tested driver represents real-world knowledge about how a device actually behaves — not how the documentation says it behaves. You can’t generate this from a spec sheet. You get it by connecting to real hardware in real installations. Every driver is a brick in a wall that competitors have to build from scratch.The AI knowledge base. Every integration teaches the system about the next one. The compounding curve is structural, not aspirational. A competitor starting today begins at integration one. We’ll be at fifty by the time they’ve set up their development environment. Their integration one will be a hundred times slower than our integration fifty.Local execution. Building local-first coordination is architecturally harder than building cloud. Most teams default to cloud because it’s easier. Then they hit the physics ceiling and discover there’s no shortcut past it. We started local. That decision is now an advantage that requires a full rebuild to replicate.The network. Every deployed gateway is a node. Every node makes the platform more valuable to every other node. Utilities want one platform covering the most devices in the most locations. Aggregators want the largest pool of controllable assets behind a single interface. More gateways make us more attractive. More attractiveness means more gateways. Network effect, rooted in physical hardware on real grid connections.These four moats compound each other. More drivers feed a smarter AI. A smarter AI drives faster integration. Faster integration deploys more gateways. More gateways attract more contributors. More contributors produce more drivers. The flywheel accelerates with every turn.Why execution is the only thing that mattersThe ideas in this post are derivable. Smart people will arrive at similar conclusions about local execution and AI-driven integration. Some probably already have.What they can’t derive is the driver library we’re building right now. The knowledge base we’re training right now. The gateways we’re deploying right now. These are assets that can only be created by shipping. Not by planning. Not by raising capital. By shipping.A head start in a compounding system doesn’t stay the same size. It grows. Every week of execution adds drivers, edge cases, knowledge, and deployed nodes. A competitor starting today isn’t six months behind us — they’re six months behind plus every driver, every edge case, every deployment, and every community contribution we’ve shipped in those six months.The gap widens. It doesn’t close.Where we are right nowWe spent six months earning the right to do this properly. The backend is rearchitected. The legacy is dead. The foundation is at first principles.Now we’re building the integration layer on top of it. The gateway software. The AI engine. The driver distribution system. Every existing driver across all our hardware is being rebuilt on the new unified architecture. No hybrid approaches. No legacy paths alongside new ones. Clean break. One platform.When this ships, we’ll have something no one in the energy industry has ever had. The ability to point an AI at any energy device, produce a working integration in minutes, deploy it instantly, and have that integration compound the intelligence of every future integration across the entire network.The coordination layer the grid doesn’t have yet.We’re building it.If you’re working in energy, building hardware, or just have devices at home that you think should be doing more — I want to hear from you. Reply to this post or find me on LinkedIn.And if you haven’t already, listen to the full episode: The Mess Is the Moat — Coordinated with Fredrik, Ep. 74 This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

Before we begin, I want to be upfront about something.The voice in this episode is mine, but it’s AI-generated.I cloned my voice using ElevenLabs, and the entire episode was produced that way. I think it’s important to be transparent about that. The words are mine. The ideas are mine. The delivery is artificial.That feels like a fitting place to start, because a big part of what I want to talk about is exactly that tension: how technology should be used as a tool, not as a replacement for what is real.This post is the written version of something deeper than a company update. It is not the pitch version. Not the investor version. This is the real version — what I actually believe, why I believe it, and what I’m asking of the people building with me.But first, something important.I did not start this company alone. I started it with my co-founders. This has always been about the team. I take my role as CEO extremely seriously, and I do everything I can to earn that position every day. But I would never pretend this is a one-man show. It isn’t. The reason we have something worth talking about is because a group of people decided to build it together.Some of you reading this are part of that team. Some of you are following from the outside. Either way, I want to be straight with you, because I think that is the only way any of this works.And to understand where we are going, you need to understand where this started.A kid in SmålandI grew up in Åseda, a small town in Småland.If you know anything about Småland, you know it tends to produce stubborn people. People who do not quit easily. That matters more than it might seem.Even as a kid, I had a certain way of disappearing into things. My first-grade teacher told my parents that I was often somewhere else in my mind. Back then they called it daydreaming. It did not feel like daydreaming to me. It felt like locking in. When I got interested in something, everything else disappeared.That trait has followed me my whole life. On some days it is a superpower. On other days it is a curse.Very early on, I wanted to understand how things worked. I tinkered with hardware, took computers apart, put them back together, built machines from parts. I was not of the C64 generation. I started with a 286 PC running MS-DOS. From that point on, I was hooked.I read everything I could get my hands on — Tolkien, Asimov, anything that expanded the world. I was deep into RPGs, tabletop games, worldbuilding, systems. I wrote text-based games in QBasic and spent absurd amounts of time optimizing memory just to make them run. Back then, 640 kilobytes was supposed to be enough for everyone. It rarely felt like enough.What all of that taught me, early, was simple: systems have rules. If you understand the rules, you can make systems do what you want.That is still basically what I do for a living.By the age of eleven, I had set up a bulletin board system in my room called The Heart of Gold, named after the ship in The Hitchhiker’s Guide to the Galaxy. This was 1991. I was connecting with people over phone lines, learning how networks worked before I fully understood what a network even was. FidoNet. The proto-internet before the internet. I still remember the sound of a 2400 baud modem connecting.I was also part of the wares scene. The pirating movement. Cracking and distributing software. It was not legal, and I am not trying to romanticize it. But it taught me things school never did: how communities self-organize, how information flows, how systems that try too hard to close themselves eventually get routed around.By my mid-teens I was running Linux Slackware and compiling my own kernels. Not casually. Obsessively. Learning what each module did, what belonged, what did not, how the full system fit together. That bottom-up systems thinking never left me.And that is where Sourceful really comes from. Not from a business plan. Not from a market map. It comes from a kid in Småland who built PCs, ran a BBS, compiled kernels, and never stopped wanting to understand how systems work.Systems under pressureThe Navy was where systems thinking stopped being theoretical.I started on fast attack craft with gas turbine propulsion. Raw speed. Raw power. In that environment, you are responsible for the whole system, not just one part of it. If something fails, you cannot call support. You cannot Google it. You fix it, or you have a serious problem.Later I moved to submarines with electric battery propulsion. A completely different environment — quieter, more confined, slower in some ways, but under even greater pressure. Weeks underwater. No help coming. The same principle applied: if you do not understand the full system, you are a liability.That shaped me profoundly.As a marine engineer, you learn that nothing exists in isolation. The electrical system affects propulsion. Propulsion affects cooling. Cooling affects air systems. Everything touches everything else. If you only understand a component but not the system, you are dangerous.And then there is the human side. Rough conditions. Broken sleep. Tight spaces. Stress. You learn to lead tired people in environments where mistakes have immediate physical consequences. In the Navy, the commanding officer stands in what we call the windshield — the position where the impact hits first.That idea stayed with me.That is how I think about leadership. When you are in the windshield, you do not get to have an off day. You do not get to be unprepared. The people around you depend on you being sharp, present, and honest about what you see and what you do not see.First principles or nothingAfter the Navy and a short period as a project leader, I did something that surprised a lot of people: I went into teaching.I became an adjunkt, teaching marine engineers subjects like pumping technology and hydromechanics. And that is where I really fell in love with first principles.I would stand at the whiteboard and derive SI units from scratch. No shortcuts. No hand-waving. I wanted students to understand where formulas came from, not just memorize them. I would spend so much time writing that I would lose my voice and wear out my arm.That period shaped how I think about everything.If you cannot derive it, you do not understand it. If you do not understand it, you cannot build on it.For a lot of people, “first principles” is a buzzword. For me, it is literal. It is years of standing at a whiteboard, reducing ideas to fundamentals until there is nowhere left to hide.That mindset later led me into a PhD, where I spent years studying how to optimize ship energy systems using machine learning. A PhD requires a specific kind of lock-in. But more than that, it taught me something about teamwork. Academic work is collaborative, but the thesis is still yours to defend. You build with others, but you also carry something alone.During that work, I encountered an idea that changed my worldview: Jevons paradox.William Stanley Jevons observed in 1865 that when coal became more efficient to use, people did not use less coal. They used more. Efficiency lowered cost. Lower cost increased accessibility. Increased accessibility increased demand.That realization broke my brain in the best possible way.Before that, I held a more traditional environmentalist mindset: conserve, reduce, use less. Jevons forced me to see that this framing was not morally wrong, but mechanically wrong. The path forward is not austerity. It is abundance. The way to outcompete fossil fuels is to make clean energy so cheap, so scalable, and so abundant that it wins on economics.That was a major shift for me. I stopped being a conservationist in the traditional sense and became something closer to an accelerationist — not the reckless kind, but the kind that believes we solve the climate problem by building overwhelming clean energy capacity.That insight became the seed of Sourceful.The real problem: coordinationBut abundance alone is not enough.You can put solar on every roof, batteries in garages, EVs in every driveway. If none of those assets can coordinate, you have not solved the problem. You have simply created millions of independent actors in a system that fundamentally requires orchestration.That is the transition we are living through now.For more than a century, the power system was one-directional: big plant, long transmission lines, passive consumer. That model is ending, not because of ideology, but because the economics changed.The grid is becoming distributed. And distributed systems only work when coordination is real.By 2030, tens of millions of electric vehicles will be connected to European grids. Every one of them represents flexible load, storage, or both. Every rooftop solar system, every battery, every heat pump adds another actor. The problem is that these actors are largely blind to each other. Your battery does not know what my inverter is doing. Your EV charger does not know what grid frequency is doing. And the grid must balance supply and demand every second.Not every hour. Not every minute. Every second.That is where most of the industry made a mistake.A lot of companies looked at this challenge and said: no problem, we will coordinate it through the cloud. Aggregate everything centrally. Send commands down through APIs. Done.Except physics does not care about your cloud architecture.A cloud roundtrip takes seconds. Grid events happen in milliseconds. You cannot coordinate physical infrastructure at grid speed through cloud APIs. It is a fundamental architectural mismatch.That is what we saw.The same systems thinking I learned in the Navy, the same first-principles approach I developed in academia, all pointed to the same answer: control has to be local. Intelligence has to be at the edge. Computation has to happen where physics actually happens.That is the thesis.What we are buildingAt Sourceful, we are building the platform that enables AI software to operate at grid speed.Local execution. At the edge. On commodity hardware. In the real physical environment where the system actually lives.We are not a hardware company. We do not manufacture hardware. We are not an app company either. We are infrastructure.The simplest analogy is AWS. AWS does not run your application for you. It provides the infrastructure that makes running applications possible. We are building the equivalent infrastructure layer for energy at the edge.And we are doing it in a way that compounds.Every new device integration. Every driver. Every deployment. Every partner. Each one makes the next one easier. The platform gets stronger as it grows. That is by design.This is not a product you ship once. It is a platform that compounds.Technology, values, and the people who build itBuilding something this hard is not just about having the right architecture. It is about the people.And I do not think you can separate who you are from what you build.I am a stoic, and I mean that in the actual philosophical sense. Marcus Aurelius. Seneca. Epictetus. Focus on what you can control. Accept what you cannot. Act with virtue. Judge yourself by your actions, not your intentions.That philosophy has outlived empires for a reason. It works.There is also the Lindy Effect — the idea that the longer something has survived, the longer it is likely to continue surviving. A book that has lasted two thousand years is more likely to last another two thousand than some management fad from last quarter. I think about that a lot. I would rather build my life and my company on ideas that have been stress-tested across centuries than on whatever happens to be fashionable in the moment.There is another word I think about often: pannben.It is Swedish. Literally “forehead bone.” In practice it means grit. Stubbornness. Refusal to quit.I found running in my thirties. I was not a natural runner. But I locked in. Eventually I ran a 10k in 36 minutes. Then an ultra marathon of 90 kilometers. Not because it came easily, but because once I decide to do something, I keep going.That matters in startups more than people admit.Talent matters. Intelligence matters. But in hard things, pannben matters more. The willingness to keep going when it hurts. To do the repetitions. To come back the next day.For over a year now, I have been waking up at five in the morning. Exercise. Journaling. Reading. Not because it is enjoyable in a Swedish winter, but because it compounds.Compounding is one of the most powerful forces I know.Small improvements are almost invisible day to day. But over time they separate people, teams, and companies. Individual discipline compounds into personal capability. Personal capability compounds into team performance. Team performance compounds into competitive advantage.That is how small teams become disproportionately effective.And health is part of that. You cannot compound if your body is failing. You cannot make sharp decisions if you are exhausted, under-recovered, and living on caffeine. Sleep, exercise, real food — this is not soft advice. It is infrastructure. Your body is the hardware everything else runs on.The team is the moatI do not want people on this journey who are here only for a paycheck.That is not a moral statement. It is a practical one. Life is too short to spend your working hours on something you do not care about. The best work comes from genuine engagement — the kind where you cannot stop thinking about the problem because it bothers you that it is not solved yet.I still think tinkering is one of the most enjoyable things in the world. That has never gone away. Whether it was optimizing heating systems in my first house, building backup power, buying an early EV, mining Ethereum, importing the first iPad, building a 3D printer, or automating a home — it is all the same impulse. Go deep. Understand the system. Make it work.That joy matters.Because you cannot grind people into excellence. I have seen attempts. It does not work. Sustainable excellence comes from people who genuinely care.A few years ago I read The Hard Thing About Hard Things by Ben Horowitz, and it resonated because I was already living it. Leadership is not about applying a neat framework. It is about making hard calls, often between bad options, and then carrying those decisions with integrity. Leadership is earned every day. It is not contained in a title.And this is why I believe something that sounds counterintuitive but feels increasingly obvious: in the age of AI, the human team matters more, not less.AI makes execution more available. Code generation, analysis, pattern recognition — increasingly these are commodity capabilities. If everyone has access to similar tools, what matters more?Judgment. Taste. Conviction. Trust. Culture. The ability to disagree productively and still commit together.That is why I say the team is the moat.Not the technology. Not the tooling. The people, and how they work together.What I’m askingSo here is what I am asking.Bring your full self to this. Not the polished version that performs well in meetings. The real version. The one that actually cares.Get a little better every day. Not dramatically. Just enough to let it compound.Speak up when something is wrong. Problems do not get cheaper with silence. In a startup, silence is one of the most expensive things there is.Hold each other accountable. Hold me accountable too. I do not want to be surrounded by people who tell me what I want to hear. I want people who tell me what I need to hear.And enjoy the building. Seriously. If there is no joy in solving hard problems with smart people, something is off.Focus on what we can control. The market will behave however it behaves. Competitors will do what they do. We do not control that. We control how sharp we are. How quickly we move. How honestly we work together.That is the philosophy.Everything else is execution.Why this mattersWe started Sourceful because we believe the energy system should work for people, not the other way around.We believe abundance, not austerity, is how we solve the climate problem.We believe open, distributed, locally intelligent infrastructure is the path there.I do not know exactly how this ends. Nobody does.But I do know this: the people building this, the team around it, and the builders who will join this platform have a real chance to create something that matters. Something durable. Infrastructure. The kind of thing that, once it exists, is very hard to replace.That is worth getting up at five in the morning for.That is worth the hard days.That is worth building.Let’s build it. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

Picture a spectacularly beautiful, cloudless spring day in California. Millions of solar panels across the state are operating at peak efficiency, churning out massive amounts of clean, zero-marginal-cost electricity. But behind the scenes, grid operators are frantically hitting the kill switch. They are actively disconnecting solar farms and shoving millions of dollars worth of pure energy straight into the dirt.Why? Because the power has nowhere to go. The electrical grid has no brain to route it, and no way to store it for when the sun goes down.In this episode of the podcast, we do a deep dive into the defining engineering challenge of our time. We explore why the greatest energy transition in human history isn’t bottlenecked by the physical hardware we put on roofs, but by the invisible information layer—the software—required to orchestrate it.Here are the core themes we unpack in the episode:1. The Hardware War is Already Over (And We Won)For decades, the systemic bottleneck of the grid was the sheer cost of generating power. That era is officially over. The manufacturing of solar panels and batteries has scaled so aggressively that the physical components are now effectively commoditized.* In just the first half of 2025, the world installed 380 gigawatts of solar capacity—the equivalent of 380 massive nuclear reactors.* Solar module prices have plummeted to a global average of roughly $0.10 per watt.* Stationary lithium-ion battery pack prices collapsed by 45% in a single 12-month period, hitting an astonishing $70 per kilowatt-hour.* Raw cell prices for Lithium Iron Phosphate (LFP) chemistry—the safest and most dominant chemistry for stationary storage—have dropped to just $36 per kilowatt-hour.2. The Symptoms of a “Dumb” GridBecause energy generation has become cheap and abundant, we’ve completely inverted the bottleneck. It is now entirely about system coordination. When you generate electricity, it must be consumed or stored the exact millisecond it is created. If the grid can’t handle it, operators are forced to curtail (throw away) the power to prevent the system from physically melting down.* In 2024, California curtailed 3.4 million megawatt-hours of renewable energy (93% of which was solar).* In the first part of 2025, Germany curtailed a record 1,750 gigawatt-hours and experienced 575 hours of negative day-ahead prices.* In Spain, solar capture prices collapsed from €61 to just €16.80 per megawatt-hour.This creates a self-reinforcing cycle of grid stress. The cheap solar drives new demand (a classic example of the Jevons Paradox, first observed in 1865), particularly from power-hungry AI data centers that demand 24/7 reliability.3. A Tale of Two BlueprintsHow do we solve this? The industry is currently split between two radically different architectural philosophies:The Centralized approach (The Mainframe): This is the top-down, massive infrastructure method. A prime example is the landmark project underway in Abu Dhabi spearheaded by Masdar. They are pairing a colossal 5.2-gigawatt solar plant with a 19-gigawatt-hour battery system. The goal is to use advanced “grid-forming inverters” to make intermittent solar behave exactly like a reliable, baseload nuclear plant.The Decentralized approach (The Internet):This is the bottom-up approach, relying on Distributed Energy Resources (DERs). This involves aggregating hundreds of thousands of small, distributed assets—like residential batteries, smart thermostats, and EV chargers—into massive “Virtual Power Plants”. We discuss movements like DePINs (Decentralized Physical Infrastructure Networks) and how protocols use cryptographic tokens to incentivize hardware deployment.4. The ZAP Gateway & The Path ForwardTo make the decentralized model work without causing physical blackouts, we need a universal translator at the edge of the network. In the episode, we discuss the “ZAP gateway”. This is a sub-€100 piece of hardware, built on the ESP32-C3 chipset, that acts as the physical bridge between chaotic software commands and fragile local grid hardware. It uses “controls forward design” to ensure that even if the internet goes down, the local battery safely manages its own limits without relying on a centralized cloud brain.5. The Regulatory SlogFinally, we touch on the painful reality of policy. While the software and hardware are ready today, regulatory frameworks are dragging their feet. In the US, FERC Order 2222 (which mandates that distributed energy can compete in wholesale markets) faces implementations delayed to 2028 or even 2030 by major grid operators. The EU faces similar sluggishness with Directive 2019/944.We have millions of the world’s fastest computers, but we are still trying to invent the router to connect them all together. That is the mission.🎧 Listen to the full episode now to hear the complete breakdown. Let me know your thoughts in the comments below—are we moving fast enough, or will the regulations keep us in the dark ages?— Fredrik This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

It’s fall 1999. A venture capitalist named John Doerr walks into a cramped office in Mountain View — thirty employees gathered around a ping-pong table that doubles as a boardroom. The company is called Google, and it’s the eighteenth search engine to arrive on the web. Doerr carries an $11.8 million check in one hand and a management system in the other. The check made Google possible. The management system made Google Google.This episode is a deep dive into that system — Objectives and Key Results — and the book that lays it all out: John Doerr’s Measure What Matters.What Are OKRs, Really?The story starts at Intel in the 1970s. Andy Grove, the legendary CEO, was staring at a crisis: the Japanese semiconductor industry was about to eat Intel alive. Grove needed every single person in the company to understand the mission and execute on it — fast. His answer was a deceptively simple framework. An Objective is what you want to achieve. Key Results are how you’ll know you’ve achieved it. That’s it.But the magic isn’t in the framework. It’s in what happens when you make everyone’s goals visible — from the newest intern to the CEO. Grove’s quote haunts the entire book: “There are so many people working so hard... and achieving so little.”The Four SuperpowersDoerr identifies four things that OKRs unlock, and we spend most of the episode working through each one with real stories:Focus. Brett Kopf built Remind, an education app, but the company was drowning — trying to do everything at once. OKRs forced them to say no. They picked three objectives and killed everything else. The company survived.Alignment. MyFitnessPal grew to 100 million users with a tiny team. How? Everyone’s OKRs were transparent and cascaded from a single mission. No one wondered what they should be working on.Tracking. Bill Gates brought OKRs to the Gates Foundation to track something most organizations struggle to measure: progress against malaria deaths. You can only improve what you measure.Stretch. YouTube set a goal of one billion hours of watch time per day. They were at 100 million. That’s a 10x target — the kind of number that makes you stare at a whiteboard and question your sanity. They hit it.Beyond Goals: The Death of the Annual ReviewThe second half of the book — and the episode — tackles something less glamorous but arguably more important: Continuous Feedback and Recognition (CFRs). Annual performance reviews are broken. A doctor changes a treatment protocol in January and doesn’t see outcome data until December. A software engineer ships a feature and gets feedback six months later. CFRs replace this with continuous, lightweight conversations.The examples from Zume Pizza and Lumeris (a healthcare company where lives are literally at stake) make the case that faster feedback loops aren’t just nicer — they’re a requirement when the world moves fast.OKRs as Coordination InfrastructureHere’s where I get personal. I run an energy infrastructure company. Every day I think about the same problem Andy Grove faced at Intel: how do you get thousands of distributed actors — solar panels, batteries, EVs, heat pumps — to coordinate toward a shared objective?The energy grid must balance supply and demand every single second. It’s a coordination problem at continental scale. And the answer, I think, starts with something Grove figured out in a semiconductor factory fifty-five years ago. Not more generation. Not bigger wires. Shared objectives with measurable key results, propagated across every node in the network.Measure what matters.Key Takeaways* OKRs are not a goal-setting exercise — they’re a coordination protocol for organizations (and systems) of any size* The four superpowers — Focus, Alignment, Tracking, Stretch — compound on each other* Transparency is the mechanism: when everyone can see everyone else’s goals, alignment happens organically* Stretch goals (10x, not 10%) change how people think about problems, not just how hard they work* Annual reviews are dead — continuous feedback is the faster loop that modern systems require* Coordination infrastructure — whether for companies or energy grids — requires shared, measurable objectivesBehind the Scenes: A New FormatThis episode is an experiment. It’s the first episode of Coordinatedproduced in what I’m calling the Radiolab format — a multi-voice, multi-track production that feels closer to audio documentary than traditional podcasting.Instead of a solo monologue, this episode features four voice layers:* Daniel — the primary host, driving the narrative with a steady British broadcaster voice* Matilda — the co-host and audience proxy, bringing energy and genuine reactions* Fredrik (narrator) — stepping in for authoritative narration and synthesis* Fredrik (expert) — the same voice but with a “tape” EQ filter, used for quoting Doerr, Grove, and other real people. It creates the feel of archival audio.How It Was BuiltThe entire production pipeline is automated — from book to finished episode. Here’s what happens under the hood:Pass 1 — The Episode Bible. The full text of Measure What Matters(~72,000 words) is split into 21 chapters. Each chapter is fed individually to Claude Opus, which extracts the most compelling stories, quotes, and emotional beats. The result is a ~28,000-word “episode bible” — a structured research document.Pass 2 — Script Generation. The bible is fed back to Claude in five separate passes (one per ACT), each with specific creative direction: the hook, the superpowers (split across two passes), the revolution, and the synthesis. Claude generates a structured JSON script with 458 segments, each tagged with speaker, production cues, and timing.Production Cues. Every segment carries metadata that controls the final mix: - Stereo panning — Daniel sits slightly left, Matilda slightly right, narrator centered - Tail-stepping — some reactions start 150-300ms before the previous line ends, creating that signature Radiolab crosstalk energy - Hard cuts — four moments in the episode where ALL sound drops to pure silence before a key revelation - Music beds— four mood tracks (tension, uplifting, contemplative, energetic) scored under narration sections - EQ presets — expert quotes get a “tape” filter (300-4000Hz bandpass) that makes them sound like archival recordingsAudio Generation. All 458 segments are rendered through ElevenLabs’ text-to-speech API, each with per-voice model selection and tuned voice settings. The stems are then assembled on a timeline — not sequentially concatenated — with overlaps, fades, panning, and music beds mixed in.Mastering. The raw mix goes through dynamic range compression and two-pass loudness normalization to -16 LUFS (the podcast standard for Spotify and Apple Podcasts).The result: 75 minutes of produced audio from a single text file input. The entire pipeline — from book to finished MP3 — runs with one command.I’m excited to keep iterating on this format. The bones are there. Now it’s about tuning the voices, the pacing, and the music to make it feel less like AI and more like radio.Full transcript available below the audio player. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

In February 2026, Y Combinator CEO Garry Tan published a short essay arguing that “don’t boil the ocean” — the most common piece of startup advice — is now obsolete. His reasoning rests on a chain of ideas stretching back 160 years, through a Victorian economist, a radical architect, and two professors betting on the price of tin. I followed that chain. It changed how I think about everything.The Paradox That Triggered a Coal PanicIn 1865, a 29-year-old English logician named William Stanley Jevons published The Coal Question. His central observation was so counterintuitive that economists are still debating it 160 years later:“It is wholly a confusion of ideas to suppose that the economical use of fuel is equivalent to a diminished consumption. The very contrary is the truth.”Making fuel more efficient does not reduce consumption. James Watt’s steam engine used 75% less coal than the Newcomen engine for the same work. And yet British coal production grew 3.5% per year for 80 straight years — from 5.2 million tons in 1750 to 292 million tons at peak in 1913. A 56-fold increase.Watt’s engine didn’t conserve coal. It made coal-powered work so cheap that steam engines migrated from mine pumps to cotton mills to railways to steamships. Every new application that became economically viable created demand that overwhelmed the efficiency gains.This is the Jevons Paradox. And it has repeated with eerie precision wherever a fundamental resource has gotten dramatically cheaper.14,000x Cheaper Light, 6,500x More ConsumptionEconomist William Nordhaus traced the cost of artificial light from the 1300s to the present. One million lumen-hours cost the equivalent of £40,800 in the 1300s. By 2006: £2.90. A 14,000-fold decline.Did humanity respond by consuming the same amount of light? The average UK resident in 2000 consumed 6,500 times more artificial light than in 1800. Computing tells the same story — cost per gigaflop dropped from $18.7 million in 1984 to three cents in 2017. Total compute consumed went up by orders of magnitude. Storage, communication, bandwidth — same pattern, every time.The cheaper the resource, the more we consume. As long as human desire for it is elastic, Jevons holds.Fuller’s Arc: From Stone to NothingWhile economists debated the paradox, Buckminster Fuller was watching the same phenomenon through a different lens. He called it “ephemeralization” — doing more and more with less and less until eventually you can do everything with nothing.Stone arches. Iron trusses. Steel cables. Wireless signals. Each generation of bridge required less material. But the number of connections, the volume of communication, the scope of what bridges enabled — that exploded beyond imagination.Fuller and Jevons aren’t contradictions. They’re two sides of the same process. We do more with less of this specific thing (Fuller’s insight), which makes the underlying capability so cheap that we consume vastly more of it (Jevons’ insight).The Ehrlich-Simon BetIn 1980, Stanford’s Paul Ehrlich — who predicted hundreds of millions would starve in the 1970s — bet economist Julian Simon $1,000 that five metals would rise in inflation-adjusted price over ten years. Ehrlich picked chromium, copper, nickel, tin, and tungsten.All five declined. Tin fell 55%. Tungsten dropped over 60%. The $1,000 basket was worth $423.93 by 1990 — despite world population growing by 800 million, the largest single-decade increase in human history. Scarcity incentivized innovation, substitution, and new discoveries. Ehrlich mailed a check for $576.07 and never bet on resource prices again.Intelligence Is Collapsing Faster Than Anything Before ItNow there’s a resource collapsing in cost that makes coal, light, and metals look like gentle declines. That resource is intelligence.Epoch AI found that LLM inference prices are falling between 9x and 900x per year, with a median of 50x. Since January 2024, that accelerated to 200x per year. GPT-3.5-level performance dropped 280-fold in cost — from $20 per million tokens to 7 cents — in two years.And here’s the Jevons effect in real time: despite per-token costs falling 280x, total inference spending grew 320% over the same period. Researchers found super-elastic demand — a 1% price decrease drives a 1.42% volume increase. Price times quantity riseseven as price falls. Demand for intelligence is, in the paper’s language, “currently un-satiated.”No previous resource — not coal, not electricity, not computation — has become cheap this fast.This Is PersonalI spend every day thinking about what happens when energy becomes the bottleneck. When intelligence is cheap but you need electricity to run it. When every home has solar and batteries and heat pumps and EVs, but nobody has figured out how to coordinate them.Germany had 457 hours of negative electricity prices in 2024. Generators paying people to take their power. Jevons would have recognized it instantly — we made electricity generation incredibly efficient, and demand exploded into categories nobody anticipated: data centers, EVs, heat pumps, AI inference clusters drawing megawatts.Tan’s essay distinguishes two responses to this moment. The zero-sum response: use AI to do the same thing cheaper, cut headcount, eke out 5% efficiency gains. The positive-sum response: attempt things that were previously impossible.Applied to the energy grid — the zero-sum response is building more solar panels and bigger batteries. Throwing hardware at the problem. The positive-sum response is asking: what if every home was a power plant? What if every battery was a grid asset? What if every EV was a node in a distributed network that could balance the grid in 200 milliseconds, from the edge, locally?The Ocean Is Already BoilingIf Jevons is right — and 160 years of evidence says he is — the demand response to near-free intelligence will be proportionally extraordinary. The organizations and founders who raise their ambitions rather than protect their incumbency will define the next era. The ones who respond with zero-sum fear will find that the ocean is boiling around them whether they like it or not.Fuller, who died in 1983, anticipated this moment: “Ephemeralization trends towards an ultimate doing of everything with nothing at all.” Intelligence approaching zero marginal cost is the logical terminus of his arc from stone to wireless.The question is no longer whether this is happening. It’s whether we’ll use the moment to boil oceans — or drown in committee.Key Takeaways* The Jevons Paradox — making a resource more efficient doesn’t reduce consumption, it unleashes demand. Coal, light, computing, and storage all followed this pattern over 160 years.* AI inference costs are collapsing faster than any previous resource: 280x cheaper in two years, accelerating to 200x per year. Yet total spending on inference grew 320%, showing super-elastic demand.* Jevons and Fuller describe two sides of the same process: we do more with less material (ephemeralization) while consuming vastly more of the underlying capability (the paradox).* The energy grid is the next Jevons battleground — cheap solar created 457 hours of negative prices in Germany, and AI-driven demand is exploding. The positive-sum response isn’t more hardware, it’s coordination infrastructure at the edge.Full transcript available below the audio player. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

This episode is different. Instead of talking about the grid, energy markets, or coordination protocols, I’m talking about this podcast — how it’s made, why I rebuilt the entire production pipeline from scratch, and what it taught me about the relationship between tools and output.The NotebookLM EraHere’s what used to happen: I’d spend a week reading papers, articles, talking to people. Then I’d dump everything into NotebookLM, hit generate, and get a thirty-minute episode. Every time. Thirty minutes. No control over length, no control over focus, and zero memory of what I’d already covered three episodes ago.It was fine. But it wasn’t mine. It didn’t sound like me. Every episode existed in isolation — no arc, no continuity. Just... content.And if you know me at all, you know that drives me absolutely crazy.Physics Before Code (Yes, Even for Podcasts)We say this at Sourceful constantly: physics before code. The physical reality of the problem has to drive the architecture. Not the other way around.So I asked myself — what’s the physics of a podcast?A podcast is a conversation. A human being talking to other human beings. That’s the constraint. Everything else is infrastructure.Once I framed it that way, the architecture became obvious. Research goes in. A script comes out — not generated audio, a script. Something I can read, edit, argue with. The script is the single source of truth. Audio, transcript, blog post — all derived from that one artifact.One command. Research in, episode out.The Memory ProblemHere’s the part that gets me genuinely excited: the system has memory.It knows what we talked about in episode twelve. It knows which topics we’ve beaten to death and which ones we’ve barely touched. When it generates a new script, it has the context of every previous episode.That means I can say: give me a ten-minute episode on grid storage, connect it to what we said about frequency regulation in episode forty-two, and don’t repeat the EV stuff from last week. And it does that. Because it has the memory.There’s an obvious irony here. I spend all day building local-first coordination infrastructure for the energy grid — systems with memory, context, and local intelligence. And then I was going home and using a cloud-only, no-memory, no-control tool to make my podcast.The cobbler’s children have no shoes, right? Not anymore.Fifteen Iterations to Sound Like MyselfI should be honest about the process, because it wasn’t “write code, done.” It was deeply iterative.The first version sounded terrible. Flat. Robotic. Like a GPS navigation system reading my thoughts.So I started tweaking. Voice stability settings. Speed. Silence between segments. Which model to use for synthesis. It turns out that AI voice cloning is incredibly sensitive to these parameters. Too much “style” and my voice drifts into accents I’ve never had. Too little speed and I sound sedated. The wrong model and prosody falls apart completely.It took about fifteen iterations to get to what you’re hearing now. It’s still not perfect. But it’s mine.Tools Shape OutputThere’s a deeper point here that I keep coming back to: the tools we use shape the things we make.If your tool gives you no control, you get generic output. If your tool has memory, your output has continuity. If your tool understands your constraints, your output respects them.We’re not just consuming AI anymore. We’re building with it — giving it context, constraints, memory. Making it an extension of how we think, not a replacement for it.Same Principles, Same ArchitectureThe same shift is happening in energy. From consuming to coordinating.Your solar panels don’t just generate power — they coordinate with the grid. Your battery doesn’t just store energy — it provides frequency response. Your EV doesn’t just drive — it balances load in your neighborhood. Every device becomes a participant, not just a consumer.That’s what Coordinated is about. Not just the energy grid. Not just markets and thermodynamics. It’s about the idea that complex systems need coordination. And coordination requires memory, local intelligence, and respect for the physics of the problem.Whether that problem is balancing fifty hertz across a continent... or making a podcast that actually gets better over time.Same principles. Same architecture. Same obsession with doing it right.This episode was produced entirely by the Coordinated pipeline — script generated by Claude Opus, voice rendered through ElevenLabs, blog post auto-generated, all from one command. The cobbler finally made himself some shoes.Key Takeaways* I rebuilt the podcast production pipeline from scratch because NotebookLM gave me no control over length, focus, or continuity — every episode existed in isolation with no memory of previous ones.* The new system treats the script as the single source of truth. Research goes in, a human-editable script comes out, and everything else (audio, transcript, blog post) derives from that one artifact.* Episode memory changes everything — the system knows what 68 previous episodes covered, so it can build on past conversations instead of repeating them.* The tools we use shape what we make. If your tool has no memory, your output has no continuity. The same principle applies to energy: devices need to be participants with local intelligence, not just dumb consumers on a cloud API.Full transcript available below the audio player. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

In this special episode of Coordinated with Fredrik, we went deep — not at the strategy layer, not at the founder layer, but at the socket level.This was a strict engineering teardown of a single question:Can NATS become the autonomic nervous system of Sourceful Energy?What follows is the architectural synthesis.The Real Problem We’re SolvingAt Sourceful, we are not just operating a backend.We are coordinating:* High-throughput mobile app traffic* A growing mesh of backend microservices* Massive telemetry streams from distributed energy assets* Smart meters* EV chargers* Solar arrays in remote geographies* Wind turbines behind unstable uplinksThis is not a traditional cloud-native architecture.It is cloud + edge + unreliable networks + financial correctness.That requires more than horizontal scaling. It requires:* Isolation of failure domains* Backpressure awareness* Autonomous routing* Dynamic topology adaptation* Edge survivabilityThis is where NATS becomes interesting.The 15 MB Binary That Changes the ConversationNATS runs as a single static Go binary of roughly 15 MB.A single node can handle 15–18 million messages per second.That sounds unrealistic until you understand the engineering choices:Concurrency Model* Lightweight goroutines* User-space scheduling* Massive TCP connection densityMemory Discipline* Zero-allocation parsing* Pointer passing instead of object churn* Minimal garbage collection pausesRouting Philosophy* Pure in-memory message routing* Disk I/O only when explicitly requestedNATS is not a heavyweight enterprise broker.It is a highly optimized, high-throughput routing engine.Selfish Optimization: Protect the System FirstOne of the most controversial ideas in NATS is “selfish optimization.”If a downstream consumer slows down:* NATS does not buffer indefinitely* NATS does not slow producers* NATS drops the connectionFrom a traditional enterprise mindset, that sounds aggressive.But in distributed energy systems, it is correct.If the router collapses:* Telemetry stops* Control signals stop* Billing APIs stop* The entire system failsProtecting the health of the transport layer is non-negotiable.The whole must survive even if individual services fail.Core NATS vs JetStreamNATS separates transient routing from durability.Core NATS* In-memory* Fire-and-forget* Ultra-low latency* No persistenceIf no subscriber exists, the message is dropped.Use this for:* Real-time telemetry* State queries* Fast internal RPCJetStream* Durable streams* Raft-based replication* Replayable consumers* At-least-once / exactly-once semanticsUse this for:* Billing events* Immutable records* Financial correctnessThe key principle:Persistence is opt-in.You only pay for disk I/O when the workload requires it.Raft Without the BottleneckMost distributed streaming systems rely on a single global consensus group.JetStream does something different:* One meta-consensus group for cluster metadata* Independent Raft groups per stream* Even per consumerIf you run 5,000 streams, you run 5,000 independent consensus groups.Why does that not collapse under overhead?Because:* Each Raft group runs as a lightweight goroutine* Heartbeats are batched* Streams are isolatedA spike in one stream does not block the others.This is horizontal scalability at the consensus layer.Subject-Based Routing Instead of IP-Based ThinkingNATS routes by subject strings, not by IP addresses.Example:telemetry.eu.germany.meter80492 Routing is powered by an optimized radix trie.This means:* No regex matching* No linear scans* Logarithmic routing complexitySubject hierarchies become your semantic network.Developers stop thinking about:* Hostnames* Ports* DNS* Reverse proxiesThey express interest in data.The infrastructure routes it.Request-Reply Without HTTPNATS supports request-reply patterns without point-to-point connections.Mechanically:* The requester generates a temporary reply subject* Publishes a message including that subject* A service processes and replies to that subject* The first response winsTo developers, it feels synchronous.Under the hood, it is fully asynchronous and multiplexed.Queue groups provide built-in distributed load balancing.This eliminates internal service meshes and east-west load balancers for microservice communication.Public ingress still requires API gateways.Internal routing becomes dramatically simpler.Global Scaling with SuperclustersInside a region, NATS uses a full mesh cluster.Across regions, it uses superclusters connected by gateways.Gateways operate in interest-only mode.If Europe is not subscribing to US telemetry:* No bytes cross the Atlantic.The moment interest appears:* Flow begins automatically.This prevents blind data mirroring and reduces egress costs dramatically.Leaf Nodes: Edge AutonomyLeaf nodes are where NATS becomes transformative for energy infrastructure.A leaf node:* Runs locally on edge hardware* Initiates an outbound TLS connection to the core* Requires no inbound firewall rules* Multiplexes all traffic over a single connectionIf connectivity drops:* Local JetStream buffers telemetry* Local control systems continue functioning* No data is lostWhen connectivity restores:* The stream synchronizes automatically* Consumers resume from correct offsetsThis enables:Autonomous edge during disconnection.Seamless federation when connected.For EV chargers, solar arrays, and wind turbines, this is critical.Decentralized Security at ScaleTraditional brokers rely on centralized authentication.That becomes a bottleneck at scale.NATS uses:* ED25519 keypairs* JWT-based trust hierarchy* Operator → Account → User modelAuthentication becomes pure cryptographic verification.No database lookups.No external latency.No central auth bottleneck.Permissions are embedded in JWT claims:* Publish rights* Subscribe rights* Data limitsRevocation can be pushed in real time without cluster restarts.For enterprises tied to Okta or LDAP, auth callouts bridge existing identity systems into decentralized JWT issuance.This allows compliance without sacrificing performance.Kafka, RabbitMQ, MQTT — Where NATS FitsKafkaDesigned for:* Durable append-only logs* Analytics pipelines* Data lakesStrength:* Historical retentionTradeoff:* Partition-bound scaling* Consumer rebalancing pauses* Operational overheadNATS:* Dynamic routing* Elastic worker scaling* Lower latency for microservicesRabbitMQDesigned for:* Complex exchange-based routingTradeoff:* Higher operational fragility under partitions* Cluster state complexityNATS:* Simpler subject routing* Gossip for cluster sync* Raft-backed durabilityMQTTBest for:* Constrained IoT devicesNATS does not replace MQTT.It embeds an MQTT broker.MQTT topics are mapped directly to NATS subjects internally.This creates a unified backbone:* Edge devices speak MQTT* Backend services speak NATS* No external translation layer requiredThe Paradigm ShiftFor decades, distributed systems have been built around:* IP addresses* DNS names* Blocking HTTP calls* Explicit service discoveryNATS introduces a different idea:Express interest in a semantic subject.Let an autonomic system route it dynamically.In a world of:* Real-time AI inference* Autonomous energy assets* Fluid containerized workloads* Distributed edge computingIP-based thinking becomes friction.Subject-based thinking becomes leverage.What This Means for SourcefulAdopting NATS is not swapping a queue.It is:* Flattening internal service meshes* Eliminating east-west load balancers* Moving complexity into autonomic infrastructure* Enabling edge-first resilience* Protecting system health by design* Running a global coordination backbone on a single optimized binaryThe goal is operational simplicity.Push complexity into the transport layer.Free engineers to focus on energy optimization logic.If we get this right:The infrastructure becomes invisible.And the grid becomes programmable. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

You know the feeling.You’re low on range. The map says there’s a charger around the corner. You pull up — relief. And then… black screen. Or “out of order.” Or worse: it looks fine, but nothing happens.That moment — that tiny spike of anxiety — is the real enemy of EV adoption.In this episode of Coordinated with Fredrik, we go deep into the engineering side of that problem. Not surface-level EV talk. Not market hype. We unpack the actual backend architecture required to build a charging network that doesn’t break — and more importantly, one that scales.This was a special deep dive made for David and the team at Sourceful Energy. The mission:How do you build a truly robust charging network using OCPP — but do it in a modern, open-source, event-driven way that fits a NATS-based architecture?Why OCPP 2.0.1 Is the Only Serious Choice TodayWe trace the evolution of OCPP from the early SOAP-based days (heavy XML over fragile 2G connections) to the WebSocket revolution of 1.6 — and then to the modern architecture of 2.0.1.Here’s the punchline:* 1.6 works, but it’s messy and vendor-fragmented.* 2.0.1 is structured, hierarchical, and event-driven.* It has native certificate handling.* It supports granular device modeling.* It enables accurate transaction handling.* It’s built for ISO 15118 and plug-and-charge.* And it cleanly prepares you for OCPP 2.1 and bidirectional charging (V2X).If you’re building from scratch in 2026, there’s no serious argument for staying on legacy 1.6 unless hardware forces your hand.The Architecture Question: Don’t Build a CSMS From ScratchThe temptation is obvious: “We’re engineers. Let’s build it ourselves.”That’s a trap.Implementing the full OCPP spec — properly — is a multi-year effort. Edge cases, retries, timeouts, certificate handling, WebSocket state management… it’s a black hole.Instead, we explore the open-source landscape:* The old monolithic Java servers.* Client-side firmware stacks.* And finally: a modern, modular TypeScript-based backend designed for 1.6 and 2.0.1 side by side.The key architectural insight:Use a modular OCPP engine as the edge gateway.Inject a custom NATS adapter.Publish clean, validated events into your own internal event-driven system.Let OCPP parsing and compliance live at the edge.Let your business logic live in your own microservices.That separation is everything.NATS + OCPP 2.0.1 = Clean Topic HierarchiesOCPP 2.0.1’s hierarchical device model maps beautifully to NATS subject structures.Instead of a generic firehose of messages, you can structure topics like:ocpp.v2.station-42.evse1.connector.temperature ocpp.v2.station-42.transactionevent ocpp.v2.station-42.status Now your billing service only subscribes to transaction events.Your analytics service subscribes to metering data.Your ops dashboard listens to error codes.Fully decoupled. Clean. Scalable.This is how you avoid turning your backend into a spaghetti monolith.Reliability Is Not a Feature — It’s the ProductUp to 25% of public chargers can be non-functional at any given time. That’s not a UX issue. That’s a systemic architectural failure.A huge reason: vague error reporting.If your charger only reports “Other error,” your operations team has no choice but to roll a truck. That’s expensive. And slow.In the episode, we talk about:* Mandating standardized granular error codes.* Using compliance tools to verify hardware implementations.* Making reliability a contractual requirement in procurement.Because here’s the uncomfortable truth:The world’s best backend can’t fix bad firmware.Architecture and hardware procurement strategy must align.The Bigger Play: From Charger to Grid AssetWe close the episode with something bigger.OCPP 2.1 introduced proper bidirectional charging (V2X). That means EVs stop being passive loads and start becoming active grid assets.If your backend is:* Event-driven* Granular* Secure* High-throughputYou’re not just running a charging network.You’re laying the foundation for a virtual power plant.And that only works if the software pipes are designed correctly from day one.If you’re building infrastructure — not just apps — this episode is for you.This wasn’t about theory. It was about architecture decisions that determine whether your network becomes resilient, scalable, and future-proof — or collapses under its own complexity.Thanks for tuning in to Coordinated with Fredrik.More deep dives coming. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

It wasn’t for a broad audience.It wasn’t for policymakers or the energy-curious.It was built for one specific person: David — a firmware engineer driving north on the E4 toward Linköping.Long highway. Solo drive. Brain half in traffic, half in state machines.So we decided to go deep into something that actually matters if you’re building charging infrastructure:OCPP — Open Charge Point Protocol.No marketing fluff. No “empowering the green transition.”Just architecture, protocol evolution, firmware headaches, and where this is all going.What OCPP Actually IsStrip it down and OCPP is simple:It is the application-layer language between a charging station (EVSE) and a backend system (CSMS).That’s it.Today, that means:* JSON payloads* WebSockets transport* TLS security* Deterministic state machinesIt is the pipe between hardware and cloud.And if you’re writing firmware, it is the difference between “works in the lab” and “survives production.”The Early Days: SOAP, XML, and PainBefore OCPP 1.6, we lived in the SOAP era.XML everywhere.Heavy envelopes.Verbose messages.Request/response HTTP only.That created a structural problem: chargers sit behind NATs and cellular modems. Servers couldn’t easily push commands down to them.So chargers had to poll.“Do you have work for me?”“Do you have work for me?”“Do you have work for me?”Inefficient. Expensive. Messy.And for embedded systems? Parsing XML on a constrained MCU was not fun.OCPP 1.6J — The WebSocket RevolutionThen came OCPP 1.6J.The “J” mattered.JSON instead of XML.Persistent WebSockets instead of pure HTTP.Bidirectional messaging.Suddenly:* The backend could push commands instantly.* Latency dropped.* Cellular data usage shrank.* Parsing became lighter.For firmware engineers, this was a major quality-of-life upgrade.And 1.6 became everywhere. AC chargers. DC chargers. Public networks. Private networks.It worked.But it wasn’t clean.The Ambiguity ProblemOCPP 1.6 had a philosophical flaw:It was ambiguous.Take a simple case.A user plugs in a cable before authorizing.The charger sends:StatusNotification: PreparingWhat does that mean?* Plug inserted?* Authorized but waiting?* Internal self-check?* Something else?The backend had to infer meaning from sequence patterns.That’s brittle engineering.And when you scale across vendors, those interpretations diverge.OCPP 2.0 → 2.0.1: The Hard ResetOCPP 2.0 tried to fix everything.It introduced a massive architectural shift — including a hierarchical device model.But the specification itself had issues:* Broken schema references* Circular definitions* InconsistenciesYou couldn’t strictly validate against it.So 2.0.1 replaced it.And this is important:* 2.0.1 is not backward compatible with 1.6.* It’s a structural rewrite.* If someone says “OCPP 2.0,” they almost certainly mean 2.0.1.Flat World vs Hierarchical WorldIn 1.6, configuration was flat.Key → Value.Like an INI file with a networking layer.In 2.0.1, everything becomes a tree.Charging Station→ EVSE→ ConnectorAnd beyond that:* Cooling systems* Power modules* Converters* Displays* Locks* SubsystemsEach defined as:* Component* Variable* AttributeThis is no longer just a protocol.It’s a digital twin of the hardware.For firmware engineers, that means:* Internal state model required* Memory allocation planning* Component indexing strategies* Careful RAM managementYou’re not just flipping relays anymore.You’re modeling the machine.Transactions: From Guessing to Determinism1.6 had separate messages:* StartTransaction* StopTransaction* MeterValues2.0.1 consolidates everything into:TransactionEventEach event includes:* Event type* Trigger reasonNow the charger can explicitly say:* CablePluggedIn* Authorized* RemoteStart* EVConnectedNo guessing.No backend inference.Just a deterministic state machine.This is one of the most important improvements in the protocol’s history.Offline Handling: The Reality of Cellular NetworksCellular drops. Forests exist. Power glitches happen.In 1.6:* Local whitelist for RFID* Vague retry behavior* Data sometimes lost* Billing inconsistenciesIn 2.0.1:* Defined queuing behavior* Sequential ordering* Transaction sequence numbers* Explicit retry logicBut here’s the firmware cost:You now need:* Non-volatile message storage* Flash wear-leveling* Circular buffer logic* Message prioritizationThis is where embedded engineering meets infrastructure-grade reliability.Ping Is Not HeartbeatOne subtle but critical distinction:WebSocket pingKeeps the TCP connection alive.OCPP heartbeatConfirms the application logic is running.You can have:* Working ping* Deadlocked firmware loopIf your watchdog logic monitors only socket state, you will miss application failures.Network uptime ≠ system health.Error Handling: Stop Using “OtherError”In 1.6, the error enum was limited.Anything outside predefined values became:OtherErrorOperationally useless.2.0.1 improves this with structured NotifyEvent messages tied to components.Then industry alignment brought standardized Minimum Required Error Codes (MRECs).Instead of “OtherError,” you can send precise fault codes for:* Ground fault* Pilot failure* UI failure* Network outageThat difference saves unnecessary truck rolls.And truck rolls are expensive.Security: From Naive to CryptographicSecurity evolved dramatically.Profile 1:Plain HTTP. Not acceptable.Profile 2:TLS — server authentication.Profile 3:Mutual TLS — charger and server authenticate each other.2.0.1 improves:* Certificate lifecycle management* CSR handling* Renewal flowsAdd secure firmware updates with signature verification, and now you have something defensible.Without that, your charger is just an exposed Linux box with high voltage attached.ISO 15118 and Plug & ChargePlug & Charge requires certificate exchange between:* Vehicle OEM* Mobility operator* Charging infrastructureIn 1.6, this required awkward tunneling and vendor extensions.In 2.0.1, it is native.If you are building serious DC fast chargers today, 2.0.1 is not optional.OCPP 2.1: The Grid Starts MovingReleased in January 2025, OCPP 2.1 extends 2.0.1 and adds:* Bidirectional charging (V2X)* DER integration* Dynamic payment support* Secure QR-based session flowsThis is where it gets interesting.When vehicles can discharge power:* V2H (vehicle-to-home)* V2G (vehicle-to-grid)The charger becomes a coordination node.And the backend becomes an orchestrator of distributed batteries.At scale, that means millions of mobile storage units participating in grid balancing.The physical charger becomes infrastructure plumbing.The real asset becomes the battery — moving down the highway.The Bigger PictureThe shift from 1.6 to 2.0.1 and 2.1 is a shift:From ambiguity → determinismFrom flat config → digital twinFrom best effort → structured reliabilityFrom charging → energy orchestrationWe are moving from “start a charging session” to:“Coordinate distributed energy assets dynamically across the grid.”And that is a completely different future.The car is no longer just consuming electricity.The car is becoming part of the grid.David, if you’re reading this — or listening back —May your TLS handshakes succeed.May your heap never fragment.May your sequence numbers stay ordered.And may your latency be low and your uptime high.See you in the next deep dive. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

The textbook version of how the internet came to be is clean. A few smart people in California connected some computers, email happened, and then we got cat videos. That version is wrong. Or rather, it is so sanitized that it misses what actually makes the story worth telling.I spent this episode going deep on the real source material. DARPA archives, technical memos from the Internet Society, oral histories from the engineers who were actually in the room. And what came out is a story that is far more chaotic, far more human, and far more terrifying than the version you got in school.The fear that started everythingThe internet does not begin with a desire to share information. It begins with the end of the world.Early 1960s. The Cuban Missile Crisis is fresh in everyone’s memory. The United States and the Soviet Union are sitting on arsenals that can flatten civilization several times over. And in the Pentagon, the generals have a very specific nightmare that has nothing to do with the bombs themselves. It is about what happens after the first bomb lands.The question was simple: if the president picks up the phone to order a counter strike, does the phone actually work?The entire US communications infrastructure at the time ran on the AT&T telephone network. Circuit switched. When you called someone, the system physically connected a series of copper wires and mechanical switches all the way across the continent. You were essentially renting a very long wire for the duration of your conversation.The problem was the switching stations. All those wires funneled through major hubs in major cities. Chicago. St. Louis. Denver. Cities that happened to be top tier targets for Soviet missiles. Take out the switching station in St. Louis and you did not just lose St. Louis. You severed the connection between the entire East Coast and the West Coast.Efficient for peacetime. Incredibly brittle for war.Paul Baran and the fishnetThis brittleness is what brought Paul Baran into the picture, and this is a name that really should be on statues. Baran was an engineer at the RAND Corporation, the think tank tasked with thinking about the unthinkable. Around 1960, he started working on the survivability problem.The result was massive. Eleven volumes of what became the On Distributed Communications memorandum. In those papers, he drew three network topologies that remain the most important diagrams in the history of computing.Centralized: a bicycle wheel. All spokes connect to one hub. One bomb, game over.Decentralized: a few bicycle wheels connected to each other. Better. But you can still isolate large chunks of the network by hitting the right nodes.Distributed: a fishnet. No center. No hubs. Every node connected to several of its neighbors. A mesh.And then Baran did something that separates the thinkers from the engineers. He did not just draw it. He proved it mathematically. He ran Monte Carlo simulations where he virtually destroyed nodes in his mesh. He simulated a nuclear attack on his own design.The results were staggering. Even if you destroyed 50% of the nodes, literally wiped half the map off the face of the earth, the remaining nodes could still maintain significant connectivity. That is deeply counterintuitive. You would think losing half the network would kill the whole thing. But the beauty of redundancy is that if the direct route is gone, the message goes left, then down, then right, then up. As long as some path exists anywhere, the message gets through.But this created a new problem. In the old system you had a dedicated wire. You knew the exact path. In a fishnet where the path changes constantly and cities might get nuked mid-sentence, how does the information know where to go?Hot potato routing and the invention of packetsBaran realized you could not use analog voice anymore. You could not send a continuous stream. If the path breaks, the stream breaks. You have to digitize it. Chop it up. He proposed taking data and cutting it into tiny digital chunks of 1,024 bits each. He called them message blocks.Then he introduced a concept he called hot potato routing. A guy working on apocalyptic nuclear survival strategy called it hot potato. And it is the perfect description. Imagine each node is a person. You hand them a message block. They do not store it. They do not think about it. They look at the address, look at their neighbors to see who is still alive and not busy, and they throw it to the best option immediately. Get it out of here.If that neighbor gets destroyed a millisecond later, the next potato goes to a different neighbor. No central commander needed. The network heals itself, packet by packet, node by node.Now here is where the story gets strange. While Baran is doing this work in Santa Monica, fueled by Cold War paranoia, a physicist named Donald Davies at the National Physical Laboratory in the UK is independently arriving at almost the exact same solution. But Davies was not thinking about bombs. He was thinking about efficiency.Davies realized that the phone system was a terrible fit for computer data. Phone calls are continuous. Computers are bursty. You type a command, burst of data, then sit there thinking for twenty seconds. Silence. Then the computer replies. Burst. If you are renting an entire highway for that, you are driving one car down it every ten minutes.Davies proposed chopping data into chunks and weaving different conversations together on the same wire. Like shuffling a deck of cards. And he gave us the word we still use: packets.The wild part is that Baran and Davies, working in total secrecy from each other on different continents with completely different motivations, both settled on 1,024 bits as the optimal packet size. The physics of information transfer pushed them both to the exact same destination. It was not just invented. It was discovered.From theory to refrigeratorsTheory is cheap though. Building the thing required money, and it required a guy named JCR Licklider. Licklider was a psychologist, which matters. He was not just a hardware person. He was interested in how humans and machines interacted. In 1962 he took over the computer research program at ARPA, and he brought a philosophy that was alien to the military. He did not see computers as ballistic calculators. He saw them as communication devices.In 1963 he wrote a memo to his colleagues with a title that still gives me chills: To the Members and Affiliates of the Intergalactic Computer Network. This man was dreaming of the cloud fifty years too early.Licklider was the evangelist. He passed the torch to Bob Taylor, who got the project approved for the most mundane reason imaginable. Taylor had three terminals in his Pentagon office, each connected to a different mainframe at a different university. None of them talked to each other. He had to physically roll his chair between them.He reportedly said: “Man, it is obvious what to do. If you have these three terminals, there ought to be one terminal that goes anywhere you want to go.”That frustration sparked the ARPANET. Taylor hired Larry Roberts as chief architect, but Roberts immediately hit a wall. The universities he wanted to connect all ran different computers speaking different machine languages. And the universities were hostile to the idea. They did not want to give up their precious computing cycles to run experimental network software.The solution came from Wesley Clark: do not ask the mainframes to run the network. Build a smaller, separate computer to handle the traffic. They called it the IMP, the Interface Message Processor.ARPA sent out a request for quotation to 140 companies. IBM laughed at them. The giants of the industry thought packet switching was unstable. Only 12 bids came back. The winner was a small consulting firm in Cambridge, Massachusetts: Bolt, Beranek and Newman. BBN.The IMP they built was a modified Honeywell DDP-516. A steel refrigerator you could probably drop off a moving truck and it would still boot up. Inside that steel fridge: 12 kilobytes of memory. The code that ran the entire early internet: 6,000 words of assembly language. No room for bloat.And the best side note from this era: when BBN won the contract, Senator Ted Kennedy sent them a congratulatory telegram. He had misread Interface Message Processor as Interfaith Message Processor. Considering they were getting a Honeywell to talk to an IBM, it was basically a religious miracle.LoOctober 29, 1969. Boelter Hall, UCLA, Room 3420. A windowless room filled with the hum of cooling fans. Charlie Klein, a 21 year old grad student, sits at a terminal. 350 miles away at the Stanford Research Institute, Bill Duvall sits at another. They are coordinating over a regular telephone.Klein types L. “Did you get the L?” Duvall checks: “Got the L.”Klein types O. “Got the O.”Klein types G. The system crashes. Buffer overflow on the Stanford side.The first message ever sent on the internet was “LO.” Like lo and behold. It was accidental poetry. The telegraph got “What hath God wrought.” The telephone got “Mr. Watson, come here.” Both scripted and rehearsed. The internet began with a crash and a fragment.Feels honest, somehow. The internet is messy. It started messy.They fixed the bug within an hour. By December they had four nodes running.The constitution of cyberspaceThe original ARPANET ran on a protocol called NCP. It was designed for a world where ARPANET was the only network and all the hardware was trusted. But by the early 70s, other networks started appearing. Satellite links. Radio networks in Hawaii. And none of them could talk to each other.Enter Vint Cerf and Bob Kahn. In 1974 they published a paper that still runs the world: A Protocol for Packet Network Intercommunication. They laid out four ground rules that I think of as the constitution of cyberspace.Each network stands on its own. You do not have to change your internal workings to connect. Just speak the common language at the gateway. Come as you are.Best effort communication. The network does not guarantee delivery. If a packet drops, the network does not panic. The sender realizes it did not arrive and sends it again. Push the responsibility to the edges.Stateless gateways. The routers do not remember anything about the conversation. They just look at the address and move the packet. This keeps them simple and fast.No global control. No president of the internet. Decentralized governance to match the decentralized topology.This philosophy gave birth to TCP/IP. IP is the envelope with the address. TCP is the certified mail receipt that puts packets in order and asks for missing ones to be resent. And then UDP, for when speed matters more than perfection. You would rather have a tiny audio glitch than a three second pause while your computer retrieves a lost millisecond of sound.The big migration happened on January 1st, 1983. Flag Day. Every node on the ARPANET had to switch from NCP to TCP/IP or get cut off. The engineers were so proud they made buttons: “I survived the TCP/IP transition.”The basements and the blizzardsWhile DARPA and the universities were building the official internet, something else was happening in basements and bedrooms. The great blizzard of 1978 in Chicago. Ward Christensen and Randy Suess, members of a local computer club, were totally snowed in. So they wired a computer to a phone line and wrote software that let people leave messages for each other digitally. The Computerized Bulletin Board System. The first message board.Throughout the 80s, thousands of BBSs popped up. But they were local. Long distance calls were expensive. Then Tom Jennings came along with FidoNet, the “poor man’s internet.” The BBSs would call each other at 4 AM when phone rates dropped to rock bottom, blast out all the mail in a compressed file, grab the incoming, and hang up. A message from LA to New York might take three days to hop across the country. But it was nearly free.And it created a unique culture. You could not have instant flame wars because your insult would not arrive until tomorrow. A slower, more thoughtful social network. FidoNet grew to 40,000 nodes and millions of users. No government funding. No central control. People desperate to communicate will build the infrastructure themselves.Scaling crises, sharks, and napkinsAs the internet grew, it started breaking.First crisis: the phone book. In the beginning there was a single text file called hosts.txt kept at SRI. It listed every computer on the internet and its IP address. If you added a machine, you had to call SRI during business hours and ask them to update the file. By 1983 this was obviously absurd. Paul Mockapetris invented DNS, the distributed domain name system. Delegated authority. You manage your own house, the system just knows how to find your front door.Second crisis: routing. The Border Gateway Protocol was literally designed on two napkins at an IETF meeting in Austin, Texas in 1989. Two engineers from Cisco and IBM sketching out how the global internet would work over lunch. And BGP runs on blind trust. If a network says “I know the way to Google,” the other networks just believe it. Which is how Pakistan accidentally took YouTube offline for the entire planet in 2008 through a simple configuration error that propagated globally via BGP.And then there are the sharks. The first transatlantic fiber cable went in 1988. It kept having faults. When they pulled the cable up from the ocean floor, they found teeth marks. Sharks have organs called ampullae of Lorenzini that detect tiny electrical twitches of fish muscles. The copper conductor running alongside the fiber to power the repeaters was generating electromagnetic fields. The sharks confused the internet with a dying fish and attacked. Google eventually had to wrap their transpacific cables in Kevlar.We built the most advanced communication system in human history and nature tried to eat it.The internet also nearly drowned itself. In October 1986, the connection between Lawrence Berkeley Lab and UC Berkeley, two sites 400 yards apart, dropped from 32 kilobits per second to 40 bits per second. You could tap Morse code faster. A vicious feedback loop: the network gets busy, a packet gets delayed, the sender assumes it is lost and sends a duplicate, doubling the traffic, making congestion worse, causing more duplicates. The internet shouting over itself, drowning in its own echoes.Van Jacobson fixed it by applying fluid dynamics. Packets are like water in a pipe. You can not force more in than the pipe can hold. His slow start and congestion avoidance algorithms for TCP taught computers to calm down and wait for acknowledgment before sending more.The dumb pipe that enabled everythingIf I had to distill this entire history into one lesson, it comes back to the end-to-end principle. The idea that the network should be dumb. A simple pipe. The intelligence lives at the edges, in your laptop, in the server.This is the only reason the web exists. Tim Berners-Lee did not have to ask permission to invent the World Wide Web. He did not have to ask the phone company to upgrade their switches to support hypertext. He just wrote software for the endpoints. The network carried it without knowing what it was.Permissionless innovation. If we had built a smart network optimized purely for voice, the way AT&T originally wanted, we would never have gotten Netflix or Zoom or Bitcoin. The network would have rejected them as improper traffic. The dumb network allowed for brilliant ideas.The uncomfortable endingBut here is the thing that keeps me up at night. We started with a mission to build a network that could survive nuclear war. A distributed, decentralized mesh. Technically, we succeeded. The protocols are decentralized.But look at the layer above. Most of the internet’s traffic flows through a tiny handful of companies. If Amazon Web Services goes down, half the internet breaks. We built a system that is technically distributed but commercially centralized. We replaced the AT&T switching hubs with massive data centers in northern Virginia.The house of cards problem. We might have just rebuilt it with better branding.We solved the engineering problem. We have not solved the human tendency to centralize power. And that tension, between the decentralized dream and the centralized reality, is not over. It is just beginning.This is Episode Notes from Coordinated with Fredrik. If this kind of thing interests you, subscribe and stick around. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

There is a number that keeps rattling around in my head since recording this episode: $8.8 trillion. That is the demand-side value of open source software according to a recent Harvard Business School study. Not the market cap of the companies selling software. The value of the code itself. The actual lines sitting in public repositories, running the global economy, maintained by volunteers, hobbyists, and a rotating cast of corporate contributors who might get reassigned next quarter.To put that in perspective, it rivals the GDP of the entire Eurozone.And here is what keeps me up at night: if I buy a physical component for our energy infrastructure, I get a warranty, a supplier, and a paper trail. If it breaks, I know who to call. But the software stack underneath all of it — the grid management, cloud infrastructure, data pipelines — that sits on a foundation that legally comes with zero warranty. None. Express or implied.That tension is what this episode is really about.When software was just the manualWe tend to think of software as the product. But go back to the 1950s and 60s, and software was an accessory that shipped with the hardware. Nobody hoarded it because there was no reason to. IBM customers formed a user group in 1955 called SHARE, and their motto was disarmingly simple: “SHARE is not an acronym, it’s what we do.” By 1959, they had collaboratively written an entire operating system. Just engineers helping engineers get million-dollar machines to work better.That culture crystallized into something close to a philosophy at the MIT AI Lab in the 1970s. Code was left open. If you needed a program for your experiment, you walked to a cabinet, copied the source from a paper tape, added what you needed, and put it back. It was communal by default.Then the world changed.A printer jam that reshaped the global economyThe proprietary turn started with a legal shift, not a technical one. IBM unbundled software from hardware in 1969 under antitrust pressure, and overnight, code got a price tag. Bill Gates fired the first cultural shot in 1976 with his open letter to hobbyists, essentially arguing that if software has value, creators deserve to get paid. Fair point from a business perspective. But to the hacker community, it felt like someone was putting fences around a public park.The real breaking point, though, was absurdly petty. Richard Stallman wanted to fix a paper jam on a Xerox printer at MIT. He had done it before on their old printer by writing a script that notified users when their print job got stuck. But the new Xerox printer ran proprietary code. When he asked a researcher at Carnegie Mellon for the source, the guy said no — he had signed an NDA. Stallman viewed this as a moral betrayal. That printer jam radicalized him.He launched the GNU project in 1983 and created the GPL, a legal hack that used copyright law against itself. Copyright restricts sharing. Copyleft mandates it. The GPL said: do whatever you want with this code, but if you distribute changes, you must keep them open under the same license. It was viral by design.The hobbyist who built the engineStallman had the philosophy, the legal tools, and the foundational programs. But by the early 90s, GNU was missing its kernel — the part that actually talks to the hardware. Their kernel project, Hurd, was stuck in architectural perfection debates. While they were building a cathedral, a 21-year-old Finnish student named Linus Torvalds posted a casual message on Usenet in August 1991: “I’m doing a free operating system, just a hobby, won’t be big and professional like GNU.”Probably the greatest understatement in the history of technology.Linus licensed his kernel under Stallman’s GPL, not for ideological reasons, but because it was a fair trade: I show you my code, you show me yours. His monolithic kernel was messy, technically “wrong” by academic standards, but it was fast and it worked. History proved that worse is better. A massive, chaotic network of volunteers connected by the internet could iterate faster than any closed corporate team.When the boardrooms noticedCorporate America eventually could not ignore it. But “free software” sounded anti-capitalist and legally terrifying to a CIO in 1997. So in 1998, at a strategy session in Palo Alto, Christine Peterson suggested the term “open source” — stripping away the moral philosophy and replacing it with a pure engineering and business argument. Less vendor lock-in. Shared maintenance costs. Better, faster, cheaper.Microsoft was terrified. Internally, their own engineers admitted Linux was competitive. Publicly, Steve Ballmer called it “a cancer.” Their strategy was embrace, extend, extinguish. But then IBM showed up as the white knight. In 2001, they announced a billion-dollar investment in Linux. Not because they cared about the OS — they made their money on hardware and consulting. It was a move to commoditize the operating system layer and destroy Sun Microsystems’ expensive proprietary Unix business. IBM told every Fortune 500 CIO: Linux is safe. You will not get fired for running it.And once it was deemed safe, it started eating the world. The LAMP stack (Linux, Apache, MySQL, PHP/Python) let startups build for zero licensing cost. Facebook, Wikipedia, WordPress — all built on free infrastructure. By 2002, Apache ran 58% of all websites.The irony peak came in 2018 when Microsoft, the “cancer” company, acquired GitHub for $7.5 billion. They finally realized the cancer was actually the cure for their own irrelevance.The fragility underneathHere is where it gets uncomfortable. XKCD comic 2347 shows all of modern digital infrastructure as a massive tower balanced on one tiny block: “a project some random person in Nebraska has been thanklessly maintaining since 2003.” It is funny until you realize it is basically a documentary.Heartbleed in 2014 showed us that the encryption library securing most of the internet’s traffic was maintained by essentially one person full-time. Log4j in 2021 scored 10 out of 10 on the severity scale and compromised 40% of business networks overnight — a logging library so boring nobody paid attention to it.But the one that should genuinely scare any CEO is the XZ Utils backdoor from 2024. This was not an accidental bug. A persona calling themselves “Jia Tan,” likely a state-sponsored actor, spent two to three years earning the trust of a burned-out volunteer maintainer. They submitted helpful patches, took load off the tired guy’s shoulders, and were gradually granted repository permissions. Once they had the keys, they injected a backdoor designed to subvert SSH authentication — the way administrators securely log into servers. If it had hit stable Linux releases, attackers would have had a master key to millions of servers worldwide.It was caught by pure luck. A Microsoft engineer named Andres Freund noticed SSH logins lagging by half a second during routine benchmarking, got curious, decompiled the binaries, and found the most sophisticated supply chain attack in history. Half a second of latency saved us.Maintainer burnout is not just a sad open source HR problem. It is a national security vulnerability.AI breaks the definitionAnd now we have AI, where the very meaning of “source code” falls apart. A traditional open source project is human-readable text files. An LLM is three things: architecture, training data, and weights. When Meta releases Llama and calls it “open source,” they give you the weights but not the training data. It is like handing someone a compiled binary without the original code. You can run it, but you cannot reproduce it or deeply audit it.The geopolitics here are loud. Meta releasing Llama was not charity. It was scorched earth. If capable AI models become a free commodity, OpenAI and Google lose their moat. Meta protects its core ad business by destroying competitors’ margins. It is IBM and Linux all over again, just at a different scale.And now nation states are playing. The UAE funds the Falcon models. France backs Mistral. DeepSeek R1 from China matched GPT-4 reasoning at a fraction of the training cost and briefly crashed NVIDIA’s stock. China is using open source as a competitive wedge against American AI dominance.Satya Nadella made a point that stuck with me: data sovereignty is not just about where your servers sit. It is about tacit knowledge. If you rely entirely on a closed API from an American tech giant, every prompt you send makes their model smarter. You are exporting your own intelligence. But if you download an open-weight model and fine-tune it locally, you keep the weights. You own the brain.For any company running critical infrastructure, this is not a theoretical debate.Grazer or gardener?The episode ends with a question I have been sitting with: when I look at our tech stack, do I see a pile of free resources to consume indefinitely? Or do I see a fragile supply chain that requires active stewardship?Because if you are not contributing back — engineering time, financial support, participating in governance — you are not a neutral user. You are part of the risk profile. You are building skyscrapers balanced on the shoulders of that tired volunteer in Nebraska.Open source is not just code anymore. It is the invisible critical infrastructure of the modern world. And infrastructure requires maintenance.Listen to the full episode of Coordinated with Fredrik wherever you get your podcasts. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

There’s a strange paradox at the heart of Europe’s energy transition.On paper, 2025 looks like a victory lap. Renewables generated nearly half of the EU’s electricity. Wind and solar alone overtook fossil fuels. A decade ago, that would have sounded like science fiction.We have become incredibly good at harvesting energy from the sky and the wind.And yet, at the same time, we are wasting staggering amounts of it.This is not a generation crisis.It’s a plumbing crisis.The Ferrari Engine in a Model T ChassisHere’s the uncomfortable truth:We built a 21st-century renewable generation fleet and plugged it into a 20th-century grid.Modern renewables are:* Variable – the sun and wind don’t follow office hours* Distributed – rooftops, fields, offshore wind farms* Digital – powered by inverters and electronicsBut the grid they connect to was designed for something entirely different:* Centralized, giant power plants* One-way power flow* Passive consumers* Heavy mechanical inertiaIt was built for coal plants pushing electrons in one direction.Now power flows in every direction, from millions of rooftops, batteries, EVs, and wind farms. The infrastructure was never designed for that.And the consequences are becoming impossible to ignore.Curtailment: Paying for Power We Throw AwayWhen the grid can’t move electricity from where it’s produced to where it’s needed, operators do something painful.They curtail.That means telling wind farms or solar plants to stop producing electricity—even when the wind is blowing and the sun is shining.And because of contractual agreements, we often pay them anyway.In 2025:* Germany, France, and the Netherlands curtailed 3.9 TWh of renewables.* Great Britain alone wasted 10 TWh—roughly enough to power every home in London for a year.* Congestion costs across seven key European countries hit €7.2 billion in a single year.Let that sink in.We are paying billions to shut off clean power and then paying again to turn on fossil fuel plants elsewhere because the wires can’t carry the energy to where demand is.This isn’t an environmental failure.It’s a coordination failure.Negative Prices Are Not a Good ThingYou may have seen headlines about negative electricity prices.It sounds like abundance. It sounds like victory.It’s not.In 2025:* Germany experienced 539 hours of negative prices.* The Netherlands saw 584 hours.Negative prices happen when there’s too much electricity in the wrong place at the wrong time and not enough flexibility to absorb it.The market is essentially screaming:“Someone please use this power.”“Someone please stop producing.”But we don’t have enough storage.We don’t have enough transmission.And we don’t have enough flexible demand.So the price collapses.Negative prices are not a sign of success. They are a stress signal.The Iberian Blackout: When Physics Fought BackIn July 2025, Spain and Portugal experienced a massive blackout affecting 60 million people.It started with what should have been a manageable grid fault.But that day, the system had very low inertia.Old grids relied on massive spinning turbines in coal and nuclear plants. Those rotating machines act like giant flywheels. They resist sudden changes and stabilize frequency.Solar panels don’t spin.They connect through inverters. They have zero physical inertia.When the disturbance hit, frequency collapsed almost instantly. Protective systems cascaded. Spain and Portugal were electrically islanded to protect the rest of Europe.This was a wake-up call.A grid dominated by electronics behaves differently than a grid dominated by heavy mechanical systems.If we don’t design for that difference, we pay for it.How We Got Here: The Postwar BlueprintTo understand the problem, we have to go back to 1949.European engineers toured the United States under the Marshall Plan and returned with a clear philosophy:* Big power plants* Centralized generation* One-way transmission* Predict and provideIt worked. It rebuilt Europe. It powered decades of growth.But it baked in a core assumption: energy flows from the center outward.Today, that assumption is broken.And nearly 40% of Europe’s distribution grids are over 40 years old.They were not built for:* EV charging* Rooftop solar* Bidirectional power flow* Real-time flexibilityWe added millions of high-performance renewable assets.We did not upgrade the roads.The Queue: 1,700 GW Waiting in LineAcross 16 European countries, 1,700 GW of renewable capacity is stuck in grid connection queues.That’s three times what Europe needs to meet its 2030 climate targets.Why?Because:* Grid studies are slow and bureaucratic* Transmission build-out takes 8–15 years* Developers submit speculative “ghost” projects to secure queue spots* Permitting and public opposition delay everythingSolar farms can be built in 1–3 years.Transmission lines take a decade.We’re trying to run a sprint while tied to a marathon walker.The Perverse Incentive: CapEx BiasHere’s the structural problem few people talk about.Grid utilities earn regulated returns on capital expenditure (CapEx). That means:* Build a €1 million transmission line → earn guaranteed returns for decades* Install €100,000 worth of congestion-solving software → no profit marginUnder most regulatory systems, utilities are incentivized to build concrete, not code.Even if software solves the problem faster and cheaper.This is not a technical limitation.It’s a regulatory one.If we don’t fix the incentive structure, we will keep choosing the slow, expensive option.The Software-Defined GridThe future grid needs four layers:1. SensingYou cannot manage what you cannot measure.We need real-time visibility at the edge of the network.2. Control & AutomationGrid events unfold in milliseconds.Human reaction times are irrelevant.Automation must stabilize frequency and manage flows instantly.3. MarketsFlexibility must be valued.If your EV or home battery provides balancing services, you should be paid.Price signals need to reach the edge.4. Physical BuildYes, we still need more wires.Especially high-capacity transmission corridors.But they must be built strategically.This is not hardware versus software.It’s hardware + software + coordination.Dynamic Line Rating: Free Capacity We’re IgnoringToday, most transmission lines operate using static ratings based on worst-case scenarios (e.g., hot, windless summer days).But in reality, lines are often cooler and wind-cooled.Dynamic Line Rating (DLR) uses real-time weather and sensor data to increase line capacity safely.The result?30–40% more capacity from the same wire.Without building anything new.That’s not incremental.That’s transformative.Virtual Power Plants: Coordination at the EdgeInstead of building new gas plants to meet peaks, we can aggregate:* EV batteries* Home storage* Smart water heaters* Flexible industrial loadsThousands of small devices can act like one large power plant.A virtual power plant (VPP) can:* Discharge during peak demand* Absorb surplus during negative price events* Provide balancing servicesAt roughly one-tenth the cost of building new physical infrastructure.This is coordination as infrastructure.The Global Race: China Is Moving FasterIn 2024 alone, China invested $83 billion in its grid.It has built 37 of the world’s 39 ultra-high voltage (UHV) transmission lines.Europe has zero.UHV lines move enormous amounts of power across thousands of kilometers with minimal losses.This is not just about climate.It’s about industrial competitiveness.If cheap renewable energy can’t reach European industry, manufacturing migrates to where it can.The grid is now geopolitics.The €584 Billion QuestionEurope needs approximately €584 billion in grid investment by 2030.That sounds enormous.But every euro invested in grid infrastructure saves roughly two euros in system costs by:* Reducing curtailment* Avoiding fossil backup* Lowering congestion costs* Increasing efficiencyThe cost of inaction is far higher.The Real Political QuestionWe talk about NIMBY—Not In My Backyard.But the coming battle is deeper:* Not under my street (new cables)* Not in my view (new pylons)* Not near my town (new substations)We have to decide:Are we willing to build visible infrastructure to enable an invisible energy revolution?Or will we let a clean energy abundance die in a bureaucratic waiting room?From Abundance to CoordinationWe no longer have an energy scarcity problem.We have a coordination problem.The next decade of the energy transition will not be defined by how many solar panels we install.It will be defined by:* How intelligently we move electricity* How well we coordinate distributed assets* How fast we reform regulation* How effectively we digitize the gridThe future of clean energy is not just generation.It’s plumbing.And the question is simple:Will we upgrade it in time? This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

On February 18th, 1745, in Como, Italy, a child was born who would quietly alter the trajectory of civilization.Two centuries later, the unit of electric potential — the volt — would carry his name.But the real story doesn’t begin with a nobleman in northern Italy.It begins with a dead frog.In the late 1700s, Luigi Galvani observed something uncanny: a dissected frog’s leg twitching violently when touched with two different metals. He believed he had discovered animal electricity — a vital force inside living tissue.Alessandro Volta disagreed.Volta argued the frog wasn’t the source of the electricity. It was merely a conductor — a wet, salty bridge between two metals. To prove it, in 1800 he stacked alternating discs of zinc and copper separated by brine-soaked cardboard. When he connected the top and bottom, current flowed — continuously.Not a spark. Not static.A steady stream.The world’s first battery.That stack of metal and wet paper — the voltaic pile — was the ancestor of every lithium-ion pack on Earth today.And I believe we are living through a moment just as significant as that day in 1800.The Constraint That Shaped 300 YearsFor three centuries, our civilization has had one fundamental limitation:We could produce energy.But we could not store it.Electricity had to be used the exact second it was generated. The grid had to balance supply and demand every millisecond. Turn on a light, and somewhere a gas turbine spins slightly faster.We were tethered to fuel.Batteries cut the tether.Energy storage is not a convenience feature. It is a structural transformation.For the first time in history, we can decouple energy production from energy consumption at scale.And the cost curves are collapsing.In 2010, lithium-ion battery packs cost around $1,100 per kilowatt-hour.In early 2026, they sit around $108 per kilowatt-hour.A 93% drop.If your rent or groceries had dropped 93% in fifteen years, you would live in a different world.This is not incremental improvement.This is the precondition for energy abundance.Why Batteries Were Stuck for a CenturyFrom roughly 1860 to 1990, battery energy density barely doubled. Over a century of stagnation.Why?Because batteries don’t follow Moore’s Law.A computer chip moves electrons — nearly massless particles. You can miniaturize pathways for electrons endlessly.A battery moves ions.Lithium ions. Sodium ions. Lead ions.Ions are physical objects with mass and volume. You cannot shrink atoms. If you want more energy, you need more material.You are constrained by thermodynamics — by the chemical bonds themselves.That’s why gasoline dominated the 20th century.One kilogram of gasoline holds around 12,000 watt-hours of energy.A kilogram of early lead-acid batteries held maybe 30.Physics was ruthless.Until lithium-ion.The Three-Person Relay That Changed EverythingLithium-ion wasn’t one breakthrough. It was a relay race across decades.* M. Stanley Whittingham discovered intercalation — lithium ions sliding between layers of a crystal without destroying it.* John Goodenough doubled the voltage by introducing lithium cobalt oxide.* Akira Yoshino removed volatile lithium metal and replaced it with carbon, making the system safe.And in 1991, Sony commercialized it.Not in a car.In a camcorder.The early lithium-ion battery cost about $7,500 per kilowatt-hour. Astronomical. But consumers weren’t buying energy — they were buying memories. Longer filming time. Portability. Experience.Consumer electronics funded the factories.Your old camcorder paid for the battery in your EV.The Learning Curve MiracleThe collapse in battery prices is not random. It follows Wright’s Law: every time cumulative production doubles, costs fall by a predictable percentage.For lithium-ion, that learning rate sits roughly between 18% and 28%.Gigafactories didn’t just scale production — they accelerated learning. Thousands of micro-improvements: thinner foils, faster rollers, optimized coatings, better yields.China’s massive overcapacity — often criticized — has actually intensified price competition and accelerated cost decline.This is how revolutions compound.The Chemistries of AbundanceLithium-ion is no longer one chemistry.The future is diversification toward abundance.LFP — Lithium Iron Phosphate* No cobalt.* No nickel.* Iron and phosphate — cheap, globally abundant.* Longer lifespan (thousands of cycles).* Much safer thermal characteristics.Modern cell-to-pack designs have compensated for lower cell energy density through structural innovation.LFP is no longer the “cheap and weak” option. It’s becoming the workhorse of global electrification.Sodium-IonSodium is everywhere. It’s in salt. In oceans. In the Earth’s crust.It’s not energy-dense enough for high-performance cars yet, but for:* City vehicles* Scooters* Grid storage* Cold climatesIt’s extraordinary.It can be shipped at zero volts. It performs well at extreme cold. It avoids many supply-chain bottlenecks.This is true materials abundance.Iron-AirFor grid-scale, long-duration storage, iron-air may be transformative.It literally rusts and unrusts iron to store energy.Heavy. Slow. Perfect for the grid.Target cost: around $20 per kilowatt-hour.At those prices, storage becomes almost infrastructural — like concrete.The Grid Is Already ChangingThis is not theoretical.In California, batteries have become the largest contributor during evening peak demand. They are flattening the infamous duck curve by storing midday solar and discharging after sunset.China installed staggering amounts of storage capacity in 2025.Saudi Arabia and Abu Dhabi are building solar-plus-storage systems designed for 24/7 dispatchability.The old critique — “renewables are intermittent” — weakens when storage becomes cheap.At projected costs of $32–$54 per kilowatt-hour by 2030, building new solar-plus-storage may become cheaper than fueling existing gas plants.That’s not moral persuasion.That’s spreadsheet logic.Energy Abundance and the Jevons QuestionThere is a paradox in economics: as efficiency improves, consumption often increases.If energy becomes cheap, we will use more of it.That is not necessarily a problem.Cheap energy enables:* Desalination at scale.* Industrial recycling.* Indoor agriculture.* AI infrastructure.* Material synthesis.* Climate adaptation technologies.Every time energy became cheaper in history — from wood to coal to oil — human living standards rose.The difference now is that the primary source is stellar.The sun sends 173,000 terawatts to Earth continuously.The constraint was never generation.It was storage.Circularity and ResilienceA “dead” battery is not waste. It is high-grade ore.Battery recycling can recover over 95% of critical materials. Before recycling, batteries can serve second-life roles in stationary storage.We move from extractive mining toward circular supply chains.Add AI into this ecosystem — optimizing dispatch, predicting degradation, orchestrating millions of distributed assets — and the system becomes self-balancing.An Internet of Energy.The Bigger FrameCivilization is a heat engine.For thousands of years, that engine ran by burning things.Now we are transitioning to storing sunlight.The battery is the bucket.And we are learning to make buckets from iron, sodium, carbon — from common materials, not rare ones.When energy becomes abundant:* Water can become abundant.* Food can become abundant.* Computation can become abundant.* Intelligence can become abundant.This isn’t utopian thinking.It’s where the cost curves point.The next time you charge your phone, plug in your car, or see a solar panel on a roof — don’t just see a device.See the early architecture of a civilization untethered from combustion.Volta would be shocked.And we’re just getting started. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

Coordinated with FredrikThere is a fundamental tension at the heart of every ambitious company.On one side: the blue-sky vision.On the other: the gritty, operational reality of making it real.Engineers understand this tension intuitively. Physics doesn’t negotiate. Gravity doesn’t care about your roadmap. Thermodynamics ignores your quarterly goals. If your calculations are off, the bridge collapses.Leadership, however, is fuzzier. There are no immutable equations. No compile errors flashing red before failure. Just drift.This episode is about how to remove that drift.It’s about execution.And it’s rooted in the philosophy behind Measure What Matters by John Doerr—and the engineering rigor of Andy Grove.The Activity TrapAndy Grove, former CEO of Intel, had a deep disdain for what he called the activity trap.The activity trap is motion without progress.Lights on late at night.Slack buzzing.Meetings stacked.Features shipping.But no real value created.Grove’s insight was simple but brutal:Knowledge is potential energy.Execution is kinetic energy.Most organizations measure effort. Few measure output.That distinction is the entire game.Operation Crush: Alignment as SurvivalIn 1980, Intel was threatened by Motorola’s 68000 chip. It was faster. Cleaner. Technically superior.The default engineering response would have been: build a better chip.But chips take years.Grove didn’t change the product.He changed the battlefield.He launched Operation Crush.Objective: Crush Motorola.Key Result: Win 2,000 design wins in one year.That specificity mattered.Not “increase market share.”Not “improve positioning.”But 2,000.Engineering pivoted to support sales.Marketing pivoted to target executives instead of programmers.The entire organism aligned around one measurable outcome.They hit it.The lesson:A technical deficit can be overcome by organizational alignment.Execution is leverage.The Four Superpowers of OKRsAt the center of this operating system are OKRs: Objectives and Key Results.They are not to-do lists.They are not performance metrics.They are directional control systems.Let’s break down the four “superpowers.”1. Focus and CommitHigh-performing teams set three to five objectives max.More than that is dilution.Focus is not about remembering what to do.It is about killing what not to do.The discipline is brutal.If your goal is engagement, you may have to say no to features users ask for—if they reduce engagement. Even good ideas must die if they don’t serve the objective.Focus requires commitment.If leadership wavers, the system collapses.If the CEO’s calendar doesn’t reflect the OKRs, neither will the company’s behavior.Signal must match structure.2. Align and ConnectAs companies scale, silos form.Engineering optimizes for elegance.Sales optimizes for revenue.Marketing optimizes for visibility.Without alignment, each team finds a local maximum while the company misses the global one.OKRs solve this through radical transparency.Everyone’s goals are visible.Dependencies become explicit instead of accidental.The model isn’t purely top-down. Healthy systems are roughly:* 50% top-down direction* 50% bottom-up initiativeThis balance protects innovation while maintaining coherence.It’s how Google built Gmail through 20% time—bottom-up ideas aligned to top-level vision.3. Track for AccountabilityTracking is not about punishment.It is about feedback loops.A healthy OKR culture treats missed goals as data, not moral failure.Google famously scores OKRs on a 0.0–1.0 scale.And here’s the twist:0.6–0.7 is the sweet spot.If you’re consistently scoring 1.0, your goals are too safe.A 0.7 means you stretched.A red is not shame.A red is signal.Some organizations institutionalize this through rituals like “selling your reds” — publicly sharing failures to invite support and problem-solving.That level of psychological safety is not optional.It is infrastructure.4. Stretch for AmazingThere are two kinds of objectives:Committed objectivesOperational promises. Must hit 1.0.(Example: payroll accuracy, uptime guarantees.)Aspirational objectivesMoonshots. You don’t know how to achieve them when you set them.This was the philosophy behind Google Chrome under Sundar Pichai.Early goals were missed.They doubled down.Then doubled again.Eventually, they won.Stretch goals force non-linear thinking. If the target is incremental, solutions remain incremental.If the target is absurd, the architecture must change.This is how YouTube moved from counting clicks to measuring watch time—and set a billion-hours-per-day target.The number wasn’t the point.The forced re-architecture was.The Human Operating System: CFRsOKRs without human infrastructure fail.This is where CFRs come in:* Conversations* Feedback* RecognitionAnnual reviews are too slow.Markets move weekly.Continuous check-ins replace yearly judgment.Leadership isn’t just about metrics. It’s about understanding:* What energizes your people?* What drains them?* Where do they want to go?Performance without humanity becomes brittle.Humanity without performance becomes chaos.The system requires both.Structure Does Not Kill SoulEven activists discovered this.Bono once feared that introducing structure into advocacy would dilute passion.Instead, it amplified it.Emotion without targets is noise.Passion with metrics becomes force.Structure is not bureaucracy.It is a vessel.Engineering the OrganizationHere is the synthesis.If you are building complex systems—power grids, distributed energy resources, intelligent controllers—you would never operate without telemetry.You measure load.You measure voltage.You monitor frequency.You debug the system continuously.Your organization deserves the same rigor.OKRs are telemetry for ambition.They translate vision into velocity.They convert hallucination into measurable progress.And they expose drift before collapse.The Final QuestionLook at your current dashboard.Is everything green?If so, you are probably managing the probable.Greatness lives in the red.The question isn’t whether you’re missing goals.The question is whether your goals are bold enough to miss.Execution is everything.The operating system is optional.The outcomes are not.—Coordinated with Fredrik This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

There is a phrase that has quietly governed modern management culture for decades:“Don’t boil the ocean.”It’s the sentence that appears whenever ambition starts to feel uncomfortable.When scope expands.When a system-level question threatens a quarterly roadmap.It’s framed as wisdom. Prudence. Maturity.But what if that advice—so deeply internalized that we barely question it anymore—has quietly become dangerous?This episode, and this essay, explores a contrarian but increasingly unavoidable thesis:In an era of collapsing intelligence costs, not boiling the ocean is how you lose.This is not a motivational slogan. It’s an economic and engineering argument.To understand why, we need to rewind nearly 160 years—back to coal mines, steam engines, and a mistake humanity has repeated every time a general-purpose resource becomes radically cheaper.The Original Mistake: When Efficiency BackfiresIn 1865, at the height of the British Industrial Revolution, a young economist named William Stanley Jevons published a book called The Coal Question.At the time, Britain was anxious about energy dominance. Coal powered everything: factories, railways, ships, empire. The assumption among policymakers was simple and intuitive:As engines become more efficient, total coal consumption will fall.After all, James Watt’s steam engine was dramatically better than the old Newcomen design—using roughly one quarter of the fuel for the same mechanical work.Efficiency should lead to conservation.Except it didn’t.Jevons observed something deeply counterintuitive:Despite massive efficiency gains, coal consumption didn’t fall at all.It exploded.UK coal production grew steadily for decades—rising from ~5 million tons in 1750 to over 100 million tons by the 1860s, eventually peaking near 300 million tons in the early 20th century.Jevons summarized the paradox succinctly:“It is wholly a confusion of ideas to suppose that the economical use of fuel is equivalent to a diminished consumption. The very contrary is the truth.”This is what we now call the Jevons Paradox:When a general-purpose resource becomes more efficient and cheaper, total consumption increases—because new uses become economically viable.Efficiency doesn’t cap demand.It unlocks it.Latent Demand and the Threshold of ViabilityWhy does this happen?Because demand for fundamental resources isn’t fixed—it’s latent.When steam power was expensive and inefficient, it was used only for extreme, high-value tasks (like pumping water out of deep coal mines). Once efficiency improved, the activation energy dropped.Suddenly it made sense to:* Put engines in textile mills* Power ships and locomotives* Mechanize entire industriesThe question was never “How much steam power do humans need?”The question was “At what price does entirely new behavior emerge?”That same dynamic has repeated itself over and over again.Light: From Luxury to PollutionNothing illustrates this better than the history of light.In the 1300s, producing a fixed amount of illumination—about one million lumen-hours—cost the modern equivalent of £40,000. Light was so expensive that people rationed candles the way we ration fuel during wartime.By 2006, that same amount of light cost £2.90.A 14,000× reduction in real cost.So did we save energy?No.Between 1800 and 2000, per-capita light consumption increased ~6,500×.We didn’t stop at lighting rooms. We lit cities. Highways. Stadiums. Parking lots at 3 a.m. We put light into pockets, shoes, keyboards, architecture. We created an entirely new problem—light pollution—because light became too cheap to care about.LEDs repeated the pattern again:* Lower wattage per bulb* Explosive growth in total lightingEfficiency didn’t restrain usage.It expanded imagination.Doing More With Less: Buckminster Fuller’s Missing HalfThis is where Buckminster Fuller enters the story.Fuller described a long-term technological trajectory he called ephemeralization:Doing more and more with less and less—until eventually you can do everything with almost nothing.His favorite example was bridges.* Roman bridges: massive stone, brute force, pure compression* Iron bridges: lattice structures, geometry, less material* Steel suspension bridges: tension, elegance, minimal mass* Eventually: radio waves, fiber optics—connection without materialThe function remains.The atoms disappear.At first glance, Fuller seems to contradict Jevons. But he doesn’t.They describe the same system from different angles:* Ephemeralization → less material per unit of function* Jevons Paradox → vastly more total units once the function becomes cheapWe didn’t save copper by inventing fiber optics.We used orders of magnitude more communication.The Great Bet: Ingenuity vs. ScarcityThis tension came to a head in the 1970s.On one side:* The Club of Rome* Paul Ehrlich* Limits to Growth, The Population Bomb* A zero-sum worldview: finite resources, inevitable collapseOn the other:* Julian Simon* Buckminster Fuller* The belief that human ingenuity is the ultimate resourceIn 1980, Simon challenged Ehrlich to a bet.Ehrlich chose five industrial metals—copper, chromium, nickel, tin, tungsten—and predicted prices would rise over the next decade as population exploded.Instead, prices fell by 57%.Why?* Substitution (fiber replaces copper)* Better extraction* Recycling* Design efficiencyIngenuity outran depletion.Intelligence Enters the EquationAll of this matters because we are now repeating the same mistake—but with something far more powerful than coal or light or copper.We are making intelligence cheap.The cost of AI inference has been collapsing at an unprecedented rate—on the order of hundreds of times per year.What cost ~$20 per million tokens in 2022 costs cents today.This is ephemeralization of cognition.And if Jevons holds—as it always has—then the implication is unavoidable:Cheap intelligence will not reduce work.It will explode the scope of what gets built.The fear narrative—“AI will take jobs”—is the same zero-sum thinking that lost the Simon-Ehrlich bet.It assumes:* A fixed amount of code* A fixed amount of analysis* A fixed amount of problem-solvingHistory says the opposite.When the cost of thinking drops, we attempt problems that were previously unthinkable.The Real Bottleneck Has MovedWhen software was expensive, the bottleneck was execution.Now execution is cheap.The bottleneck is vision.This is why incrementalism is suddenly dangerous.Optimizing for 1.05×—cutting support costs, shaving headcount, marginal automation—is defensive thinking applied to an abundance problem.As Astro Teller famously put it:“It’s often easier to make something 10× better than 10% better.”Why?Because 10% forces you to argue with legacy constraints.10× forces you to throw them out.Energy, AI, and the Literal OceanIn energy, this becomes especially clear.If AI helps unlock controlled fusion—abundant, clean, baseload power—the question isn’t “How much cheaper is my electricity bill?”The question is:* What becomes possible when energy is no longer the constraint?Desalination at planetary scale.Carbon capture as infrastructure.Terraforming, not conservation theater.This is Jevons again—at civilizational scale.The Actual ChoiceBuckminster Fuller framed it starkly:Utopia or oblivion.Not because technology guarantees utopia—but because fear guarantees stagnation.The tools are arriving whether we are psychologically ready or not.The only remaining decision is whether leaders choose:* Scarcity thinking and protectionism* Or positive-sum ambition and constructionSo the real strategic question becomes:Where are you still optimizing for 1.05× when the physics now allow 10×?What ocean are you refusing to boil—not because it’s impossible, but because it used to be?Because the water is ready.The apparatus exists.And timid incrementalism is no longer neutral—it’s a risk. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

Coordinated with Fredrik — Episode RecapOn January 30, 2026, SpaceX filed what looked like the most boring piece of regulatory paperwork imaginable. An FCC application. A string of numbers. The kind of thing you scroll past.Except this one was for permission to launch one million orbiting data centers. And in the preamble, they called it “a first step towards becoming a Kardashev Type II civilization.”That is not normal language for a permit application. That is a declaration of intent for a different species.This episode digs into what that filing actually means, and why it forces anyone working in energy to confront a question that used to be reserved for philosophy seminars: are we wiring the planet, or wiring the solar system?Three visions, one fork in the roadThe episode walks through three competing models for where energy goes from here. They sound like they belong in different centuries, but all three are showing up on balance sheets right now.The Earthbound Optimizer. This is Professor Mark Jacobson’s model out of Stanford. His thesis is that we can run 100% of civilization on wind, water, and solar. Not just electricity. Everything. Transport, heating, industry, agriculture, the military. All of it, with existing technology. No fusion. No miracle batteries. No carbon capture.The physics behind it is surprisingly straightforward. Combustion is terrible at converting energy into useful work. A gasoline car turns only 17-20% of its fuel into motion. The rest is heat and noise. An electric motor runs at 90-95% efficiency. A heat pump moves three to four units of heat for every one unit of electricity you put in. Jacobson calculates that simply by electrifying everything, we cut global energy demand by 56.4%. The upfront cost is around $61.5 trillion, but annual energy costs drop from $17.8 trillion to $6.6 trillion. That is a six-year payback with an infinite tail of savings. Any board would fund that project in a heartbeat.So why haven’t we done it? Because the model assumes 80% of daily electrical loads can be shifted within an eight-hour window. Charging your car at 2 AM instead of 6 PM? Easy. Asking a steel mill or a data center training an AI model to pause for eight hours? That is where the model meets reality.The Orbital Industrialist. This is the Musk play. He looked at the seven-year wait for a new substation in Virginia, the zoning fights, the interconnection queues, and decided that the grid is a political problem. Rockets are a physics problem. He prefers physics problems.In a sun-synchronous orbit, solar panels get 99% uptime. No clouds, no night, no atmosphere scattering the light. A panel in space generates six to eight times more energy per year than the same panel on Earth. The whole idea is to move the heavy compute, the training runs that take months and consume staggering amounts of power, off the planet entirely. Train the model in orbit where energy is constant and free. Beam the finished weights back to Earth. Learning happens in the sky. Thinking happens on your phone.The catch is cooling. Space is a perfect insulator. There is no air for convection. The only way to dump heat is radiation, and to radiate at the scale of gigawatt data centers you would need radiator panels the size of Gibraltar. Silicon chips melt long before the radiator reaches efficient operating temperature. You might need entirely new semiconductor materials, gallium arsenide or silicon carbide, that can run at 300-400 degrees Celsius. It is not just about launching servers. It might mean reinventing the chip.The whole bet rides on Starship driving launch costs from $2,700 per kilogram down to $200, maybe eventually $10. At $10 per kilo, you can launch heavy, cheap, standard server racks. Mass stops being a constraint. The engineering tradeoffs change completely.The Cosmic Architect. This is the Dyson swarm endgame. Not a solid shell around the sun (that is physically impossible), but trillions of individual satellites orbiting in dense formation, each capturing a sliver of sunlight. Musk’s million satellites would capture roughly 0.00000000004% of the sun’s output. A rounding error on a rounding error. But the expansionist logic says once you start, you do not stop.The theoretical blueprint is called the Mercury Loop. You land self-replicating mining robots on Mercury, which is rich in metals and sits right next to the sun. They mine the surface, build thin-foil solar collectors, and use electromagnetic railguns to shoot them into orbit. Those collectors beam energy back down to power more mining. It is an exponential feedback loop. Researchers at Oxford calculated you could dismantle the entire planet in about 31 years.Even at that scale, thermodynamics wins. The Landauer limit means every bit erased generates heat. A Dyson swarm eventually cooks itself if it thinks too hard.The Jevons Paradox sitting in the middle of all thisThis is the tension that runs through the entire episode and connects directly to how any energy company should think about the next decade.Jacobson argues that efficiency leads to sufficiency. Electrify everything, coordinate the loads, and demand goes down. We can get by with less.The expansionist view says the opposite. William Stanley Jevons noticed in the 19th century that when steam engines got more efficient, coal consumption went up, not down. Cheaper energy means more uses for it. If you unlock cheap orbital compute, demand does not flatten. It explodes into virtual worlds, planetary simulations, uses we cannot even conceive of yet.If Jacobson is right, energy companies are optimization businesses. You squeeze value out of a more or less static system. If Musk is right, you are preparing for a grid that needs to double, then triple, then quadruple. It is not a conservation problem. It is a throughput problem.The one thing all three visions agree onWhether the power comes from a rooftop in Palo Alto, a satellite 500 kilometers up, or a ring of collectors around the sun, the bottleneck is always the same: coordination.Jacobson’s model only works if 80% of load is flexible. That requires massive demand response, virtual power plants, automated dispatch. Space solar needs laser downlinks, ground stations, collision avoidance for a million moving objects, all managed in real time. Even the Dyson swarm needs orchestration at a scale that makes today’s grid look like a toy.The hardware is not the hard part. The connective tissue is. California proved this in 2024. They hit 117% renewable coverage in some intervals. Battery storage grew 2,100% in five years. But they also threw away 3.4 million megawatt hours of clean energy because they could not move it in space or time. Germany spent three billion euros just on redispatch, paying plants to turn down in one place and up in another to manage congestion.The electrons are there. The infrastructure to get them to the right place at the right time is what is lagging.The ownership question nobody wants to talk aboutJacobson’s world is distributed. Rooftop solar, community wind, local batteries. Hard to monopolize sunshine when it falls on everyone’s roof.The orbital and Dyson worlds are centralized by nature. You need to be a trillion-dollar entity to launch rockets at scale. You need to own the mass drivers on Mercury. It recreates the dynamics of the oil industry. A few players control supply, everyone else is a customer.We are choosing between energy democracy and energy tycoons. Or some hybrid of the two.So what does this actually mean?We receive 10,000 times more energy from the sun than we currently use. The scarcity is not natural. It is a scarcity of infrastructure and coordination.Musk’s orbital play is, at its core, a hedge against our own dysfunction. A bet that we are too slow at building transmission lines, too tangled in zoning fights, too bad at aggregating distributed resources to keep up with what AI demands. So he is routing around the zoning board entirely.Maybe he is right about that. But today, right now, the fight is still on Earth. It is the last 10% problem. It is making the load follow the sun. It is the boring, unglamorous work of connecting millions of devices into something that behaves like a single coordinated system.Whether the future is on Earth or in orbit, the operating system for the energy transition is the same: coordination software, protocols, aggregation. The unsexy layer that makes any of this actually work.We are currently deciding, in boardrooms and regulatory filings, whether to wire the planet or wire the solar system. And every battery you aggregate, every flex load you optimize, is a vote in that election.Keep coordinating.Listen to the full episode on [Coordinated with Fredrik]. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

In October 1990, a Stanford biologist sat at his desk and wrote a check for $576.07. He put it in an envelope. He addressed it to an economist at the University of Maryland. He did not include a note. No congratulations, no concession speech, not even a “good game.” Just the check, sealed and mailed.That silence is deafening when you know the backstory. Because that check was never about the money. It was the settlement of a decade-long wager about the fundamental nature of reality itself. And the argument it represents, between people who look at a finite planet and see walls closing in and people who look at the same planet and see a launchpad, has been raging for over two centuries. Right now, in the age of AI and climate tipping points, it is reaching a fever pitch.This is the story behind our latest episode of Coordinated with Fredrik. We called it “The Spiral” because the clash between these two worldviews is not a pendulum swinging between optimism and pessimism. A pendulum returns to the same points. A spiral goes around, but with each revolution, it moves up an axis. It progresses. Each side is forced to incorporate something the previous round missed, and the stakes get higher with every turn.The bet that started with too many people and ended with no noteThe man writing the check was Paul Ehrlich, author of The Population Bomb, a book that sold two million copies and predicted hundreds of millions of people would starve to death in the 1970s and 1980s. Ehrlich had been on The Tonight Show with Johnny Carson roughly 20 times. He got a vasectomy to set an example. He told an interviewer that he would take even money England would not exist by the year 2000. He was, in every sense, the public face of environmental doom.On the other end of the envelope was Julian Simon, an economist who had written The Ultimate Resource, arguing that the human mind is the only resource that matters. Where Ehrlich saw mouths to feed, Simon saw minds that create. In 1980, Simon issued a public challenge: pick any raw materials, any timeframe longer than one year, and I will bet you the inflation-adjusted price goes down.Ehrlich and two colleagues chose five metals. Chromium, copper, nickel, tin, and tungsten. They placed a $1,000 bet with a payoff date of September 29, 1990.During that decade, the world added more than 800 million people, the largest single-decade increase in human history. Demand exploded. And every single metal fell in price. Tin dropped over 70 percent. Tungsten fell by half. Hence the check.For a lot of people, that was the end of the story. Optimists won, pessimists lost, case closed, let’s drill some oil. But the surface reading is dangerously incomplete. If you run the same bet over different decades, the results flip completely. A study found that Ehrlich would have won 61.2 percent of all possible ten-year intervals between 1910 and 2007. From 2000 to 2010, with the China boom driving metal prices parabolic, Ehrlich would have wiped the floor with Simon. It was not a definitive victory. It was a single data point in a much larger war between two operating systems for viewing the world.The Club of Rome and the model that keeps tracking realityThe modern version of the limits worldview began in a villa in Rome in 1968, when an Italian industrialist named Aurelio Peccei gathered scientists and economists because he believed all of humanity’s problems were interconnected. Peccei was not some ivory tower philosopher. He had been tortured by fascists in the anti-resistance movement during the Second World War. He had seen civilization come apart and get rebuilt. He called the interconnected mess of global problems the “problématique,” and his group became the Club of Rome.Four years later, a team of 17 MIT researchers built a computer model called World3 and ran it on room-sized mainframes. They tracked five variables: population, food production, industrial output, pollution, and resource depletion. The key was not just the variables but the feedback loops and delays between them. More factories mean more food, more food means lower mortality, lower mortality means more people, more people means more factories. That is the engine of civilization. But more factories also mean more pollution, which degrades soil, and more resource extraction, which gets progressively harder and more expensive. The system starts to eat itself.The “standard run” scenario, business as usual with no major policy changes, projected overshoot and collapse around the 2040s or 2050s. Not by the year 2000, as critics endlessly claim. If you actually look at the charts from the 1972 book, all the curves keep growing well past 2000. The myth that the Club of Rome predicted the world would end in 2000 is the most persistent straw man in the history of this debate.And the model has tracked reality with unsettling accuracy. In 2008, an Australian physicist named Graham Turner compared 30 years of actual data against the original 1972 curves. The match was terrifyingly close. A 2014 update was bleaker: the data indicated the early stages of collapse could occur within a decade. A 2020 study concluded that without major changes, economic growth will peak and then rapidly decline by around 2040. We are living inside the window of their original prediction right now.The man who saved a billion lives and still said it was temporaryWhile the Club of Rome was modeling collapse, an agronomist from Iowa was busy proving them wrong with his bare hands. Norman Borlaug had gotten to college on a wrestling scholarship. He ended up developing semi-dwarf, high-yield wheat varieties through a technique called shuttle breeding, growing two generations per year by alternating between locations in Mexico.The test came in the mid-1960s, when India and Pakistan teetered on the brink of exactly the catastrophe Ehrlich was predicting on television. Borlaug shipped his seeds. They arrived during the Indo-Pakistani War. The results were staggering. Pakistan’s wheat yields nearly doubled within five years. India went from famine threat to grain surplus so fast that local governments had to close schools and use classrooms as temporary granaries because they ran out of storage.The Congressional Gold Medal credits Borlaug with saving over one billion lives. He is the single greatest data point in the techno-optimist argument. But here is the thing that both sides tend to forget: Borlaug himself was not a blind optimist. In his Nobel Prize acceptance speech, he called his Green Revolution “a temporary success” and “a breathing space.” He warned about population growth in nearly every speech he gave for the rest of his career. He knew he had not solved the problem forever. He had bought us a few decades to get our house in order.A quantum physicist, a basement, and a philosophy built on thermodynamicsThe modern acceleration movement exploded out of a very unlikely origin. In 2022, a French-Canadian quantum physicist named Guillaume Verdon quit his job at Google, moved into his parents’ basement in Quebec, sold his car, bought $100,000 worth of GPUs, and started a movement on Twitter under the pseudonym BasedBeffJezos.The name was a pun on Jeff Bezos. The philosophy was a direct shot at Effective Altruism, the movement associated with AI safety and existential risk. Where EA said slow down, we might destroy ourselves, Verdon’s movement said speed up, or we definitely will. He and his co-founders called it effective accelerationism, or e/acc.The intellectual foundation is built on the work of MIT biophysicist Jeremy England, whose theory of dissipative adaptation proposes that under certain conditions, matter spontaneously organizes itself into more complex structures because those structures are better at spreading energy around. A forest dissipates far more solar energy than a desert. Life, in this framing, is a mechanism the universe evolved to increase entropy faster. Intelligence and technology are even better mechanisms. A data center takes organized energy and converts it into waste heat and information. It is, thermodynamically speaking, a machine for accelerating entropy.E/acc takes this and runs with it. They argue that civilization is a higher-order dissipative structure, that resistance to acceleration is metaphysically misguided, and that humanity’s cosmic duty is to climb the Kardashev scale from a Type 0 civilization to one that harnesses the energy of an entire planet, then a star, then a galaxy. Energy consumption is not a vice. It is a moral virtue.This moved from fringe Twitter into the heart of Silicon Valley strategy at startling speed. In October 2023, Marc Andreessen published his Techno-Optimist Manifesto, a 5,200-word essay that used the phrase “we believe” 113 times and called sustainability, the precautionary principle, and trust and safety the enemies of progress. After the 2024 US election, tech figures began explicitly connecting e/acc principles to deregulatory politics.The geographic split is not a coincidence. Limits thinking is a European movement. The word “décroissance” comes from French. Kate Raworth’s Doughnut Economics was adopted as an official planning framework in Amsterdam. The precautionary principle is baked into EU regulation. E/acc is a Silicon Valley movement, full stop. Its founders are tech workers. Its patron saints are venture capitalists. Its cultural habitat is X. American frontier mythology, libertarian philosophy, and venture capital’s fundamental business model, exponential growth or death, created the conditions for its emergence.Both sides are right, and both sides are dangerously wrongThe hardest part of this story is that neither tribe has the full picture.The Jevons Paradox sits at the center of the conflict like an oracle telling both sides exactly what they want to hear. In 1865, William Stanley Jevons observed that more efficient steam engines did not reduce coal consumption. They increased it, because cheaper energy opened up new uses. We see this playing out with LED lights (90 percent more efficient, yet global light pollution has exploded) and now with AI. From 2010 to 2020, data center energy use stayed flat despite massive computing growth because chips kept getting more efficient. AI shattered that trend. US data center power consumption hit 183 terawatt-hours in 2024 and is projected to reach 426 terawatt-hours by 2030. In Ireland, data centers now consume 22 percent of total national electricity.On this point, the limits crowd is right. You cannot efficiency your way to lower total consumption.But the optimists have their own devastating exhibit. Solar energy followed a cost curve that almost nobody predicted. A watt of solar cost $76.67 in 1977. Today it is around 30 cents. A decline of over 99 percent. By 2018, solar became the cheapest new electricity source in most markets. If energy becomes nearly free and clean, a lot of supposed limits start to dissolve. Water scarcity? We have oceans full of salt water and the main barrier to desalination has always been energy cost. Resource scarcity? With enough cheap energy you can brute-force almost any material back into its constituent elements.And peak oil, once the limits tradition’s most concrete prediction, became a cautionary tale. US production peaked in 1970 exactly as predicted, declined for 38 years, and then fracking and horizontal drilling blew past the old peak. We went from worrying about peak supply to seriously discussing peak demand.But then there is phosphorus. This is the resource nobody talks about at dinner parties that may be the single biggest bottleneck for human civilization. Morocco controls over 70 percent of remaining economically viable phosphate rock reserves. Ninety percent goes to fertilizer. There is no synthetic substitute. You cannot 3D print the element phosphorus. Without it, Borlaug’s entire Green Revolution collapses. Estimates say we have 50 to 100 years at current consumption rates. The e/acc philosophy does not have an answer for this one.And biodiversity loss is not a model prediction. It is a measurement. A 73 percent decline in monitored wildlife populations since 1970. Once a species is gone, its web of ecological relationships is gone forever. That is thermodynamic irreversibility. You cannot unscramble that egg.The spiral keeps turningThe pattern repeats across two centuries, each turn incorporating the truth of the previous one while reacting against it. Malthus identified the problem of geometric growth against linear resources. He was right in principle and wrong about timing. The Industrial Revolution and Borlaug proved technology could break through Malthusian limits, but Borlaug himself warned the fix was temporary. The Club of Rome modeled the unintended consequences of that breakthrough and found the system delays and feedback loops that lead to overshoot. They were right about the structure of the problem but underestimated technology’s ability to postpone the date. Now e/acc is arguing that the solution to the problems of technology is more technology, faster. They are driving the AI boom and the clean energy transition. But they may be blind to the hard geological and biological cliffs we are racing toward.There is a group of thinkers trying to build a bridge. The ecomodernists, led by people like Stewart Brand, the man who created the Whole Earth Catalog and then did a complete 180 on nuclear power, genetic engineering, and urbanization. Their argument is decoupling: use advanced technology to shrink humanity’s physical footprint so that nature can recover. Dense energy, vertical farming, precision agriculture. Use the acceleration toolkit to achieve the goals of the limits movement.That synthesis may be the only path that actually works. You cannot run a modern energy company or a modern economy with only the limits mindset. That leads to stagnation, to a culture of no, to degrowth, which, whatever its intellectual merits, is politically impossible in any functioning democracy. People will not vote to make themselves poorer. But you also cannot lead with pure acceleration. That leads to recklessness, to ignoring externalities, to driving a thousand miles an hour toward a cliff because you have blind faith that innovation will magically build a bridge just before you go over the edge.You need the optimist’s toolkit and the pessimist’s diagnostic. The diagnostic tells you what the hard constraints are: phosphorus cycles, carbon budgets, grid stability, biodiversity thresholds. The toolkit tells you how to solve for those constraints with innovation, new cost curves, new business models.That $576.07 check Ehrlich mailed in 1990 would be worth about $1,400 today. A small price for a public lesson in humility. But the underlying wager has never been settled. The limits model says we are entering the downturn right now. Resources should get scarcer, prices should spike, the system should unravel. The acceleration model says AI and robotics and maybe fusion are about to unleash abundance that makes all previous resources irrelevant.The tension between those two futures has never been tighter. The bet is still on. We are all waiting to see who ends up writing the next check.This post accompanies our latest episode of Coordinated with Fredrik, where we take the deep dive into the full history, the data, and the people behind both movements. Listen wherever you get your podcasts. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

There is a moment in every startup’s life where growth stops feeling like progress.You hired smart people.You raised money.You shipped something that works.And yet—everything feels slower, noisier, more fragile than when you were three people in a room.This episode is a deep dive into the most dangerous phase in a startup’s life: the transition from a scrappy founding team to a 10–15 person company.We unpack:* Why productivity mathematically collapses as teams grow* The psychological traps founders fall into (hero syndrome, identity foreclosure)* Why Slack becomes a liability at scale* What a minimum viable operating system for a 10-person company actually looks like* How founders must shift from doing work to designing systemsThis is not about motivation.It’s about mechanics.If you feel like you’re constantly firefighting, this episode explains why—and how to stop.The Garage Myth Dies at 10 PeopleEvery founder remembers the garage phase.Three or four people.One shared brain.No process, no meetings, no documentation—and somehow everything works.That phase ends brutally around 10 people.Not because anyone is incompetent.But because implicit coordination stops working.There’s a simple formula behind this:N × (N − 1) ÷ 2That’s the number of communication paths in a team.* 3 people → 3 connections* 5 people → 10 connections* 10 people → 45 connections* 15 people → 105 connectionsNothing “feels” different when you hire the 7th or 8th person.But the communication network has already exploded.You’re no longer in a team.You’re running a distributed system—without having designed it as one.Biology and Math Are Both Against YouThis breakdown isn’t just organizational. It’s biological.Anthropologist Robin Dunbar showed that humans have hard cognitive limits on stable group sizes.Two thresholds matter here:* ~5 people: a support clique (everyone knows everything)* ~15 people: a close group limitThe 5–15 range is a no-man’s land.Founders try to manage a small tribe with garage-era instincts.The result is chaos—and the founder becomes the bottleneck.The Bottleneck Founder PatternWhen founders don’t adapt, the same symptoms appear every time:1. Decision QueuesWork stalls while everyone waits for the founder to approve tiny things.The founder becomes a toll booth.2. Team PassivityHigh-performers stop thinking.They wait.They become order-takers instead of owners.3. “Swoop and Poop” ManagementThe founder disappears, then reappears with opinions and changes—without context.Nothing kills morale faster.Crucially:This is not because founders are bad people.It’s because of identity conflict.Identity Foreclosure: Why Letting Go Feels Like DyingMost founders—especially technical ones—built their identity around being the builder.Writing code.Solving hard problems.Getting instant dopamine from things that work.Leadership doesn’t give that feedback.Managing people is:* Delayed gratification* Ambiguous outcomes* Often invisible when done wellAs Paul Graham describes it:founders are trapped between the maker schedule and the manager schedule—and both suffer.So founders compensate by becoming heroes.They jump in.Fix the bug.Save the day.And accidentally teach the team:“Don’t worry. I’ll always fix it.”That’s not leadership.That’s dependency creation.From Firefighter to Fire ChiefThe key shift is this:Stop holding the hose. Start building the fire station.A firefighter fights fires.A fire chief ensures:* Training* Equipment* Water pressure* StrategyTouch the hose only when the building is about to collapse.This transition feels like grief.You’re letting go of the identity that made you successful.But without it, the company never scales.Giving Away Your LegosFormer Facebook leader Molly Graham has a perfect metaphor:Growing a company is like giving away your Legos.You built the thing.You know every brick.Now someone else will build with your pieces—badly, at first.Hovering makes it worse.Her rule:If you’re doing the same job you did six months ago, you’re the bottleneck.Growth requires repeatedly firing yourself.Why Slack Becomes the EnemySlack feels efficient—until it isn’t.Research shows:* 23 minutes to regain focus after an interruption* Even 5-second interruptions triple error ratesAt 10 people:* Decisions live in DMs* Context is fragmented* No single source of truth existsFounders become archaeologists, digging through chat logs to understand why something happened.The Minimum Viable Operating SystemThis episode argues for a deliberately minimal stack, not enterprise process.1. Linear for ExecutionLinear integrates directly with GitHub.Status updates happen automatically.No nagging.No manual reporting.Work updates itself.2. Notion for MemoryNotion becomes institutional memory.Rule:If it’s discussed, it’s documented.This shifts the company from tribal knowledge to durable knowledge.Meetings That Don’t Suck: L10-LiteInstead of heavy frameworks like EOS, the episode recommends a single weekly leadership meeting:60–90 minutes. Same agenda. Every week.Agenda:* Wins (psychological momentum)* Scorecard (5–7 key metrics)* Priorities (on/off track)* IDS: Identify, Discuss, SolveMost meetings report status.This one resolves bottlenecks.You leave with decisions, owners, and deadlines.Delegation That Actually WorksDelegation is not assigning tasks.It’s assigning outcomes.Instead of:“Change the button color.”Say:“Customers can’t find the buy button. Fix that.”Frameworks discussed:* CEO Bubble: what only the founder should do* Decision Zones: green / yellow / red decisions* MSCL test: mandate, stakes, edge, leverageMost founders stay busy because they’re hiding in low-leverage work.The Real Shift: From Doing to DesigningAt three people:Your output = your work.At ten people:Your output = the system you designed.This is the hardest lesson.Teaching feels slow.Letting go feels dangerous.But founders who make this shift are ~3× more likely to reach a successful exit.Closing ThoughtIf you’re constantly firefighting, the problem isn’t effort.It’s architecture.The fire won’t disappear.But if you don’t build the fire station, you’ll be holding the hose forever.And eventually—you’ll run out of water. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

Most engineering failures don’t come from bad algorithms or insufficient data.They come from something much more basic:We didn’t define the system properly.That’s what this episode of Coordinated with Fredrik is about — system boundaries, thermodynamics, and how we should think about a home once it stops being a passive consumer and starts behaving like an active energy systemWhere does a system begin — and where does it end?This sounds almost philosophical, but it’s one of the most practical questions an engineer can ask.Every system needs a boundary. Without one, it’s impossible to reason about control, optimization, or even responsibility. This is true for software systems, mechanical systems, and very much so for energy systems.Ludwig von Bertalanffy, the father of systems theory, once said:“The boundaries of a system are not given in nature but are determined by the observer.”That’s true in many domains — but energy is special.In energy systems, the boundary is not arbitrary.It is physical, legal, and enforced.That boundary is the electricity meter.The meter is not a billing deviceWe tend to think of the electricity meter as something purely administrative — a device that exists to calculate our bill. But that’s a mistake.The meter is the point of common coupling (PCC) between your home and the grid.Everything you consume passes through it.Everything you export passes through it.It is where:* Ownership changes* Responsibility changes* Grid physics ends and home physics begins* Billing, tariffs, export limits, and fuse constraints applyIn thermodynamic terms, it is the membrane between two systems.Once you see the meter this way, the right question stops being “What is my inverter doing?” and becomes:What crosses this boundary, when, and under what constraints?That single shift changes everything.Why the old model worked — and why it brokeHistorically, homes were boring.They were passive loads.Power flowed in one direction.Individual behavior didn’t matter much.From the grid’s perspective, you could aggregate thousands of homes and get remarkably accurate forecasts. The system was statistically predictable because nothing interesting happened at the edges.So our tooling reflected that worldview.We read registers.We polled Modbus TCP.We collected telemetry.And for a long time, that was enough.The moment homes stopped being predictableThen we added things.Solar PV at the edges of the grid.Batteries that store energy over time.Electric vehicles with large, deadline-driven loads.Heat pumps with thermal inertia and weather-dependent efficiency.Suddenly:* Power flows both ways* State matters (SOC, temperature, availability)* Timing matters more than magnitude* Homes can go from “doing nothing” to exporting 8 kW in secondsFrom the grid’s point of view, a home that used to be a smooth, boring signal becomes bursty, stateful, and hard to predict.A house might sit at zero net flow for hours — perfectly balanced by solar and storage — and then abruptly inject a large amount of power when a battery fills up or a cloud passes.The old statistical assumptions no longer hold.A short detour into thermodynamics (the useful parts)Thermodynamics gives us the correct mental model for all of this.Clausius summarized the first and second laws in a single sentence:“The energy of the universe is constant. The entropy of the universe tends to a maximum.”Everything that happens inside a home — or any site — sits inside that frame.The first law: accountingEnergy doesn’t disappear. It transforms.For a home:* Energy can be stored chemically (batteries)* Stored thermally (hot water tanks, slabs, buildings)* Converted between electrical and thermal forms* Exported or imported across the meterPower is just energy per unit time.Storage is what happens when generation and consumption don’t align in time.In that sense, storage isn’t a device category.It’s a consequence of time mismatch.The second law: usefulnessThe first law tells us energy is conserved.The second law tells us not all energy is equally useful.Electricity is high-quality energy.Low-temperature heat is low-quality energy.You can easily turn electricity into heat.You can’t easily turn heat back into electricity.This is why heat pumps matter so much: they don’t create heat — they move it, exploiting temperature differences to deliver more heat than the electrical energy they consume.None of this is optional. Software that ignores the second law will always look good in simulations and fail in reality.From signals to systems: Site, Device, DERThis is where thermodynamics meets software architecture.SiteThe site is the system boundary.Everything behind the meter.A site has:* Objectives (cost, comfort, self-consumption, grid services)* Constraints (main fuses, export limits, tariffs)* State that evolves over timeOptimization only makes sense at this level.DeviceA device is something you can communicate with.It has:* Protocols (Modbus, REST, cloud APIs)* Registers* Firmware versions* Vendor quirks and bugsDevices answer the question:What can I technically talk to right now?That’s necessary — but insufficient.DER (Distributed Energy Resource)A DER is a logical abstraction.It represents capability, constraints, and state — independent of protocol.A battery DER might represent:* Total capacity* Current SOC* Charge/discharge limits* EfficiencyWhether that battery consists of one module or twenty cells doesn’t matter unless it affects system behavior.DERs answer the real question:What can this resource do for the system?Devices are how you talk.DERs are what you reason about.Why this abstraction mattersOnce you define:* The boundary (the site)* The resources (DERs)* The constraintsControl stops being reactive.The problem becomes:What should the energy flow across the meter look like over time?The grid doesn’t care how your system is wired internally.It cares about magnitude, direction, and timing at the boundary.In that sense, the meter becomes the objective function.Homes are no longer loadsA modern home has:* State* Constraints* Objectives* Time-coupled decisionsThat’s not a load.That’s an agent.We inherited an energy system architecture from a time when homes were boring. They aren’t anymore. That creates real challenges — but also real opportunities.And none of them can be addressed without going back to first principles and defining the system correctly. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

Episode SummaryWe tend to think about infrastructure in physical terms: wires, pylons, transformers, steel and copper. But modern systems—especially energy systems—are held together by something less visible and just as critical: the messaging layer.In this episode, we trace the hidden history of how machines learned to talk to machines. From Wall Street trading floors to oil pipelines in the desert, from telecom switches in Sweden to rage-coded weekends in Silicon Valley, this is the story of frustration-driven innovation.We explore:* Why synchronous “telephone-style” software broke at scale* How publish/subscribe became the software equivalent of a system bus* Why RabbitMQ, Kafka, NATS, and MQTT exist—and what specific pain each one was born to solve* The architectural tradeoffs between smart brokers and dumb pipes* Why replayability, liveness, and reliability are fundamentally different goals* How modern systems increasingly combine all of these tools* And why the next architectural leap will come from today’s friction pointsThis episode isn’t about choosing the “best” messaging system.It’s about understanding why each one exists, and what happens when you use the wrong tool for the wrong kind of problem.Key Concepts* Messaging as the nervous system of physical infrastructure* Subject-based addressing and decoupling* Smart broker vs. dumb broker architectures* Append-only logs and replayability* Control planes vs. data planes* Edge constraints and low-power networks* Friction as a signal for architectural evolutionMentioned Systems & Ideas* TIBCO and the original information bus* AMQP and the open-standard rebellion* RabbitMQ and Erlang’s “let it crash” philosophy* Kafka and the log as the source of truth* NATS and the “dial tone” model* MQTT and constraint-driven protocol design* ZeroMQ, Pulsar, Redis Streams (briefly)The Invisible GridHow Messaging Systems Became the Nervous System of Modern InfrastructureClose your eyes for a moment.(Not if you’re driving—but mentally.)When we talk about infrastructure, we picture the physical grid: copper wires, transformers humming in empty fields, pylons cutting across landscapes. It’s tangible. You can touch it. You can see it rust. You can watch a tree fall on it.If that grid fails, everything stops.But there is another grid—one we almost never visualize.An invisible grid, running inside software.A nervous system made of messages.And just like the physical grid, when this system clogs, desynchronizes, or collapses, the lights go out anyway—no matter how much copper is in the ground.Modern energy systems, financial markets, cloud platforms, and industrial control loops don’t merely use software. They depend on it at the level of physics. Signals must arrive on time. Control decisions must propagate. State must remain coherent across thousands or millions of moving parts.This post is about how we got here.Not as a clean, planned evolution—but as a genealogy of frustration.The Original Sin: The Telephone CallEarly software systems communicated the same way humans did: by calling each other directly.Application A opens a connection to Application B, waits for it to respond, sends data, and blocks until it hears back. This is synchronous coupling—the software equivalent of a phone call.It works fine for two systems.It collapses at scale.On a trading floor—or an energy grid—one event must fan out to many consumers: risk engines, dashboards, control systems, settlement layers. In the telephone model, the sender must call each one, sequentially.If any receiver is slow or unavailable, everything backs up.Latency accumulates. Failure cascades.In finance, you go bankrupt.In energy, you destabilize the grid.This brittleness created the first great insight.The Software Bus: Publish, Don’t CallIn the mid-1980s, an engineer looked at a computer motherboard and asked an uncomfortable question:Why is software dumber than hardware?A CPU doesn’t “call” the graphics card. It broadcasts onto a system bus. Whoever is listening picks up the signal. The sender doesn’t care who receives it—or if anyone does at all.That idea became publish/subscribe.Instead of sending data to addresses, you publish it to subjects.Instead of knowing who consumes it, you just agree on what it means.This decoupling was revolutionary. It gave us the first real software nervous system.And it worked—so well that it created the next problem.When Middleware Ate the BudgetBy the early 2000s, large enterprises had dozens of incompatible messaging systems. Each vendor had its own protocol, its own servers, its own licensing model.Banks were spending absurd portions of their IT budgets not on business logic—but on plumbing.The rebellion that followed wasn’t technical at first.It was economic.Why don’t we have a TCP/IP for messaging?That question led to open standards. And open standards led to open source.RabbitMQ and the Power of “Let It Crash”RabbitMQ emerged from a near-perfect alignment between problem and tool.The problem: routing messages reliably, flexibly, transactionally.The tool: Erlang—a language built for telecom switches that cannot go down.Erlang’s philosophy is radical: don’t prevent failure—contain it.Instead of one giant program sharing memory (where one bug burns the house down), Erlang runs millions of tiny isolated processes. If one crashes, a supervisor instantly replaces it.Failure becomes routine. Boring. Managed.RabbitMQ embodies this mindset. It is a smart broker: it routes, retries, buffers, tracks acknowledgements, and guarantees delivery.It is a post office.And like all post offices, it has limits.Kafka and the Log That Changed EverythingWhen LinkedIn tried to track everything, the post office model broke.Too much sorting. Too much state. Too much overhead.The breakthrough was deceptively simple:Stop routing messages. Start recording history.Kafka treats data as an append-only log—an immutable sequence of events. Producers write to the end. Consumers read at their own pace. The broker doesn’t track who’s done what.This aligns perfectly with disk physics. Sequential writes are fast. Replays are free. History becomes an asset.In this model:* The log is the source of truth* Databases are just materialized views* You can replay the past with new intelligenceKafka isn’t a post office.It’s a newsstand.NATS and the Dial ToneThen came cloud platforms, microservices, and another frustration.Messaging systems had become pets—delicate, stateful, needy.But cloud infrastructure demands cattle—replaceable, disposable, boring.NATS was born from that tension.Its original design was ruthless:* No persistence* No buffering for slow consumers* No guarantees beyond “best effort right now”If you’re too slow, you’re dropped.If no one’s listening, the message vanishes.This sounds dangerous—until you realize what it’s for.Control planes. Heartbeats. Service discovery. Real-time signals where the latest state matters more than history.NATS is not a database.It’s a dial tone.MQTT: Innovation Under ConstraintThe most elegant designs often come from the harshest constraints.MQTT was built for oil pipelines in the desert, running over satellite links so slow and expensive that saving two bytes mattered.The result was a protocol stripped to its bones:* Tiny headers* Persistent low-power connections* Explicit handling of unreliable networks* A “last will and testament” for dead devicesYears later, the same properties made MQTT perfect for smartphones.From oil rigs to billions of pockets.Today, MQTT is the language of the edge.Synthesis: No Winners, Only TradeoffsThere is no perfect messaging system.Each of these tools exists because an engineer hit a wall:* Too slow* Too heavy* Too expensive* Too fragileThey encode those frustrations into architecture.That’s the real lesson.Modern systems don’t pick one.They compose:* MQTT at the edge* Kafka for history and analytics* NATS for control and coordination* RabbitMQ for transactional workDifferent pipes for different fluids.The Real QuestionEvery major evolution in messaging came from irritation.A system that made engineers sigh.A component everyone dreaded touching.A piece of infrastructure that fought back.So here’s the closing thought:Where is that friction in your system today?That’s not technical debt.That’s a signal.The next nervous system will be built there. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

Most conversations about supplements are useless.They start too early, aim too low, and ask the wrong question.This is not a beginner’s guide. This is not “eat your vegetables” or “drink more water.” This is a discussion for someone who has already done the hard part.You wake up early.You train consistently.Alcohol is gone.Food is clean.Sleep is a priority.In other words: the foundation is already built.So the real question becomes uncomfortable, almost heretical:When you’re already operating at the 95th percentile of discipline, does supplementation actually move the needle — or are you just creating expensive urine?This episode of Coordinated with Fredrik was built to answer that question with one lens only: return on investment.Not vibes. Not biohacker cosplay. ROI.The Core Thesis: No Magic Pills, but Real Feature UpgradesAfter combing through meta-analyses, randomized controlled trials, and clinical protocols published through early 2025, the answer is not a naïve yes.It’s a qualified yes.Supplements do not replace fundamentals. They do not compensate for poor sleep, weak conditioning, or a chaotic life. But even a perfect modern lifestyle leaves gaps — gaps created by:* Soil depletion* Indoor work* Chronic cognitive load* Latitude and lack of sun* Stress-induced mineral lossThe opportunity lies in biological arbitrage: small chemical inputs that produce disproportionate output for a high-functioning executive.Out of all the noise, four compounds consistently survive scrutiny.Not exciting. Not exotic. Just effective.The Core Four1. Creatine: Decision Fatigue InsuranceCreatine has undergone one of the most dramatic rebrandings in modern science.Once dismissed as “gym bro powder,” it is now increasingly understood as cognitive fuel.The reason is simple: the brain is an energy-hungry organ. When ATP runs low, processing speed, attention, and working memory degrade. Creatine acts as the fastest phosphate recycling system in the human body — a rapid charger for neural energy.Recent meta-analyses show statistically significant improvements in:* Short- and long-term memory* Attention span* Processing speedThe benefits are strongest under metabolic stress: sleep deprivation, high cognitive load, intense work periods. In one striking study, a single high dose of creatine largely preserved cognitive performance during 21 hours of wakefulness.For a CEO, this is not about muscle.It’s about staying sharp when the margin for error is thin.Protocol* Creatine monohydrate only* 3–5 g daily* No loading phase* Consistency over intensityKidneys: safe.Hair loss: unsupported by evidence.Marketing variants: ignore them.2. Magnesium: The Off SwitchIf creatine is about output, magnesium is about recovery.Magnesium quietly governs over 300 enzymatic reactions, including energy metabolism, neural signaling, and muscle relaxation. Yet 50–60% of adults in the Western world are insufficient.Why?* Mineral-depleted soil* Stress-driven magnesium loss* Caffeine and training increasing excretionFor high performers, magnesium deficiency is almost structural.Recent imaging studies link adequate magnesium intake to structurally younger brains, fewer white matter lesions, and improved cognitive resilience. One targeted form, magnesium L-threonate, has shown particularly strong effects on brain magnesium levels due to its ability to cross the blood–brain barrier.Choosing the form* L-threonate: cognitive longevity, focus, brain health* Bisglycinate: sleep quality, relaxation, recovery* Oxide: avoid (unless you want a laxative)Take it in the evening. Make it a ritual.3. Omega-3s: Structural Maintenance of the BrainOmega-3s are not supplements. They are building materials.EPA and DHA are structural components of neuronal membranes. Their status can be measured directly through the omega-3 index — a four-month rolling average of membrane composition.Targets matter:* 8%: associated with cardiovascular and cognitive protection* 10–12%: what aggressive optimizers aim forHigher omega-3 index levels correlate with:* Reduced cardiovascular events* Slower cognitive aging* Improved mood stabilityPlant-based omega-3s (ALA) do not convert efficiently. For DHA, conversion is effectively negligible. If you don’t eat fatty fish, supplementation is non-negotiable.Key insight: do not guess. Test. Adjust dosage based on blood data.Potency vs purity is a real trade-off:* Fish oil: higher doses, contamination risk* Algae oil: cleaner, lower doses* Fish roe: superior bioavailability, lower required intakeThere is no ideology here. Only measurement.4. Vitamin D: System-Wide RegulationVitamin D is misclassified. It is not a vitamin in function — it is a hormonal regulator.It directly influences the expression of over 1,000 genes. Above the 37th parallel, winter synthesis is effectively zero. For Northern Europe, deficiency is structural.Beyond bone health, recent data links adequate vitamin D levels to:* Slower telomere shortening* Lower dementia incidence* Improved immune and metabolic regulationThe mistake is dosing without context.High-dose vitamin D without:* Magnesium (cofactor)* Vitamin K2 (calcium traffic control)…creates risk.Safe stack* Vitamin D3: often 5,000 IU+ (test to confirm)* Vitamin K2 (MK-7): 100–200 mcg* Adequate magnesium intakeThis is an ecosystem, not a pill.The Stoic Frame: Don’t Major in the MinorsSupplements are the final 5%.They polish the machine — they do not build it.Zone 2 cardio.Sleep.Strength training.Consistency over decades.Miss a day? Nothing breaks.Forget a week? No catastrophe.The stoic advantage is restraint.The goal is not obsessive optimization. The goal is protecting cognitive capital — the asset that compounds everything else you do.The real biohack isn’t copying anyone’s protocol.It’s knowing your own numbers.Measure. Adjust. Repeat.That’s the whole game.This post is adapted from the podcast episode transcript and reflects the discussion as presented in Coordinated with Fredrik . This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

I’ve been thinking about this paradox since my PhD days. Back then I was trying to make ships more fuel efficient, believing it would reduce emissions. Then I discovered William Stanley Jevons and his 1865 book “The Coal Question.” It nearly broke me.Jevons noticed something counterintuitive about James Watt’s steam engine. Watt made steam engines roughly four times more efficient than the old Newcomen engines. Common sense says this should have reduced coal consumption. The opposite happened. Coal use exploded.Why? Because efficient steam power suddenly became cheap enough to use everywhere. Factories, railways, mines. The efficiency didn’t save resources. It unlocked demand that hadn’t existed before.“It is wholly a confusion of ideas to suppose that the economical use of fuel is equivalent to a diminished consumption. The very contrary is the truth.”That quote has stuck with me for years.Now look at AI. In 2024, data centers consumed about 415 terawatt hours globally, roughly 1.5% of all electricity. Projections put that at 945 terawatt hours by 2030. Virginia already sends 26% of its electricity to data centers. Ireland is at 21-22%.DeepSeek showed you can train competitive models far more cheaply. Did that reduce compute demand? No. It opened the door for more companies to train more models. Same pattern as Watt’s engine.Here’s where I land on this: the paradox isn’t a warning. It’s a description of how economic systems work. Fighting it is pointless. The question is whether we can meet rising demand with clean, abundant energy.Solar is already the cheapest electricity source available. The sun has always been Earth’s power plant. Even the coal Jevons worried about is just ancient stored sunlight.We’re going to use more electricity. A lot more. That’s not a crisis. That’s an opportunity to finally build the energy system we should have had all along.More on this in future episodes.——SHOW NOTES:Episode recorded Sunday morning, Kalmar, SwedenThe History* William Stanley Jevons, Liverpool-born economist, published “The Coal Question” in 1865* UK consumed 93 million tons of coal annually at the time, nearly all of Britain’s energy supply* Coal production had grown 3.5% per year for the previous 80 years* British coal production didn’t peak until 1913, almost 50 years after Jevons wrote his warningThe Famous Quote “It is wholly a confusion of ideas to suppose that the economical use of fuel is equivalent to a diminished consumption. The very contrary is the truth.”The Steam Engine Story* Thomas Newcomen developed the atmospheric engine in 1712, less than 1% efficient* James Watt (born 1736) was given a Newcomen engine to repair in 1763* Conceived the separate condenser idea in 1765, patented in 1769* Watt’s engine was roughly 4x more efficient than Newcomen’s* Result: coal consumption exploded because steam power became economical for everythingThe Paradox More efficiency didn’t reduce coal use. It made steam power cheap enough to deploy everywhere. Textile mills, railways, factories. The efficiency unlocked demand that didn’t exist before.Modern Examples* More fuel-efficient cars → people drive more* LED lights → we install far more lights* Air conditioning → billions of new users as it became affordableAI and Data Centers Today* 2024: Global data centers consumed 415 TWh (1.5% of global electricity)* 2030 projection: 945 TWh (nearly 3% of global electricity)* Data center electricity growing 15% per year (4x faster than overall electricity growth)* US: 4% of total electricity goes to data centers* Virginia alone: 26% of state electricity to data centers in 2023* Ireland: 21-22% of national electricity to data centersThe DeepSeek Parallel Early ChatGPT models might be the Newcomen engine of AI. Useful for specific tasks but extremely inefficient. DeepSeek and other breakthroughs are making AI cheaper and more efficient. According to Jevons Paradox, this won’t reduce compute demand. It will increase it.The Optimistic Take Rising electricity demand isn’t a problem if we meet it with abundant, cheap, clean sources. Solar is now the cheapest electricity source on earth. The sun has always been Earth’s power plant. Even fossil fuels are just stored sunlight. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

In this episode, I share a candid, thinking-out-loud reflection on how fast agentic coding is evolving—and what that speed means if you’re building and leading a company right now.From a CEO’s perspective, I talk about why old mental models for software development no longer fully apply, and why it’s no longer enough to ask how we build, but who (or what) should be doing which parts of the work. I don’t have all the answers—and I’m very explicit about that—but I believe it’s better to share pragmatic, half-formed thoughts than to wait for perfect clarity.Over the last few weeks, my own workflow has shifted dramatically. I describe moving from “sitting in a terminal pressing yes” to running agents on sandboxed virtual machines, orchestrating long-running tasks, and increasingly treating AI systems less like tools and more like collaborators. At times, it genuinely feels magical—and at other times, cognitively overwhelming.“If you can suddenly build things that weren’t possible just weeks ago, you can’t rely on old standards of thinking.”I also explore what feels like the next inflection point: moving beyond single agents toward true multi-agent orchestration, where planning, coordination, and even delegation are increasingly handled by AI itself. Today, humans still act as the operator-in-the-loop. Tomorrow, that role may shift upward—toward taste, direction, and judgment rather than execution.Throughout the episode, I reflect on what this means for engineers, “intelligence workers,” and small teams—especially in high-impact domains like the energy transition, where I’m grateful to be building with a 10-person team using these new tools.This episode is less about conclusions and more about momentum. It’s a snapshot from an extremely exciting moment in time—shared in public, in real time, while everything is still changing. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

For the first time, this isn’t AI-generated or scripted — it’s a live, unscripted monologue recorded early in the morning, just before heading to work. No plan, no outline. Just me speaking freely, partly for anyone listening, and partly for myself.I wanted to try this format because the last few weeks have felt intense. The pace of change in AI — especially what many are calling the agentic era — has accelerated in a way that feels qualitatively different from past hype cycles. This time, it feels real. And with that comes both excitement and stress.One recurring feeling I talk about is how traditional software moats are eroding. Having large codebases, teams of engineers, and long-lived products used to be a clear advantage. Increasingly, those same assets can also become constraints. The more you own, the more you have to maintain — and the slower you can move. There’s a growing case for minimalism: owning less, building faster, and treating software as something that can be created, adapted, and discarded quickly.Internally, we’ve been pushing agentic coding tools hard, especially Claude Code. I’ve experimented with AI-assisted development for a long time — Cursor, VS Code integrations, Codex, and others — but something clicked recently. Around early December, the combination of Claude Code, Opus 4.5, and effectively unlimited usage changed how productive these tools felt. It finally started working the way it should. That shift has been genuinely mind-altering.Zooming out, I’m broadly optimistic. The evolution of AI and agentic tools is good for society and the economy, even if the transition will be messy. Uncertainty creates opportunity, and builders who enjoy being at the frontier are likely to thrive. We may even be closer to AGI than many previously thought, which means second-order effects we haven’t anticipated yet — including in areas like healthcare, where people closest to real problems could now build their own solutions at almost no cost.Of course, most people won’t take advantage of this. Just like with the internet, the tools will empower a minority — but that minority is no longer blocked by access to capital, engineering teams, or specialized skills. Ideas and the will to execute are becoming the main bottlenecks.This episode is short, informal, and experimental. I’m not yet sure if this format will continue, but it felt worth trying. Sometimes thinking out loud is the best way to understand what’s actually changing.More to come — maybe. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

A War Room, Not a HolidayThis episode of Coordinated doesn’t begin with a product launch or a funding announcement. It begins in silence.December 26, 2025. The food is gone, the house is quiet, and most of the tech industry is still mentally offline. That’s when Fredrik, CEO of Sourceful, opens Claude Code “just to see what it can do.”What follows over the next four weeks—through January 22, 2026—is not incremental progress. It’s an existential reckoning.This episode is a case study in leadership during what increasingly feels like an agentic inflection point: a moment when the constraints that defined software work for decades simply collapse.The Snowy EpiphanyAt first, the logs show curiosity. Then surprise. Then stress.By the morning of December 26, Fredrik is no longer experimenting—he’s replicating core parts of Sourceful’s codebase. Rust pipelines. Test suites. Simulated runs of the ZAP platform. All in a single weekend.This is not empowering. It’s destabilizing.If a CEO—someone who does not identify as a traditional developer—can do this in days, what does that say about the barriers the organization assumed were real?The realization lands hard: the moat was never syntax. It was time. And time just collapsed.That emotional whiplash mirrors what others in the industry were reporting at the same moment. Andrej Karpathy described feeling both 10× more powerful and simultaneously behind. Speed had increased, but so had the ceiling—and the floor had dropped out entirely.January 3rd: The ManifestoOn January 3rd, the private reckoning becomes public.Fredrik posts a blunt, unambiguous message internally. It is not inspirational fluff. It is a directive:* Do not identify as a developer.* Identify as an engineer who solves problems.* There is no leaning on tenure.* I expect a few terminals crunching away all the time.The framing is binary: adapt, or the company is doomed. Harness this shift, or miss the most exciting opportunity of a lifetime.This wasn’t happening in a vacuum. Across the industry, leaders were independently reporting similar accelerations. David Holz and Tobi Lütke both described shipping more in weeks than in years through what’s now casually called “vibe coding.”This wasn’t hype. It was structural.From Coding to OrchestrationOne signal makes the shift undeniable: Stack Overflow traffic falling back to 2008 levels.The era of searching for syntax is ending. We are moving from task execution to system orchestration.For Sourceful—working with real-world energy grids, physical infrastructure, and regulatory constraints—this matters deeply. The work was never about typing code. It was about understanding systems. The new tools simply strip away the pretense.The Infrastructure of OneYou can’t demand 10× speed without providing a rig.That leads to one of the most concrete takeaways of the episode: the Mac Mini employee.The idea is simple:* Your laptop is for you.* Your agent needs a persistent home.A headless Mac Mini, running 24/7, accessed remotely, becomes a permanent agent workstation. No sleeping when the laptop closes. No lost context. No runaway cloud bills.For long-running simulations, inference loops, and overnight experimentation, a $600 machine outperforms many cloud setups on both cost and continuity.This isn’t theoretical. It’s operational leverage.Why Monorepos Win in an Agentic WorldAnother reversal emerges: monorepos.Microservices and repo fragmentation were designed to reduce human cognitive load. Agents have no such limitation. They want context—all of it.In a monorepo, an agent can see:* The API* The frontend that consumes it* The database schema underneath* The simulation models upstreamHallucinations drop. Refactors become holistic. A change to how solar output is modeled can propagate cleanly across the entire stack in one pass.Fragmented repos blind your most powerful worker.The RALF Loop: Engineering Over TypingAt the heart of the episode is the RALF loop (Read–Analyze–Loop–Fix), popularized by Jeffrey Huntley.The premise is uncomfortable but increasingly true: typing code is becoming a commodity. Engineering—thinking, specifying, anticipating failure—is not.In a RALF workflow:* Humans define the problem, constraints, and edge cases.* Agents write tests.* Agents write code.* Agents refactor.* Agents repeat until the spec is satisfied.Weeks of work collapse into hours—not because quality drops, but because feedback loops vanish.Persistence, Memory, and CouncilsThe biggest bottleneck isn’t intelligence. It’s memory.Tools like persistent Claude harnesses (internally nicknamed “snake oil”) compress massive conversation histories by up to 95% without losing decisions or constraints. That makes week-long engineering conversations viable again.For high-stakes systems, one agent isn’t enough. The episode outlines LLM councils: multiple agents independently proposing solutions, with a judge agent synthesizing and critiquing them.It’s peer review at machine speed.The Skill FlipOne of the most unsettling shifts discussed: designers are becoming top-tier builders.As Ryan Nystrom observed at Notion, product designers—armed with taste, empathy, and agents—are shipping production-ready systems. The AI handles syntax. Humans supply judgment.If your value was “I know React best,” you’re exposed.If your value is architecture, coherence, and system thinking, you’re more valuable than ever.This aligns with David Heinemeier Hansson’s updated triad:* Computers make it work* AI makes it fast* Humans make it beautifulStrategy: Platforms, Not FeaturesZooming out, the episode lands on a critical business implication.Generic SaaS is getting crushed. If an agent can build it in a week, it will race to zero.The value concentrates at the extremes:* Cultural, story-driven consumer apps* Deep, physical-world platforms: energy, biotech, defense, infrastructureSourceful lives on the second edge. You can’t “vibe code” grid compliance in Germany or physical connections to inverters.That’s why the warning from Gokul Rajaram matters: build platforms, not features. Features are faucets. Platforms are the pipes in the walls.The Open-Source DilemmaThe episode doesn’t dodge the dark side.As agents ingest public code wholesale, some maintainers—like Mark Schmidt—are closing source entirely, arguing that open code now directly trains replacements.For companies with real IP—physics models, grid algorithms—the question is unavoidable: what stays open, and what must be protected?There’s no easy answer. But pretending the trade-off doesn’t exist is no longer viable.Money as ExhaustThe episode ends not with tactics, but philosophy.Borrowing from Foz, Fredrik frames money as exhaust—not the goal, but the byproduct of a well-built engine.Obsess over the real problem. Build the strongest system possible. Let the exhaust take care of itself.That December stress wasn’t weakness. It was signal.While others unplugged for the holidays, Sourceful spotted the shift early. The task now isn’t panic—it’s institutionalization: using agents, infrastructure, and process to sustain speed without burning out humans.Or as the episode signs off:Don’t chase the exhaust.Build the engine.Listen to the full episode of Coordinated by Fredrik for the complete discussion and practical details behind this transformation.https://www.vibekanban.com/ This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

The most uncomfortable truth in crypto is not that the technology fails. It’s that it works exactly as designed—right up to the moment a human touches it.This episode begins with a paradox that should unsettle every technically literate founder: the largest losses in crypto history are not caused by broken cryptography, failed audits, or consensus bugs. They are caused by moments of trust, urgency, habit, desire, fear, and authority. The protocol holds. The person breaks.What we explored is not “how hacks happen,” but why sophisticated, rational, engineering-minded people lose generational wealth in systems that are mathematically secure. The answer is brutally simple: security thinking rarely extends beyond the protocol layer.Crypto didn’t eliminate trust. It relocated it—onto the human holding the keys.The Gap Between Secure Systems and Vulnerable OperatorsEvery case we examined followed the same structure. The cryptography was sound. The exploit occurred elsewhere.A prominent investor loses $24 million because a telecom employee accepts a $500 bribe. A DeFi CEO signs an irreversible transaction because habit overrides scrutiny. A startup founder loses $50,000 because optimism and social pressure disable skepticism. A software engineer loses six Bitcoin because romance becomes leverage.Different attacks. Same weakness.The attack surface is not the chain. It is identity, attention, emotion, routine, and fear.This is where engineering intuition often fails. Engineers expect adversaries to attack systems where entropy lives—in code, math, randomness. Instead, attackers go where predictability lives: human behavior.Old Cons, New InterfacesNothing about these scams is novel.They are Ponzi schemes, advance-fee frauds, honey traps, authority impersonation, and extortion—centuries-old psychological weapons. Crypto simply gives them three properties that make them devastating: speed, irreversibility, and pseudonymity.A forged letter becomes a deepfake Zoom call. A bribe becomes a SIM swap. A blackmail envelope becomes a hotel room and a QR code. The medium changes. The playbook does not.What has changed is the payoff structure. A single successful attack can move millions in minutes, across borders, beyond recovery. That incentive justifies patience, sophistication, and hybrid digital-physical operations.The episode makes this explicit: modern crypto crime is layered. Social engineering enables technical exploitation. Technical compromise enables physical coercion. Digital footprints enable real-world targeting.Once you see this, “cybersecurity” feels like a dangerously incomplete word.The Engineering Failure Mode No One Likes to AdmitThe most important insight of the episode is also the most uncomfortable for founders: personal security posture is a first-order business risk.It does not matter how secure your protocol is if your keys can be coerced, copied, or socially engineered. It does not matter how many audits you pass if a single person can sign away irreversible value under pressure.This is why the conversation keeps returning to habits. Clicking confirm without reading. Trusting urgency. Believing authority. Treating personal devices as safe by default. Fragmenting security across accounts and wallets instead of thinking systemically.Engineers are trained to remove single points of failure from machines. Many still tolerate them in themselves.From Cryptographic Resilience to Human-Aware SecurityThe episode doesn’t end in paranoia. It ends in design.Security that works in this environment must assume humans are fallible under stress. That assumption changes everything. It leads to layered authentication that resists phishing, device separation that limits blast radius, multi-signature schemes that survive coercion, decoy strategies that reduce physical escalation, and operational habits that prioritize verification over speed.More importantly, it reframes the goal. The goal is not perfect security. The goal is to become an economically unattractive target.Criminals optimize for return on effort. Harden enough layers—technical, procedural, physical, reputational—and the model breaks. They move on.That is engineering logic applied correctly.The Unresolved QuestionThe episode closes on a question that deliberately remains unanswered: is individual hardening enough?As long as single humans can be coerced into signing irreversible transactions tied to their physical safety, there is a systemic problem that personal discipline cannot fully solve. Multi-sig wallets help. Social norms help. But the incentive landscape remains.The real battle may not be fought at the wallet level at all, but at the intersection of identity, custody, and social systems. Until then, the uncomfortable truth stands:In crypto, the weakest link is not the protocol.It is the person who believes they are rational under pressure. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

Modern leadership discourse is saturated with abstractions: vision, values, strategy, alignment. These words are cheap. They float easily above reality. What is scarce—vanishingly scarce in large organizations—is consequence.This episode of Coordinated revolves around a single, brutal question: who actually pays when decisions go wrong? Not rhetorically. Not reputationally. Literally.Nassim Nicholas Taleb popularized the phrase skin in the game, but the idea is far older than Taleb, older than capitalism, older than corporations. It is the civilizational insight that systems decay when decision-makers can externalize downside. When gains are private and losses are socialized, fragility is not an accident. It is the default outcome.In complex, high-stakes domains—energy, finance, infrastructure, defense—this asymmetry becomes lethal. Failures are not theoretical. They cascade. They compound. They don’t politely wait for quarterly reports.The central claim of this episode is simple and uncomfortable: leadership without personal exposure is not merely unethical; it is an engineering flaw .Moral Hazard Is Not a Personality ProblemWe like to explain systemic failure through individual bad actors. Greedy executives. Incompetent managers. Corrupt consultants. This is comforting and mostly wrong.The real culprit is structure.When executives collect upside through bonuses, stock options, or prestige, while the downside is absorbed by employees, customers, taxpayers, or “the market,” behavior predictably shifts. Risk migrates into the tails. Maintenance is deferred. Redundancy is cut. Catastrophic failure is postponed just long enough for the decision-maker to exit.Economists call this moral hazard. Taleb calls it fragility. Engineers recognize it immediately: remove feedback and the system lies to you.The absence of skin in the game is not a character flaw; it is a broken feedback loop. People do not learn. Organizations do not adapt. Errors accumulate invisibly until they rupture.This is why so many post-mortems sound identical. The same mistakes, recycled under new branding. The same “unexpected” failures that were, in fact, inevitable.Skin in the Game as a Truth FilterOne of the most important reframings in the conversation is this: skin in the game is not primarily about punishment. It is about epistemology.When your livelihood, reputation, or capital is on the line, reality corrects you quickly. Delusion becomes expensive. Wishful thinking burns.Taleb’s famous line captures it perfectly: “Don’t tell me what you think. Tell me what’s in your portfolio.”In organizations, this translates cleanly. Trust the arguments of people who will personally live with the consequences. Discount those who won’t. The best signal of belief is exposure.This is why consultants are structurally dangerous in mission-critical systems. They are paid upfront, insulated from long-term outcomes, and rarely present when consequences arrive. Their feedback loop is severed. They do not learn. The system pays the tuition.Skin in the game restores learning by re-coupling belief and consequence.Engineering, Not EthicsThe episode deliberately avoids framing this as a moral sermon. Ethics without enforcement are decoration.Historically, civilizations understood this viscerally. Hammurabi’s Code was ruthless but effective: if a builder’s house collapsed and killed the owner, the builder paid with his life. If it killed the owner’s son, the builder’s son died. Horrifying, yes. But as a safety incentive, unmatched.Modern societies rightly reject collective punishment and inherited guilt. Kant’s moral philosophy replaces outcome-based justice with duty and intent. But Kant’s framework assumes moral saints. Systems cannot.Skin in the game is the pragmatic synthesis. It does not require virtue. It requires alignment. Even flawed humans behave responsibly when their own downside is real.This is why skin in the game is best understood as systems design, not ethics. It is how you force honesty without trusting character.Leadership Credibility Is Bought, Not ClaimedIn infrastructure companies, trust is not built through vision decks. It is built through shared exposure.A founder who has their capital locked into the company, who uses the product personally, who absorbs pain before asking others to sacrifice—this sends a costly signal. Costly signals are the only credible ones.This is why symbolic gestures matter when they are real. When leaders cut their own compensation before cutting staff. When they remove executive luxuries during downturns. When they are subject to the same risks they impose on others.The late Nintendo CEO Satoru Iwata halving his salary to avoid layoffs wasn’t charity. It was leadership legitimacy. The organization responded with loyalty, innovation, and endurance.Leadership without visible vulnerability collapses under stress. People can smell asymmetry.Decentralization Is Not a Preference. It Is a Requirement.Skin in the game scales poorly under centralization. Distance dilutes consequence.Complex systems—energy grids, logistics networks, software platforms—require decisions to be made close to where reality manifests. Local engineers, operators, and teams hold knowledge headquarters never will.Hayek called this the knowledge problem. Taleb gives us the enforcement mechanism: empower the exposed.Decentralized authority paired with local accountability produces faster adaptation and fewer catastrophic failures. Centralized command paired with insulated leadership produces paralysis and brittle systems.History, from military disasters to grid collapses, confirms this relentlessly.The Founder Advantage Is TimePerhaps the most underappreciated dimension of skin in the game is time horizon.Founders with deep exposure think in decades. Hired executives with short vesting cycles think in quarters. The difference is not subtle.Research consistently shows that CEOs approaching equity vesting dates cut maintenance, inflate earnings, and pursue flashy acquisitions that destroy long-term value. They get paid. The system weakens.Institutionalizing skin in the game—long vesting periods, mandatory shareholding, post-tenure exposure—is how organizations manufacture a synthetic founder mindset.Without it, short-termism is not a risk. It is a certainty.The Price of CredibilityThis episode ultimately circles one unavoidable conclusion: leadership without consequence is fragile, and often fraudulent.The price of credibility is not rhetorical skill or moral posturing. It is personal exposure. Time. Capital. Reputation. Visible vulnerability.Skin in the game is the mechanism that forces alignment between words and reality. It turns leadership from a role into a risk-bearing function.If you are not exposed to the downside, you are not truly in charge. You are managing optics.And systems built on optics do not survive contact with reality.This essay is based on the episode “Skin in the Game” from the podcast Coordinated with Fredrik, exploring leadership, infrastructure, and accountability through the lens of systems design and philosophy This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

There’s a quiet failure mode creeping into modern leadership: you look faster and smarter because AI outputs are clean and confident, but your actual judgment gets weaker—because you stop doing the hard work that builds judgment in the first place. In this episode of “Coordinated with Fredrik”, we unpack a framework we call The Vigilant Mind Playbook—not as “AI wellness,” but as competitive strategy: how to keep an asymmetric edge when everyone has the same models.TL;DRAI doesn’t just change productivity. It can change how your brain allocates effort—and how your organization converges into the same strategic blind spots as your competitors. The playbook is about building “cognitive sovereignty by design”: protecting human judgment, forcing friction at critical decision points, and preventing the “algorithmic hive mind” from turning your strategy into a commodity.The core tension: optimization versus judgmentRight at the start, the episode frames the dilemma in a way every exec will recognize: in complex sectors (like energy), you live on ruthless optimization—yet the technology that boosts optimization can quietly erode the one asset you cannot replace: independent judgment.A line that lands like a punch: we’re watching a shift where “optimization replaces wisdom and performance becomes a substitute for truth.”That’s not just philosophical posturing—it’s a warning about how leaders start making decisions when “polished output” becomes a proxy for understanding.Cognitive debt: when convenience becomes a tax on your executive functionThe episode introduces a concept it calls cognitive debt, described as something that shows up not only in behavior but in neurological measures—specifically citing an MIT experiment using EEG brain scans while participants tackled complex writing/strategy tasks.The claim (as described in the episode) is blunt: the AI-assisted group showed weaker neural activity—less engagement in regions tied to attention, working memory, and executive function—while the non-AI group did the full “effort payment” themselves.Then comes the part executives should actually fear: when AI was removed, participants struggled to recall their own arguments and access the deeper memory networks needed for independent thinking; their brains had adapted to outsourcing.The episode translates that into an organizational risk: imagine analysts who rely on AI summaries for complex standards or technical domains—what happens when the model makes a subtle error, or the situation demands a contrarian insight the model can’t produce? You may no longer have the “neural infrastructure” left to spot the mistake.The psychology: AI as a System 1 machine (and your addiction to certainty)The playbook leans hard into behavioral science: Kahneman’s System 1 vs System 2 framing—fast/effortless intuition versus slow/deliberate reasoning—and labels AI as the “ultimate System 1 facilitator.” It gives instant answers and lets you short-circuit the productive struggle where real insight forms.This is where the episode drops a Greene-style provocation: “The need for certainty is the greatest disease the mind faces.”In other words: the AI doesn’t just give you information—it gives you a hit of certainty, and certainty is the drug that kills skepticism.The episode also references a Harvard Business Review study (as described in the conversation) where executives using generative AI for market forecasts became measurably over-optimistic—absorbing the machine’s confidence as their own—while a control group forced into debate and peer review produced more accurate judgments.The “algorithmic hive mind”: when your whole industry converges into the same mistakesThe risk scales. The episode names it: the algorithmic hive mind—not just individual laziness, but organizational homogenization. If every company uses the same foundational models trained on the same data and optimized for the same metrics, strategic edge “evaporates.”You get convergence, shared blind spots, and “optimized average performance”—and that’s exactly when you become fragile: the moment you need a truly different strategy, you realize your thinking has been homogenized by the tool.The episode uses the 2010 flash crash as the illustrative analogy: algorithmic homogeneity + feedback loops + speed = tiny errors amplified into systemic chaos, only stopped when humans hit circuit breakers.The point isn’t finance trivia. It’s the structural warning: when everyone’s automation aligns, it can amplify errors faster than humans can react.The executive move: institutionalize cognitive sovereignty (don’t “hope” for discipline)The episode’s most practical shift comes late: stop treating this as personal productivity hygiene and treat it as governance.It proposes moving AI oversight out of “IT” and into core strategy, potentially via an oversight body—a “cognitive sovereignty officer or counsel”—tasked with auditing AI use, assessing cognitive debt, and flagging over-reliance before it becomes a crisis.Then it gets concrete: structural mandates like mandatory human approval for high-risk decisions (pricing changes over a threshold, AI-suggested firings reviewed by HR/legal), described as “circuit breakers” that prevent automated mistakes from spiraling.And finally: stress test your AI like banks stress test financial models—against scenarios it hasn’t seen—so you confront limitations while still using the tool aggressively.Culture: reward the dissenter, or your org walks off a cliff politelyOne of the sharper cultural notes: people who question AI outputs can face a competence penalty—seen as “not trusting the tech.” Leadership has to shatter that dynamic and explicitly reward the person who challenges the machine.The episode also highlights a trust problem: “a third of employees actively hide their use of AI,” driven by stigma and fear of looking replaceable—creating a strategic weakness because best practices and flaw-spotting don’t circulate.In a Greene lens: secrecy is not “edgy.” It’s organizational self-sabotage when it prevents shared learning and accountability.The closing provocation: competence becomes cheap; wisdom becomes the only edgeThe episode ends with the thesis you probably want tattooed on your operating system:Most competitors will become faster, louder, and more confident with AI—but few will become wiser. When models are ubiquitous, competence becomes a commodity; the only non-replicable asset left is wisdom and contrarian judgment.And that’s the episode in one sentence: AI can optimize—but it can’t replace wisdom, because wisdom is governance over your own mind. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

For roughly a hundred years, power meant scale. Who could drill the most oil, burn the most coal, build the biggest power plant. The 20th century rewarded those who controlled supply through sheer size and centralized ownership.That era is over.We are now entering what I call the Coordination Century. And the companies that understand this shift will define the next hundred years of energy infrastructure.The Efficiency TrapHere’s something that took me years to fully internalize, even with a PhD in energy optimization: you cannot conserve your way out of an energy crisis.In the mid-1800s, William Stanley Jevons studied coal usage in Great Britain and noticed something that defied all logic. James Watt had just developed a steam engine that was vastly more efficient than anything before it. Less coal required for the same work. The obvious prediction? Coal consumption would drop.The opposite happened. Coal became so cheap per unit of work that entirely new applications emerged. Factories, trains, ships, industrial processes nobody had imagined. Consumption exploded. The efficiency gains accelerated the Industrial Revolution.This pattern repeats everywhere. Make cars more fuel-efficient, and people buy SUVs and commute longer distances. Make computing cheaper, and we invent entirely new industries that consume more electricity than previous generations could have imagined.The lesson is uncomfortable but essential: making energy cheaper or more efficient does not reduce consumption. It creates new uses. Humanity always finds ways to absorb abundance.Which means Sourceful cannot be an efficiency company. We have to be a coordination company.The Abundance ParadoxHere’s the part that should make you sit up straight: the generation problem is solved.The sun delivers roughly 1,000 times Earth’s entire energy needs every single hour. Solar panels and large-scale batteries have crossed a critical economic threshold. In most markets worldwide, building new renewables is now cheaper than running existing coal or gas plants.The hardware exists. The technology works. And yet we have a crisis.In southern Sweden — a near-perfect laboratory for the energy transition — we saw over 400 hours of negative electricity pricing in 2024. Some municipalities experienced closer to 700 hours. Think about what that means: energy generators paying the grid to take their electricity because the system is overloaded.A homeowner with solar panels gets paid to use more energy or simply shut off their system. Not because we lack power, but because we cannot coordinate it.The physics of electrical grids demand that supply and demand balance every single second. When an uncoordinated wave of solar hits the grid at noon and base-load plants cannot ramp down fast enough, system stability becomes threatened.This is not an energy crisis. It is a coordination collapse masquerading as an energy crisis.The hardware is ready. The operating system is broken.300 Million Mobile Power PlantsConsider the electric vehicle. Traditional utility executives see EVs as liabilities — millions of people plugging in after work, crashing the grid. A blackout waiting to happen.We see something completely different.A parked EV sits idle 95% of its life. The global fleet will grow from roughly 58 million vehicles today to over 300 million by 2030. Collectively, those vehicles will carry an estimated 2,800 gigawatt-hours of flexible storage capacity.That is more flexible storage than the entire legacy European grid was designed to handle.This is not a problem. This is the largest distributed battery deployment in human history, waiting to be orchestrated.The multi-trillion dollar question becomes: how do you coordinate millions of energy decisions made by private citizens across the globe, minute by minute?OAuth for EnergyToday, if you buy a Tesla Powerwall, it communicates with your Tesla EV and the Tesla app. A closed ecosystem. It has limited or zero communication with a competitor’s solar inverter or the real-time pricing signals from your local grid operator.Everything is siloed. Proprietary. You need permission to interact with anything outside that wall.This centralization stifles innovation and limits the value consumers can generate from assets they already paid for.What we are building at Sourceful is an open coordination primitive for distributed energy resources. Think of it like OAuth — when you log into a third-party app using your Google account, you do not give that app your password. You grant specific, temporary permissions. Access to your name and email. Nothing more. And you can revoke it instantly.Our coordination layer works the same way, but for energy assets. The owner controls the permissions. They decide who gets to access the flexibility of their battery or their EV charging schedule, for how long, and for what compensation. They can revoke access at any time.This is energy sovereignty. And once people experience true ownership over the value their assets generate, they will never go back to being passive consumers in a centralized model.The Numbers That MatterTheory is worthless without proof. In dynamic markets like Sweden, a standard home setup with solar and a small battery generates between $800 and $1,500 per year through a combination of bill reduction and actual revenue from grid services.Optimal setups — larger storage, connected EVs, multiple revenue streams — have generated up to €4,000 annually. These are verifiable figures backed by bank statements showing payouts from grid operators.How? By providing services the grid desperately needs: frequency response and peak shaving.The grid frequency (50 Hz in Europe, 60 Hz in North America) is the vital sign of balance. Deviate too much and you get blackouts. The grid needs assets that can respond in seconds to inject or absorb power. A massive centralized power plant takes minutes to adjust. A network of thousands of distributed batteries coordinated instantly can provide this service with unmatched speed. Grid operators pay a premium for that responsiveness.Peak shaving is simpler: reducing demand during the highest-cost hours (typically 5-8 PM) so utilities do not need to fire up expensive peaker plants for just a few hours. Coordinated home batteries and EVs can pause charging or draw from storage instead of the grid.The UK utility-scale battery market offers a cautionary tale. Everyone built for frequency response. Everyone optimized for the same niche. Revenues dropped 76% in two years as the market saturated.Distributed coordination avoids this trap. Geographic diversity. Asset heterogeneity. A home in a rural area handles voltage support. An EV fleet in a city does intraday arbitrage. Suburban batteries handle frequency response. The value per connection increases exponentially, not linearly.Why the Infrastructure MattersGrid coordination requires machine speed. Electricity balances every second. Rewards need to be paid instantly to change behavior.We need rails that can handle millions of tiny transactions at near-zero cost. Solana delivers 400-millisecond block times, finality in about 2.5 seconds, and sustained throughput of 3,000-5,000 transactions per second. Average transaction cost: $0.00025.Compare that to issuing 50,000 payments on a congested Ethereum mainnet at $5 per transaction. The economic model collapses before it starts.The technical choice is deliberate and performance-driven. But here is what matters more: blockchain is a tool, not the product. We use it where it creates non-negotiable value — transparent rules for compensation, efficient global rewards distribution, and robust authentication. The core narrative remains focused on energy coordination itself.The Mission AheadWe are heading to Solana Breakpoint 2025 in Abu Dhabi not as observers but as operators. The goal is simple: establish Sourceful as the definitive emerging power in the energy and infrastructure sector.Every conversation forces a choice between two futures. Centralized proprietary control, where utilities dictate terms and consumers remain passive. Or distributed permissionless participation, where asset owners control their flexibility and capture its value.The 20th century rewarded those who controlled supply. The 21st century will reward those who coordinate complexity.We are building the operating system for energy abundance. And we are just getting started.Listen to the full briefing on “Coordinated with Fredrik” wherever you get your podcasts. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

In this week’s deep dive on Coordinated with Fredrik, we tackled the $222 billion elephant in the room. That is the estimated amount manufacturers spend annually on maintenance caused by aging equipment.As engineering leaders, we face a brutal dilemma: How do we embrace Industry 4.0 and predictive analytics without ripping out the legacy systems that are actually keeping the lights on?. We hear so much about “digital transformation,” but in sectors like energy and water, reliability is everything3. You cannot just beta-test a substation.The answer isn’t to replace the past; it’s to coordinate it with the future.The “Operational Museum” and Trapped DataMost industrial environments are what I call “operational museums”. You have equipment from the 80s, 90s, and today, all trying to coexist. The biggest cost here isn’t just the spare parts; it’s the trapped data.We still see critical machine data confined to clipboards and paper logs. When data is entered manually hours later (or never), it becomes worthless for automation. You can’t build a predictive maintenance algorithm on data that is late or inaccurate.The 45-Year-Old Backbone: ModbusTo solve this, we went technically deep into the grandfather of industrial connectivity: Modbus.It was published in 1979 by Modicon. Why are we still talking about a protocol that predates the World Wide Web? Because it is royalty-free, ruthlessly simple, and it runs on everything.However, integrating Modbus is where the “coordination” headache begins:* The Architecture: It uses a Master-Slave architecture where slaves are passive—they only speak when spoken to.* The Addressing Trap: One of the biggest debugging time-sinks is the difference between the documentation (one-based indexing) and the actual message packet (zero-based addressing).* The Security Void: Modbus has zero native security. If you are on the network, you can read or write to any register.The Strategic Pivot: The GatewaySo, how do we bridge a 1979 protocol with 2025 AI analytics? We don’t rip out the machines. We invest in intelligent gateways.The strategy discussed in this episode involves using gateways to translate raw, local Modbus data into modern, IT-friendly protocols like MQTT or OPC UA.* OPC UA adds context and security, modeling the data so a register isn’t just a number, but a defined temperature value.* SunSpec standardization helps manage multi-vendor solar plants by ensuring registers are always in the same place.The Human FactorFinally, we can’t ignore the biological component of the system. Automation fails when the people on the floor reject it. We looked at a case study where foundry workers rejected a “cobot” because they saw it as a threat.Successful coordination means redefining the human role from “operator” to “supervisor and augmenter”The TakeawayDigital transformation isn’t a binary choice between old and new. It is an “and” strategy. By using gateways to coordinate legacy reliability with modern analytics, we can turn dormant data into immediate operational levers.Listen to the full episode to hear the technical breakdown of RS-485 best practices and why “daisy chaining” is non-negotiable. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

There’s a dangerous seduction in the frontier of AI right now. Models grow bigger; their outputs shimmer with coherence; their corporate parents brag about breakthroughs every quarter. We’re told these systems are “intelligent,” “aligned,” “safe enough,” and increasingly marketed as replacements for human reasoning itself.This episode of Coordinated with Fredrik steps directly into that fog and lights it on fire.We didn’t talk about hype cycles, product announcements, investment theses, or the standard “AI will change everything” fluff. Instead, we went straight for the jugular: What does it mean to build critical infrastructure on top of systems that, at their core, are statistical ghosts?That is not a philosophical question. It’s an engineering one—and it cuts right to the bone of every power operator, every defense agency, every financial risk desk, and every CEO foolish enough to mistake fluency for reliability.The transcript doesn’t pull punches, and neither should the blog.The Core Problem: AI Is Brilliant in All the Ways That Don’t Matter When Something BreaksLLMs are extraordinary mimickers. They compress the entire textual history of the species into latent space and spit out answers that sound like understanding.But sounding right and being right are not synonyms.The models are built on a single objective: predict the next token. Everything else—tone, narrative, logic, persuasion—emerges as a side effect of statistical training. There is no causal model of the world underneath. No physics. No grounding. No internal consistency check. No understanding of error.This design philosophy works beautifully until you hit the real world, where reality doesn’t care about your linguistic confidence or your probability distributions.And the deeper we went in this episode, the more the foundational cracks widened.Safety Layers Built on the Same Fragile Foundation Will Fail TogetherThe conversation lays out a bleak but necessary warning: the AI safety techniques the industry relies on today—RLHF, reward models, fine-tuning, interpretability tools—aren’t independent safety layers. They are all built on the same substrate, and therefore:When one fails, all fail. Simultaneously. Catastrophically.The term from safety engineering is “correlated failure modes,” and if you work in nuclear plants, aviation, or grid stability, that phrase is synonymous with nightmares.To put it plainly:We aren’t stacking safety layers; we’re stacking different expressions of the same statistical fragility.You can’t build a reliable defense-in-depth system when every layer is made of the same soft metal.The Ghost in the Machine: Why LLMs Break Under StressOne of the most unsettling parts of the discussion is how LLMs behave during long-chain reasoning.A small mathematical misstep early in a multi-step problem doesn’t simply produce a slightly wrong answer. It cascades. It compounds. And the model has no internal mechanism to realize something’s gone off the rails.A ghost predicting its own hallucinations.This is why they fail at:* precise multi-step math* edge-case logic* novel reasoning* rare-event forecasting* high-risk decision chains* anything that requires causal coherence rather than textual familiarityThese limitations aren’t bugs. They aren’t “work in progress.” They are structural. The architecture is fundamentally reactive. It cannot generate or test hypotheses. It cannot ground its internal model in reality. It cannot correct itself except statistically.This makes it lethal in systems where one wrong answer isn’t an embarrassment—it’s a cascading blackout.The March of Nines: Where Statistical AI Meets the Real World and LosesIn energy—and in any mission-critical domain—the holy grail is reliability.Not 90%. Not even 99%. You need the nines.99.9%99.99%99.999%Each additional nine is an order of magnitude more difficult than the last.What this episode exposes is that statistical systems simply cannot achieve that march. It’s not an optimization problem. It’s not an engineering inefficiency. It’s not a matter of “more compute.”The long tail of rare events will always break a system that learned only from the past.LLMs are brilliant at the median but brittle at the edges.Critical infrastructure lives at the edges.So What’s the Alternative? Systems That Don’t Just Predict—They Learn.The second half of the episode explores a radically different paradigm: agentic, experiential AI—embodied systems that learn like animals, not like autocomplete machines.Richard Sutton’s OAK (Options and Knowledge) architecture is one such blueprint. It insists on:* learning from interaction* forming internal goals* developing durable skills* building causal models over timeBiology didn’t evolve by reading a trillion documents. It evolved by experiencing the world, failing, adapting, and iterating through an adversarial outer loop that never ends.LLMs are not alive in that sense. They have no loop. No surprise. No grounding. No experiential correction. No self-generated goals.If we ever want truly reliable artificial intelligence, we will need systems that build themselves through interaction rather than ingestion.This is not convenient for companies chasing quarterly releases—but reality doesn’t care about convenience.Energy Systems Are Too Important to Outsource to Statistical GuessworkIf you’re coordinating a decentralized energy network, integrating millions of DERs, predicting rooftop solar volatility, or steering a virtual power plant through a storm, you don’t need mimics.You need agents that understand cause, effect, uncertainty, and consequence.You need intelligence that does not collapse under rare events or novel conditions.You need systems that don’t just summarize history—they survive it.That’s the central tension exposed in this episode:Will the world keep chasing fast, brittle, statistically impressive tools?Or will we pay the safety tax required to build something real?It’s an engineering question with civilization-level implications.Final ThoughtThe future of infrastructure—energy, mobility, logistics, defense—is about to collide with the limits of statistical AI.This episode isn’t a warning. It’s a calibration. A reframing. A demand for seriousness in a time when the world is drowning in hype.If we’re going to build the next century, we need systems we can trust in the dark when everything else is failing.Statistical ghosts won’t get us there.The Infographic from NotebookLM: This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

The history of how humanity manages data is not just a technological timeline; it is fundamentally a story about our changing relationship with time and memory in digital form. From the moment we started scratching cuneiform on clay tablets, we sought to capture a record. The modern database system, however, has evolved from a simple recorder of the “now” to a dynamic, four-dimensional steward of the entire “story”.Join us on this grand odyssey through the architectural revolutions that redefined data storage.Act I: The Age of Rigidity (1890s – 1990s)The first mechanical memory was born in 1890 when Herman Hollerith adapted punched cards for the U.S. Census, dramatically cutting the tabulation time and laying the groundwork for IBM. This mechanical era gave way to electronic databases in the 1960s, driven partially by the NASA moon race.This period was dominated by Navigational Databases:* Hierarchical Models: IBM’s Information Management System (IMS, 1968), created for the Apollo program’s monumental Bill of Materials, organized data as parent-child trees.* Network Models: The CODASYL standard allowed more flexible, spiderweb-like relationships using physical pointers.These systems were blazingly fast for the queries they anticipated, but they were rigid; accessing data in an unanticipated way was cumbersome.The Relational HegemonyThe watershed moment came in 1970 with E.F. Codd’s introduction of the Relational Model. Codd proposed separating the logical schema (tables, rows, and columns) from the physical storage, enabling users to declare what data they wanted without knowing how to navigate the pointers. This led to the creation of SQL, the lingua franca of data, and established Relational Database Management Systems (RDBMS) as the gold standard by the 1980s.RDBMSs guaranteed ACID transactions (Atomicity, Consistency, Isolation, Durability) and were universally powered by the B-Tree data structure.Act II: The Great Architectural Pivot (2000s)The relational model excelled at transactional integrity, but it came with a fundamental flaw for the web era: its reliance on B-Trees mandated an update-in-place philosophy, which destroyed the historical context of data. A bank balance was simply overwritten.The “big data” explosion—the need to ingest millions of machine-generated events per second—broke the B-Tree architecture. Why?* Random I/O on Writes: Updating a B-Tree requires modifying scattered internal nodes, leading to excessive random I/O operations and severe bottlenecks.* Write Amplification: To ensure durability, RDBMSs often write data multiple times (to the Write-Ahead Log, and then to the B-Tree page), doubling or tripling the I/O load.The solution lay in the emergence of NoSQL and a fundamental architectural divergence: the Log-Structured Merge (LSM) Tree.MetricB-Tree (RDBMS)LSM Tree (TSDB/NoSQL)ImplicationsWrite PatternRandom I/O (Update-in-place)Sequential I/O (Append-only)LSM is superior for ingestion speed.Write AmplificationHigh (WAL + Page splits)Lower (Sequential flush)LSM minimizes immediate I/O but pays a deferred cost during compaction.Read AmplificationLow (Direct seek)Higher (Must check multiple files)LSM reads may degrade if compaction lags.LSM trees, popularized by Google’s BigTable, prioritize sequential disk writes, which are orders of magnitude faster than random writes. Data is first written to a fast in-memory Memtable and simultaneously appended to a sequential Write-Ahead Log (WAL) for durability. When the Memtable fills, it is flushed to disk as an immutable Sorted String Table (SSTable). Background Compaction processes merge these files, discarding old data.This architecture was not just a technical optimization; it was a philosophical acceptance that storage is cheap, but random I/O is expensive, dictating the software design of the last decade.Act III: Time Becomes the Primary DimensionAs storage costs plummeted, the focus moved from merely recording the “now” to capturing the “forever,” treating time not just as an attribute but as a primary dimension. This drove the evolution of Time Series Databases (TSDBs).TSDBs are designed specifically for data points with timestamps (like sensor readings or metrics) and are typically append-only, using columnar formats and heavy compression.GenerationKey SystemInnovation / MechanismFixed-Size Era (Gen 1)RRDTool (1999)Used a circular buffer (Round Robin Archive) with a fixed size, automatically downsampling and overwriting old data to maintain history.Scalable Era (Gen 2)OpenTSDB (2010)Built on HBase (Hadoop/BigTable), it introduced tags (key-value pairs) attached to metrics, solving the scale problem.Cloud-Native Era (Gen 3)InfluxDB (2013)Used a specialized LSM variant (TSM Engine) and advanced compression techniques, achieving a 12x reduction in storage requirements.Prometheus (2012/2015)Introduced the pull model (server scrapes metrics from applications) and the powerful query language PromQL.Solving High CardinalityAs microservices and ephemeral infrastructure rose, the High Cardinality Problem emerged: tagging metrics with unique IDs (like container_id or user_id) caused indexes to explode in size. Modern TSDBs address this by:* Columnar Storage: Systems like GreptimeDB and QuestDB shift away from inverted indexes, storing data by column to leverage vectorized execution (SIMD) for scanning billions of rows quickly.* Hybrid Partitioning: TimescaleDB (built on PostgreSQL) uses “hypertables” partitioned by time and secondarily by a spatial dimension (like device ID), constraining B-Tree index size.Act IV: The Pursuit of Narrative IntegrityThe quest for a perfect history extends beyond performance into the realm of semantic correctness and verifiable truth.* Temporal Databases: These systems track Bi-temporal data: Valid Time (when a fact was true in the real world) and Transaction Time (when the fact was recorded in the database). This is critical for regulated industries requiring auditable retroactive corrections.* Event Sourcing: This philosophy stores the entire stream of events—the immutable log—that led to the current state (e.g., Deposited($50) rather than simply updating the Balance). The current state is derived by replaying the log, providing perfect auditability.* Immutable Ledgers: Taking auditability one step further, ledgers like immudb ensure that even an administrator cannot tamper with history. They utilize Merkle Trees (Hash Trees) to generate a single “Root Hash”. If a single byte in a historical record changes, the root hash changes, providing cryptographic proof of unaltered history.Conclusion: The Polyglot FutureThe evolution of database systems has demonstrated a clear trajectory: from the destruction of history in mutable B-Trees to the preservation and cryptographic verification of the complete narrative.Today’s architects embrace polyglot persistence, realizing that different problems require specialized tools. An application might use a distributed relational database (NewSQL) for consistency, an LSM-based TSDB (like InfluxDB) for high-speed metrics ingestion, and a graph database (Neo4j) for relationships.Furthermore, the cloud era has shifted database management from on-premises hardware to Database-as-a-Service. Managed, serverless offerings like Amazon Aurora automatically scale compute and storage, promising that you only pay for what you use and never worry about provisioning.The database has evolved from a static file system into an elastic, living record of truth. The saga continues, driven by the persistent human desire to efficiently organize and trust the world’s information. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

In the vast history of human value, from ancient ledgers to global finance, few concepts are as fundamental, or as fragile, as trust. Our journey to store and transfer value has always been an experiment in trust architecture. We have moved from physical behemoths that anchored island communities to complex algorithms designed to secure global, anonymous markets. The thread linking these disparate systems reveals a profound truth: the basis of all money is a shared, unmovable block of trust.Let us trace this philosophical odyssey—from the giants of limestone on Yap to the ghost in the machine designed by the godfather of digital privacy.The Ledger of Giants: Yap’s Social BlockchainImagine wealth so immense it rarely moved. For centuries on the Micronesian island of Yap, value was held in enormous limestone disks called Rai stones (or fei), some weighing up to four tons and measuring 12 feet across. These were not coins slipped into a pocket. Instead, they were placed in public view—in front of meeting houses or along village paths—acting as physical markers for a communal account.The Yapese system was, in essence, a distributed ledger system. Ownership wasn’t established by possession, but by proclamation. When a trade occurred, the transfer was announced publicly, and the entire community—the collective memory of the people—would nod in assent, updating their mental or written ledgers.This sophisticated social technology solved several foundational problems that we associate with modern digital systems:* Distributed Consensus: No single authority controlled the record; disputes were settled by comparing the memory of the community and going with the majority record of truth.* Immutability: The stone’s history, preserved through oral tradition, made its past transactions tamper-proof by social contract.* Proof-of-Work (PoW): Value was accrued not just from size, but from the immense labor, risk, and sacrifice—including the deaths incurred—during the 400-kilometer quarrying voyages from Palau. This dangerous effort enforced scarcity, keeping inflation in check for centuries.The most famous example is the sunken Rai stone: one that was lost at sea but continued to be traded for over a century because the community agreed its value endured. The physical location of the stone didn’t matter, as its value resided entirely in shared agreement—a profound separation of the token from the ledger.Shadows in the Machine: David Chaum’s Cypherpunk VisionWhile the Yapese relied on intense social bonds and collective memory, the digital age required a solution for trust between strangers—and a defense against surveillance. Enter David Chaum, born in 1955, often dubbed the “godfather of digital currency”.Chaum’s vision, rooted in the burgeoning Cypherpunk movement, was not just about building code but creating a philosophical bulwark. He foresaw a “panopticon nightmare” where unchecked tracking turns citizens into suspects. His goal was to make cryptography the “invisible armor of the individual,” shielding identities and transactions.His odyssey in digital anonymity began in the early 1980s:* Mix Networks (1981): These “cascades of servers” shuffled messages like cards, obscuring senders and recipients, making communication a “ghost dance” against surveillance.* Blind Signatures (1982): Considered Chaum’s crown jewel, this allowed a trusted intermediary (like a bank) to “sign” a digital coin without viewing its origins or destination. This enabled value to flow anonymously yet remain verifiable, much like cash in a crowd.* Ecash (DigiCash, 1990s): This was the incarnated vision—digital money promising that a $10 digital bill, whether spent on pizza or protest, would vanish into the ether without revealing the user’s identity.Chaum sought security without identification. He abstracted secrecy, aiming to decouple economic action from identity, fostering markets “untethered from coercion”. However, his early systems, like DigiCash, leaned on trusted intermediaries (banks), introducing a semi-centralized scaffold that incumbents eventually managed to burn down due to “risk” fears (money laundering) and slow user adoption.Chaum’s vision ultimately served as the “ur-text” for the Cypherpunk movement. His work directly inspired systems like Tor and PGP. While Bitcoin emerged later, sacrificing full anonymity for transparency, Chaum’s push for absolute privacy continues today through efforts like the XX Network, which tackles metadata creep using quantum-resistant mixing techniques.From Vulnerability to AlgorithmThe transition from Yap’s social ledger to digital architecture was driven by the need to solve two fatal flaws inherent in any human-based consensus system, both of which destroyed the Rai system:* The O’Keefe Inflation Attack: In the late 19th century, Irish-American trader David O’Keefe mass-produced Rai stones using superior modern tools, bypassing the traditional, costly proof-of-work. This technological shock debased the currency because the new stones lacked the crucial narrative of sacrifice and traditional labor.* The German Ledger Attack: German colonial administrators, seeking to coerce the Yapese to build roads, used black paint to mark the most valuable stones as “seized”. This simple act, backed by a credible threat of violence, broke the social consensus, forcing the islanders to comply.Blockchain, starting with Bitcoin in 2008, offers an engineered solution to these ancient problems. It replaces social consensus with cryptographic consensus, placing trust in mathematics, economics, and code, rather than in fallible humans or vulnerable communities.* Solving O’Keefe (Dynamic PoW): Bitcoin’s Proof-of-Work uses a difficulty adjustment. If a modern “O’Keefe” joins the network with superior computing power (mining ships), the protocol automatically increases the complexity of the puzzle, ensuring that new blocks are created at a slow, predictable, and scarce rate.* Solving the German Problem (51% Attack): To “paint a black cross” (censor transactions) on the blockchain, an attacker would need to control over 50% of the entire global network’s computational power, making the attack economically irrational and mathematically prohibitive.The result is unforgeable digital scarcity. Bitcoin’s 21 million coin limit is guaranteed by protocol, not by legal enforcement, creating value through designed scarcity and consensus.The Persistence of Paradox: Trust Migrates, It Doesn’t VanishThe journey from stone blocks to cryptographic chains represents humanity trading social trust for mathematical proof. But this doesn’t eliminate trust; it merely redistributes it. We shift trust from central banks and feudal lords to protocol designers, software developers, and the economic incentives that govern the mining network.This transition brings Yap’s philosophical debate into the modern era:* Yap: A high-trust, identity-based system where reputation was paramount and community memory was the ledger.* Blockchain: A low-trust, pseudonymous system built for a global environment where participants do not need to know each other.Furthermore, the tension between “Code is Law” (absolute immutability) and “Social Consensus” (human judgment as the final arbiter) continues to rage in the crypto world. The DAO hack and subsequent hard fork of Ethereum proved that when technical rules yield unacceptable outcomes, the community changes the rules—a social decision dressed as a technical upgrade. As Vitalik Buterin acknowledged, social considerations ultimately protect any blockchain in the long term.Despite the ideals of decentralization, new forms of centralization inevitably emerge. Whether due to network effects, economies of scale, or simple human preference for convenience, power gravitates toward mining pool operators, core developers, and centralized exchanges.The Unending SearchThe arc from Yap’s unmoving stones to digital architecture is not purely progress; it is recursion. Both systems function because value is a collective idea, a social construction sustained by belief. The stone at the bottom of the ocean was real because the community agreed it was real. Bitcoin is real because we agree it is real.The blockchain is the latest attempt to build a shared, incorruptible memory. It offers protection against institutional betrayal by replacing human fallibility with algorithmic consistency. Yet, in fleeing the fragility of social consensus, we create rigidity. We gain consistency, but we lose the flexibility, mercy, and human judgment that defined the Yapese community’s ability to adapt.As algorithms audit our afternoons and CBDCs loom like digital dragnets, Chaum’s prophecy remains relevant: true sovereignty requires untraceable value. The question now isn’t whether technology can scale, but how we will balance privacy and transparency, freedom and security, rigidity and wisdom.The ledger watches. The choice remains: does your next transaction liberate, or merely log?The journey from heavy stone blocks to chains of light reveals that the greatest engineering challenge is not code, but cooperation. Every monetary system is simply a management protocol for the fundamental tension between individual autonomy and collective coordination. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

Welcome back to “Coordinated with Fredrik.” This week, we’re tackling an idea that forces us to question everything: the simulation hypothesis. It’s the mind-bending concept that our entire perceived world—the sky, our friends, and our existence—is not ultimate reality but an elaborate digital illusion, akin to a sophisticated computer program.This idea is far from new. It has deep philosophical roots, stretching back to Plato’s allegory of the cave, where prisoners mistook shadow projections for reality. Later, René Descartes refined this skepticism with the thought experiment of an evil demon manipulating every experience we have. The modern simulation hypothesis, however, represents a fundamental pivot from metaphysical doubt to a technological and probabilistic argument.The Statistical Modesty of Bostrom’s TrilemmaThe modern debate was formalized in 2003 by philosopher Nick Bostrom, who put forward a rigorous simulation argument framed as a trilemma. He argues that if technological progress continues, one of three “unlikely-seeming” propositions must be true:* The Great Filter: Almost no civilization will reach a posthuman stage capable of creating realistic simulations of conscious minds (perhaps because they destroy themselves).* The Great Disinterest: Advanced civilizations reach this technological stage but choose not to run “ancestor simulations” of people like us, possibly due to ethical scruples or lack of interest.* The Simulation: If the first two propositions are false (meaning advanced civilizations survive and run many simulations), then the number of simulated conscious beings would statistically far outnumber those in “base reality.” Therefore, we ourselves would almost certainly be among the simulated minds.Bostrom’s logic rests on two key assumptions: first, that consciousness can arise from computation (known as substrate-independence); and second, that future “posthuman” civilizations will command enormous amounts of computing power. If these premises hold, Bostrom claims that believing we are the special, original race among trillions of simulated ones is an act of statistical hubris.The Case for Code: Glitches and Digital PhysicsWhy do serious thinkers—like entrepreneur Elon Musk, who famously stated it’s “most likely that we’re in a simulation”, and astrophysicist Neil deGrasse Tyson, who puts the odds around “50-50”—find this plausible?Part of the momentum comes from the realization that physical reality behaves mathematically, suggesting the universe might fundamentally be information. This is the core of “digital physics”.Some peculiar observations resonate with the simulation notion:* Quantum Strangeness: Quantum mechanics features an “observer effect” where measurement affects a particle’s behavior. This is reminiscent of a video game that only “renders” an object in detail when a player is looking at it, potentially to save on resources. The simulation might use randomness and indeterminacy as a resource-saving shortcut.* Error-Correcting Codes: Theoretical physicist James Gates found unexpected mathematical structures within the equations of supersymmetry that were akin to error-correcting codes—the same kind that fix data errors in browsers. This bizarre finding suggests the fabric of reality might have computational aspects, as if the universe’s “software” has a built-in debugging feature.* A Discrete Universe: In a simulation, everything must ultimately be represented in finite bits on a grid. This aligns with the speculative idea that spacetime might be “pixelated” at the incredibly tiny Planck scale.The Refutation: Thermodynamics Defeats Ancestor SimsWhile the idea is intriguing, rigorous scientific analysis, particularly recent quantitative research, provides mounting evidence against the physical feasibility of a universe simulation operating under our known laws.Physicists and information theorists have applied fundamental physical principles (like the Bekenstein bound and Landauer’s principle) to calculate the resources required. The gap is staggering:* Computational Impossibility: Simulating quantum-level reality demands $10^{62}$ times more computing power than currently exists globally. Exact simulation of the universe’s $10^{80}$ particles requires tracking $2^{(10^{80})}$ states—a number larger than the information storage capacity of the universe itself.* Energy Requirements: Simulating the entire visible universe requires encoding $~10^{123}$ bits, demanding approximately $10^{94}$ joules—an amount that exceeds the total energy content of the observable universe.* Time Constraints: Even a dramatically reduced, low-resolution simulation of Earth would consume $10^{35}$ to $10^{65}$ years of computing time per simulated second, vastly exceeding the universe’s age.In short, these calculations show that simulating our universe using physics like ours is fundamentally prohibited by energy conservation and entropy constraints; it is not merely a technological challenge.Furthermore, attempts to detect “glitches” have so far yielded nothing. The most rigorous experimental proposal suggested searching for a computational lattice by looking for rotational symmetry breaking in the arrival direction of ultra-high-energy cosmic rays. After years of monitoring, no predicted anisotropy has been detected, constraining this specific model of a simulation.This has led prominent theoretical physicist Sabine Hossenfelder to label the simulation hypothesis “unscientific” and even pseudoscience, arguing that mixing science with metaphysical speculation about omnipotent simulators is a mistake.Life and Meaning in a Digital RealityIf we were to discover the truth—that we are simulated—what would it mean for our lives?Philosopher David Chalmers argues against the notion that life would become meaningless. He contends that a simulated world is not an “illusion” but a “digital reality”. Our struggles, relationships, and joys are authentic to us. As he states, even if we’re in a perfect simulation, this is not an illusion; “Everything is just as meaningful as it was before”.The hypothesis also transforms theological concepts. An intentional creation by intelligent “outsiders” is analogous to a Creator, though the architects of our reality might be fallible, perhaps even immature, beings—not omniscient deities. Physicist James Gates even mused that the simulation idea “opens the door to eternal life or resurrection,” as our consciousness, being digital code, might be backed up or run again.Finally, the simulation hypothesis creates a profound conundrum regarding free will. Are our choices mere algorithms, making us “puppets” predetermined by code? Or, is our consciousness a “player” existing in a higher reality, making autonomous choices that influence the simulation—perhaps even instructing the simulation to “render” reality, thus elevating free will to a causal agent in physics?The Unanswered QuestionThe simulation hypothesis remains one of the most compelling and unresolvable intellectual debates of the modern era. While science has proven that simulating our universe under our physics is physically impossible, the hypothesis can always retreat to the unfalsifiable loophole that the simulators operate under physics we cannot conceive.However, the value of the idea isn’t in providing an answer we can confirm, but in the questions it forces us to ask. It spurs scientific creativity, pushing researchers to devise experiments at the boundary of physics and information. Whether our cosmos is base reality or one of infinitely nested simulations, we still face the same fundamental task: to live our lives as authentically and fully as we can. For us, as Chalmers suggests, the virtual reality we inhabit is real enough.The simulation hypothesis is like looking at a highly complex clockwork mechanism and wondering if the clockmaker used a pre-built computer program to design the gears, only to realize the sheer size and complexity of the resulting clock requires the program itself to be bigger than the finished clock—leaving us to marvel at the ingenuity (or impossibility) of the original design. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

Welcome to “Coordinated with Fredrik.” We often talk about the strategies founders need to deploy to win funding, but today, we are turning the tables. We are deconstructing the individual Venture Capitalist—the person sitting across the table—to understand the economic “machine” they operate within. The core truth is this: a VC’s behavior, which often appears baffling or “irrational” to outsiders, is a direct, rational response to the structural and financial incentives that govern their existence.The VC’s 10-Year Clock and the Power Law TyrannyEvery decision a VC makes is governed by a defined fund lifespan, typically a closed-end investment vehicle of 7 to 12 years. This “10-year clock” is non-negotiable and fundamentally shifts a VC’s personality and risk profile depending on the fund’s current phase.1. The Investment Period (Years 1-4): This is the “hunting” season where the VC is optimistic and eager to deploy capital, actively “build[ing] the portfolio”. They can underwrite a 10-year vision for a company to mature.2. The Management & Harvest Period (Years 5-10): The focus shifts entirely to managing, supporting, and exiting existing investments to return capital to LPs. A founder pitching in Year 7 of the fund is meeting a “farmer” or “mortician” who cannot make a new investment requiring another decade to mature. They prioritize speed over strategy for their current portfolio.The financial structure is the notorious “2 and 20” model. The “2” is the management fee (typically 1% to 2.5% of committed capital), which serves as the firm’s “salary” to cover operational costs and salaries. The “20” (carried interest, or “Carry”) is the “real payday”—a 20% share of the fund’s profits that General Partners (GPs) keep. Critically, this carry is only paid after the fund returns 100% of the initial capital to its LPs.This compensation structure creates the “Tyranny of the Power Law”. Because VCs must return the entire fund before seeing a dollar of profit, they are mathematically forced to find “really. large. exits.”. A $50 million acquisition might thrill a founder, but for a VC managing a $100 million fund, that profit is a “drop in the bucket”. They need one or two companies to return $100 million, $500 million, or even $1 billion to “return the entire fund” by themselves. This forces VCs into an “all-or-nothing” or “growth-at-all-costs” mindset. A stable, profitable, $50 million-revenue business is often considered a “zombie” because it doesn’t fit the Power Law. VCs are incentivized to push founders to take massive, company-ending risks to become a $1 billion “unicorn”.The Partnership Hierarchy: From Analyst to GPThe VC firm is a rigid hierarchy, and where an individual sits on this ladder dictates their job and incentives.• Junior Staff (Analysts/Associates): These entry-level roles are paid primarily from the management fee (the “2”). Their job is a “50-60 hours per week” grind of “screening pitch decks,” “deal sourcing,” and meticulous CRM updates. The role is often an “up-or-out” 2-3 year position, functioning as high-energy, low-cost labor to “filter signal from noise” for partners. Most junior VCs never see carry.• Partners (GPs): General Partners receive the “lion’s share of the carry”. Their primary responsibilities shift entirely away from analysis to fundraising from LPs, making final investment decisions, and taking board seats. For these senior partners, their entire financial motivation is the large, leveraged bet on the 20% carry, which will only pay off 8 to 12 years in the future, if at all. They often make low base salaries relative to their potential carry, with their own “capital commit” (1-3% of the fund) showing LPs they have “skin in the game”.The “black box” of the partner meeting, where a founder’s fate is decided, reveals internal politics are intense. The decision often boils down to the “point partner” (the founder’s internal champion) successfully arguing against colleagues in a “Real Decision-Making Arena”. This internal political debate is crucial, as illustrated by the recent dynamics at Sequoia Capital, where strategic misalignment and cultural failures led to leadership change, emphasizing that founders are pitching a specific partner who must then win an internal debate.Sourcing, Bias, and the FOMO-FOLS DyadVCs are obsessed with “proprietary deal flow”—deals sourced directly that are “less competitive” and usually mean “better terms”. However, deal flow is overwhelmingly driven by “Relationships”, with around 60% of VC deals originating from an investor’s personal network or referrals. This network-first approach, while an “efficiency play”, is the “primary engine of systemic bias in the industry”. If a VC’s network is homogenous, their pre-vetted deal flow will be equally homogenous.When evaluating nascent early-stage companies where there are “no concrete financial metrics”, VCs rely on heuristics and pattern-matching. Team is consistently the most critical factor, often cited as 95% of the decision at the seed stage. With little data, evaluation defaults to “gut feelings”, seeking “representativeness” (does this founder look and sound like a previous winner?). This reliance on “gut feel” is often a euphemism for “similarity bias” and perpetuates exclusion, which is why companies founded solely by women receive only 2% of all VC investment.To mitigate this bias, influential VCs like Mark Suster advocate to “Invest in Lines, Not Dots”. A “dot” is a single pitch meeting driven by “limited thought, limited due diligence”. A “line” is a “pattern of progress” observed over time, where a VC meets the entrepreneur early and watches their performance over 15 meetings or two years. This approach replaces biased gut feeling with observed data on the founder’s “tenacity” and “resiliency”.The internal decision to invest is a tug-of-war between two primal psychological fears:1. FOMO (Fear of Missing Out): The VC’s “biggest fear,” driven by regret over passing on a company that became a “really. large. exit.”. This drives VCs to invest.2. FOLS (Fear of Looking Stupid): The “avoidance of scrutiny,” driven by the fear of LPs questioning a bad investment. This drives VCs to say no.A founder successfully raises capital when they manufacture more FOMO than FOLS, often by showing the VC that other investors are interested, triggering a “bandwagon” effect and herd mentality.The Rise of the Solo GPThe traditional VC firm model is being challenged by the rise of the Solo General Partner (Solo GP). These are single individuals who raise a fund and make unilateral investment decisions. Their advantages are speed and flexibility (”a motorcycle” that weaves through traffic while larger firms are “18-wheelers”). Many Solo GPs leverage personal brands and expertise—like Aarthi Ramamurthy (Schema Ventures) who uses her podcast as a sourcing engine, or Marc Cohen (Unbundled VC) who “build[s] in public”. This new guard is competing on transparency and personal brand rather than the Assets Under Management (AUM) and management fees of the old guard.The truth remains that the job of a VC is a “tough business”, requiring patience (carry takes a decade to realize), relentless networking, and the painful necessity of turning away 98 out of 100 opportunities.The Zombie Trap: Managing FailureThe founder-VC relationship, particularly after the check is wired and a VC takes a board seat, is where the Power Law is brutally enforced. High valuations create expectations for matching performance, leading to intense pressure to pursue “growth metrics over unit economics”. If a company doesn’t fail but also doesn’t “go big,” it enters the “VC Zombie Trap”.A “zombie startup” is profitable and growing, perhaps 20% a year, but will never provide the 100x return the VC needs. VCs are incentivized to kill their zombies because they cannot waste time, energy, or “remaining capital” on a company that won’t “return the fund”. This is a “feature and not a bug of the power-law driven venture capital system”.The candid reality, as noted by Fred Wilson, is that aggregate VC returns are often “disappointing,” sometimes underperforming the NASDAQ. VCs are forced to be optimistic hunters, yet their job is a massive, leveraged bet that only pays off, if at all, years down the line.--------------------------------------------------------------------------------The mechanisms VCs use to cut through the noise—reliance on networks, pattern-matching, and the psychological battle between FOMO and FOLS—are also the mechanisms that introduce bias and structural rigidities into the ecosystem. What does this mean for the future of capital deployment?We have detailed how the VC machine operates, but the conversation doesn’t stop here. If you want to dive deeper into the tactics founders use to navigate the partner meeting “black box,” the specific “unwritten rules” of the VC ecosystem (like the “frieNDA”), or how the new wave of Solo GPs is disrupting this status quo, let us know! This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

The material provided centers on Johan Norberg’s arguments presented in his book, The Capitalist Manifesto, which serves as a defense of global capitalism, detailing its role in human progress, countering popular criticisms, and arguing for economic freedom in the modern era.This week on “Coordinated with Fredrik,” we dive into The Capitalist Manifesto with author Johan Norberg, who argues compellingly that the free market remains the best engine for human progress, despite recent global shocks and widespread political hostility.Norberg, whom Swedish Public Radio once noted as possibly the only person “particularly keen on globalization now”, lays out the dramatic, yet often unheralded, successes of the past two decades. Since he wrote his first defense of global capitalism in 2001, extreme poverty has been reduced by 70 per cent—amounting to over 138,000 people rising out of poverty every single day. This staggering progress, which also includes drastic reductions in child mortality and increased global life expectancy, confirms that we need more capitalism, not less.Addressing New Opposition and Old MythsThe nature of the opposition has changed drastically, moving from the political left (like Attac in 2001) to a “new generation of conservative politicians” and right-wing populists who now sound very much like the earlier critics. The common narrative shared across political extremes is that global capitalism has primarily benefited the rich and China, while wages stagnated and jobs disappeared in the West.Norberg argues this worldview is based on misunderstanding the economy as a zero-sum game. In reality, the decline in factory jobs is mostly due to automation and increased productivity (taken by “R2-D2 and C-3PO”) rather than competition from China. Furthermore, economic success—from the simplest item like a cup of coffee to large-scale innovations—comes from the complex, voluntary cooperation enabled by free markets, utilizing the localized knowledge of millions of actors. It is competition and freedom of choice that force capitalists to behave well.Bailouts, Breakdowns, and the Battle for BeliefWe explore modern threats, from the idea that the “Reagan/Thatcher era is over” to the dangers of crisis management. The pandemic showed that protectionism is perilous; resilience is achieved through diversified supply chains and the decentralized ingenuity of entrepreneurs adapting quickly to new needs.Crucially, Norberg challenges the notion that capitalism destroys the planet or makes us miserable. The solution to environmental problems, including climate change, is not “degrowth” (which, during the 2020 pandemic experiment, cut emissions by only 6% while plunging millions into poverty) but rather the prosperity and technology generated by the market. Data suggests that economic freedom is positively correlated with environmental performance, and “talking about money” is the key to accelerating the transition to greener, more efficient methods.Want to understand why the greatest catastrophe to hit humanity would be stopping economic growth, or explore why central bank attempts to prevent all economic risk have led to the rise of “zombie companies”? Tune in to the full discussion to discover how economic freedom is crucial for our future and why the unpredictable nature of capitalism is far superior to centralized command. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

For those of you who caught the special episode of “Coordinated with Fredrik,” you know this one was personal. My wife Madeleine and I, both having backgrounds in academia and entrepreneurship, decided to speak directly to our children, Gustav (15) and Louise (12), about what truly matters: finding their unique purpose (their Life’s Task) and achieving lasting happiness.This wasn’t just typical parental advice; it was a distillation of psychological and historical principles, heavily informed by the wisdom of thinkers like Robert Greene, whose work on Mastery I rely on daily.Here are the key takeaways from the episode, which I hope serve as a lasting roadmap for them—and perhaps for you, too.1. Finding Their Phenomenal UniquenessThe core message we shared with Gustav and Louise is that they are, scientifically and historically speaking, a phenomenon. Their DNA is unique and has never occurred before, and their early life experience is entirely their own. This uniqueness is their primary source of power.We emphasized that their purpose, or their Life’s Task, is found by digging back into childhood to discover their primal inclinations—what the sources call “impulse voices” or “seed emotions”. These are the early signs of what they genuinely loved or hated.This quest isn’t just a hobby; it’s essential because when they connect emotionally to this inner “grain” or dominant form of intelligence, they gain the massive energy required for discipline. This emotional engagement means the brain learns two, three, or four times faster than when they are merely paying “half attention”. This deep connection serves as their internal radar or compass, cutting out distractions and providing an overall framework for experimentation.2. The Great Divide: Real Fulfillment vs. The Instant RushThe pursuit of happiness is often misdirected today. We explained the difference between the “real sublime” and the “false sublime”.The real sublime is about profound, sustained fulfillment that comes from genuine effort, connecting them to something larger than their own ego, such as deeply immersive creative work or deep reflection. This is the lasting feeling they should chase.The false sublime, in contrast, is the quick, addictive high. This is where social media and instant gratification come in. These external sources provide an illusion of transcendence but require “more and more and more” to maintain the rush. The problem, especially for teenagers, is that this digital noise constantly distracts them. We stressed that anxiety is a crucial signal. Instead of rushing for the quickest answer (often provided by technology like ChatGPT) to relieve that anxiety, they must learn to embrace the turmoil and think harder. This challenging process is what develops the “muscle in the brain” necessary for creative, “alive thinking”. Relying on quick answers, Greene warns, can lead to a generation who stops thinking deeply.3. Mastering the Social World: Friends and AwarenessAs they navigate the complexities of high school and young adulthood, they must understand that they live in an “invisible realm” of power dynamics where people often wear masks. Power is not about domination, but about having a degree of control over one’s environment.We urged them to develop radical self-awareness. They must pay attention to themselves—what they love, what they hate—rather than being attuned only to what others like.Crucially, we focused on nonverbal communication. Because humans evolved communicating nonverbally for vastly longer than with symbolic language, we are wired with an “amazing sensitivity” to cues. They must develop the practice of “muting” the words and watching people—their eyes, their feet, their posture. The fake smile is a prime danger sign, often revealing a lack of genuine interest or the presence of a toxic personality. If they are constantly immersed in the virtual realm, this vital social muscle atrophies.4. The Urgency of “Death Ground”Finally, we wanted them to internalize a sense of urgency. As a founder and an academic, I know that progress doesn’t happen when the pressure is off.Drawing on strategic concepts, we introduced the idea of “death ground”. This means creating a feeling of intense necessity, acting as if their backs are against the wall and they must win or die. This barometric pressure releases incredible, unexpected energy. The reality is, they cannot fool themselves into thinking they have infinite time. Robert Greene’s own recovery from a stroke serves as a powerful testament to how a moment of crisis can heighten the appreciation for life and transform tragedy into relentless determination and focus.We want Gustav and Louise to feel that urgency now to avoid wasting the unique seed of power they possess. The feeling of being alive is a “wondrous experience,” and they should not take it for granted.This episode was built on the premise that clarity—about self, about others, and about the forces vying for their attention—is the key to a happy, successful life. If you found these reflections valuable, I highly recommend tuning in to the full episode of “Coordinated with Fredrik” for a deep dive into these tools. Would you like to explore the specific tactics for mastering nonverbal communication or the detailed process of how to handle career anxiety using these principles? This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

Sweden, a high-tax welfare state with a population of only 10.5 million people, has managed to become a global startup superpower known as “Silicon Valhalla”, producing unicorns like Spotify, Klarna, Skype, and King. Stockholm ranks second globally for producing unicorns per capita, trailing only Silicon Valley, while attracting high levels of venture capital rivaling Israel. The nation achieves this by synthesizing its robust social safety net—which structurally de-risks entrepreneurship by providing universal healthcare and tuition-free education—with targeted, pro-capitalist incentives like low corporate taxes and favorable tax treatment for employee equity.The foundation of this success was laid in the late 1990s when the “Home PC Act” (Hem-PC) allowed companies to give employees tax-advantaged home computers. This scheme resulted in 850,000 computers being purchased, reaching nearly 25% of Sweden’s households and creating a digitally native generation. This was coupled with early investment in broadband and fiber-optic networks, positioning Sweden with 28 broadband subscriptions per 100 people by the mid-2000s, far ahead of the 17 per 100 in the US.This digital foundation, combined with a debt-free talent pipeline from free university education, fuels continuous innovation. Successful exits from first-generation unicorns (Spotify, Klarna) created “founder factories” that recycle capital and operational expertise back into the ecosystem. For instance, Klarna alumni have founded 62 startups.To ensure startups can compete for talent, Sweden introduced Qualified Employee Stock Options (QESOs) in 2018. This reform drastically reduced the effective tax on employee equity from approximately 81% (when taxed as salary) to capital gains rates of 25–30% upon sale. This policy surgically addressed the high-tax paradox, making high-risk labor tax-advantaged.Today, the ecosystem is pivoting towards the digital energy transition. Sweden’s nearly 98% fossil-free electricity grid and deep software talent position it to become a leader in cleantech, focusing on AI-optimized grids, V2G technology, and battery manufacturing.The synthesis of high societal security and high capital reward is fascinating. Ready to dive deeper into the mechanics of this paradox? Let’s explore how Sweden’s cultural concept of lagom (just the right amount) translates into flat corporate hierarchies and high labor productivity—achieving 28% higher efficiency than the OECD average—and how this consensus culture accelerates, rather than hinders, innovation. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

The core concept presented in Skin in the Game: Hidden Asymmetries in Daily Life by Nassim Nicholas Taleb is that true effectiveness, fairness, and accountability require a symmetry in human affairs. This means that if an individual stands to gain rewards from an action, they must also be exposed to corresponding risks. This principle is vital not just for justice, but fundamentally, for acquiring reliable knowledge and developing a robust filter against “bull***t”.Skin in the Game (SiTG) is defined as having a share of the harm, paying a penalty if something goes wrong, rather than simply having an incentive. The absence of this symmetry allows for the proliferation of hidden asymmetries, exemplified by “the Bob Rubin trade”. This describes a situation where agents, such as bankers, make steady profits by exploiting concealed risks, and when systemic blowups occur (like the 2008 crash), they “invoke uncertainty” while transferring the risk to taxpayers who pay for their losses. Without SiTG, systems cannot learn because individuals are not victims of their own mistakes; instead, “systems learn by removing parts, via negativa“.This filtering mechanism is crucial in evaluating expertise, distinguishing “theory and practice”. The author posits that the modern world is increasingly populated by the Intellectual Yet Idiot (IYI)—a class of people “better at explaining than understanding” and who lack accountability for their abstract, complex recommendations, such as those advocating failed regime changes. In contrast, rationality is rigorously defined as that which permits survival for the collective, not mere logical consistency in abstract models.Another powerful demonstration of asymmetry is the Minority Rule, where a small, intransigent minority that possesses “soul in the game” can dominate the preferences of the flexible majority. This dynamic explains phenomena from why nearly all liquids are kosher in the U.S. to the widespread preference for non-GMO foods: the flexible majority will accommodate the choices of the stubborn minority, provided the cost difference is not substantial.Ultimately, SiTG is about honor, justice, and existential commitment, underscoring the ancient notion that if you give an opinion that others follow, you are morally obligated to be exposed to its consequences.The material touches on complex, deep dives into probability, noting that concepts like ergodicity reveal why probabilistic estimates for a single individual over time diverge wildly from the average of a population. Would you like to delve into the mathematical logic of risk-taking and ruin, or perhaps explore the Lindy Effect to understand why time—not peer reviewers—is the ultimate arbiter of lasting ideas? This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

A podcast episode made on the transcript and insights from Huberman podcast. Here is the full episode: https://www.hubermanlab.com/episode/robert-greene-a-process-for-finding-achieving-your-unique-purpose—Welcome to the ultimate strategic blueprint for self-mastery, based on the insightful discussion between neuroscientist Andrew Huberman and renowned author Robert Greene. This episode, titled “A Process for Finding & Achieving Your Unique Purpose,” explores the intersection of self-exploration, human interaction, and history.Finding Your Unique Purpose: The Life’s TaskAccording to Robert Greene, author of best-selling books including The 48 Laws of Power and Mastery, finding your sense of purpose—what he calls your life’s task—concentrates your energy and gives everything a direction. Huberman considers Mastery a brilliant exploration and practical tool for thinking about and pursuing one’s purpose.The process of finding this purpose is not a mystery, but it is difficult and requires a specific archaeological process. Greene suggests reflecting on childhood “seed emotions” or “impulse voices”—those primal inclinations that manifest when you are four or five years old, such as a deep interest in words, abstract patterns, or animals.Your unique DNA and early life experiences make you a phenomenon and one of a kind, which is your ultimate source of power. Following this natural inclination, or the “grain in your brain,” is where your power lies. When you are emotionally engaged in a subject connected to this primal inclination, the brain learns at a much faster rate.The Pursuit of True Fulfillment: Real vs. False SublimeGreene is currently writing a book on the sublime, defining it as what lies just outside the limiting “circle” of societal conventions. The Real Sublime refers to profound, transcendental experiences, such as immersive creative work, that connect you to something larger than yourself and provide sustained fulfillment. Greene’s own stroke experience, which put him on the threshold of death, was the quintessential sublime experience.In contrast, the False Sublime involves fleeting, addictive highs from external sources like drugs, alcohol, shopping, or online rage. These provide a temporary sense of transcendence but are not lasting illusions, requiring more and more consumption. The connection to one’s true purpose, however, is a visceral, emotional, and physical feeling—like swimming with the current—that allows you to withstand moments of boredom because you feel the deep overall connection.Power, Anxiety, and the Urgency of “Death Ground”Greene defines power not as dominance, but as a deep, primal need to feel a degree of control over your immediate environment and influence others. Suppressing this innate desire for control only leads to passive-aggressive or covert means of seeking power. Learning the subtle dynamics of power is essential for a social animal, and much of The 48 Laws of Power is designed for defense against manipulation.The willingness to engage in effort and push through difficult creative processes often hinges on managing anxiety. Greene shares that his writing process is 95% pain, utilizing anxiety as a signal to keep refining and improving until the work reaches the necessary level of quality.This drive is tied to the strategic concept of “Death Ground,” a strategy inspired by Sun Tzu, where necessity presses in and your back is against the wall. This pressure—the urgency of realizing you could die or lose everything tomorrow—unlocks energy and focus that you normally lack. Greene’s stroke in 2018 served as his own “death ground” moment, which, while challenging, instilled a profound appreciation for life’s urgency and spurred neuroplasticity—the brain’s ability to rewire and adapt through effort.Navigating Modern Relationships and CommunicationWhen seeking long-term romantic relationships, Greene stresses that a convergence of deep, unchanging character values is critical. He emphasizes paying attention to what signals deeper character, such as a mutual love for animals, shared approach to money, or sense of humor, as opposed to superficial interests or mere admiration.The key to navigating all human interactions lies in mastering nonverbal communication. Since humans evolved without symbolic language for a vast period, we are hardwired to have an amazing sensitivity to nonverbal signals like posture, tone of voice, and micro-expressions. Greene warns that relying on virtual interactions, social media, and AI-driven platforms causes the crucial muscle of social skills to atrophy, leading to superficial relationships and diminished empathy. Mastering nonverbal cues, such as distinguishing a genuine smile (which lights up the whole face, including the eyes) from a fake one, helps one avoid toxic, deceptive individuals.The discussion also touched on the current crisis in masculine and feminine ideals, where confusing societal signals and the erosion of positive role models leave young people feeling lost. Greene advocates for emphasizing positive traits—such as inner resilience, quiet calm, and confidence in masculinity—to counteract negative stereotypes. The key, in all aspects of life, is developing self-awareness—listening to those inner voices of frustration or delight that guide you toward your authentic path.If you’re seeking to discover the unique seed of purpose within you, develop true influence, and navigate life’s deepest challenges with resilience, the wisdom shared by Robert Greene offers an indispensable roadmap. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com

Today, we dive into a story that has defined data science for forty years: the Forecasting Paradox. It’s the counterintuitive empirical observation that complex, sophisticated models consistently failed to beat simpler, more robust statistical methods when trying to predict the future. We track this paradox from its canonical proof through the massive M-Competitions, revealing how the debate over complexity vs. accuracy was settled—and then immediately overturned—based entirely on the structure of the data.Part I: The Sad Conclusion: When Simplicity Became LawFor decades, the intuitive assumption was that a complex model, capable of capturing intricate patterns, should inherently be a better predictor. However, the empirical evidence gathered by pioneers like Spyros Makridakis through the M-Competitions proved the opposite.The M3-Competition (published in 2000) provided the foundational, canonical proof of the Forecasting Paradox. Testing 24 methods across 3,003 diverse real-world time series, the results delivered what was called the “sad conclusion”: statistically sophisticated and complex methods, including early Artificial Neural Network (ANN) models, were consistently outperformed by simple approaches.The ultimate victor of M3 was the relatively obscure Theta method. This decompositional approach, which broke the series into simple trend and curvature lines, cemented the “simplicity mantra” as the dominant, evidence-based paradigm for nearly twenty years. This finding was later demystified by Rob J. Hyndman and Billah, who proved the Theta method was mathematically equivalent to Simple Exponential Smoothing (SES) with drift—a clever parameterization of one of the simplest methods available.Why Simplicity Won: The Trap of OverfittingThe paradox held because of a crucial distinction: “goodness-of-fit” versus “forecasting accuracy”. Complex models, with many parameters, would inevitably fall into the central sin of overfitting. They learned not just the underlying true structure (the signal), but also the random historical fluctuations (the noise), mistaking error for a repeatable pattern. When forecasting, the model projected this learned noise into the future, making erratic and inaccurate predictions.Simple models like Exponential Smoothing, conversely, are mathematically incapable of modeling complex noise; they act as robust “noise filters”. A 2018 PLOS ONE paper reaffirmed this, demonstrating that modern Machine Learning methods were “dominated” by traditional statistical benchmarks when tested on M3 data, often achieving better in-sample fit but worse out-of-sample accuracy.Part II: The Hybrid Breakthrough (M4)The M4-Competition (2018), instigated by Makridakis to rigorously test modern ML, represented a massive escalation, using 100,000 independent time series.The M4 results confirmed the persistence of the paradox: the six pure Machine Learning methods submitted all performed poorly, failing to beat the simple statistical “Comb” benchmark (a simple average of three statistical methods).However, the competition was won by Slawek Smyl’s novel Exponential Smoothing-Recurrent Neural Network (ES-RNN) hybrid model. This marked a pivotal moment, signaling the end of the “forecasting winter” where simple models reigned. The ES-RNN’s success proved that the path forward was not simplicity or complexity, but a synthesis of the two. The model used classical statistical Exponential Smoothing to pre-process the data (handling seasonality and level) and present a cleaner, more stationary residual series to the Recurrent Neural Network for complex modeling.This refined the historical understanding: pure complex models still failed due to overfitting the noise, but robust, noise-filtered hybrid models could finally surpass the simple benchmarks.Part III: The Overturn: Complexity Wins in the Global Era (M5)Just two years later, the M5-Competition (2020) provided a complete and shocking reversal. The dataset was fundamentally different: 42,840 hierarchical, interrelated sales series from Walmart, critically including explanatory variables like prices and promotions for the first time.In this environment, the paradox was definitively broken. Pure Machine Learning methods dominated the leaderboard, with the winning entry being an ensemble of LightGBM (Light Gradient Boosting Machine) models. These pure ML methods significantly outperformed all statistical benchmarks and their combinations.The Triumph of the Global ModelThe shift was enabled by a change in paradigm from “local” to “global“ modeling.* Local Models (The Old Way): Traditional statistical methods treated each of the 42,840 series independently, building a separate, “blind” model for each.* Global Models (The New Way): The M5 winners trained a single, massive LightGBM model on all 42,840 series simultaneously.This allowed for cross-learning, where the complex model learned global patterns (e.g., the general effect of a promotion or price elasticity) and generalized that knowledge across different products and stores. The M5 problem was high-signal and multivariate, meaning the simple, local statistical models were too “dumb” to utilize the rich external features and cross-sectional information, allowing the complex Global ML models to dominate.The Takeaway: A Context-Dependent FrameworkThe 40-year debate is resolved not as a binary choice, but as a context-dependent framework:* Paradox is TRUE for Local, Univariate Data: If you are forecasting a single, isolated time series with limited features (like the M3/M4 context), simple statistical methods or robust statistical-ML hybrids (like ES-RNN) are the state-of-the-art choice. Pure ML will likely overfit and fail.* Paradox is FALSE for Global, Multivariate Data: If you have thousands of related series and rich external features (like the M5 context), a Global ML model (like LightGBM) is the essential tool for utilizing all available signals.Coordinated with Fredrik Pro-Tip: The Metric WarsThe evolution of the M-Competitions was not just about the models, but the rules of engagement. For decades, the comparison was built on metrics like the Mean Absolute Percentage Error (MAPE) and sMAPE. These metrics, however, are now understood to be flawed—especially because they explode or become undefined near zero values and tend to be asymmetric, disproportionately punishing complex, volatile models. The M1-M3 paradox’s magnitude was likely an artifact of this flawed metric. Rob J. Hyndman solved this by introducing the Mean Absolute Scaled Error (MASE), a scale-free, symmetric, and robust metric. Makridakis adopted MASE for M4 and M5, ensuring a fairer evaluation and creating the necessary level playing field for ML models to be accurately tested and eventually triumph in M5. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit frahlg.substack.com