Skip to content
488: The Playful Portland Programming Paradigm episode artwork

Jan 6, 2026 30 min

488: The Playful Portland Programming Paradigm

from The Bike Shed Hosted by thoughtbot

Aji and Joël talk all things hackerthons and why taking the time to have a little having fun with your work every now and then is important for developers. Our hosts reflect on their shared experience at a previous RailsConf where they entered a mini hackerthon, why playing around in a creative sandbox is crucial to honing your skills as a dev, and how programming on the fly can teach you a lot about your own strengths and weaknesses. — Read more about Joël’s entry in Rails Rumble, as well as their participation in the RailsConf 2022 mini Hackerthon. Thanks to our sponsors for this episode Judoscale - Autoscale the Right Way (check the link for your free gift!), and Scout Monitoring. Your hosts for this episode have been thoughtbot’s own Joël Quenneville and Aji Slater. If you would like to support the show, head over to our GitHub page, or check out our website. Got a question or comment about the show? Why not write to our hosts: [email protected] This has been a thoughtbot podcast. Stay up to date by following us on social media - YouTube - LinkedIn - Mastodon - BlueSky © 2025 thoughtbot, inc.

Episode details come from the podcast's RSS feed.

Embed this episode

Ready to play

488: The Playful Portland Programming Paradigm

0:00 30:47
of matches

Transcript Auto-generated, may contain errors

Autoscaling is a simple concept. Automatically add resources when needed, and automatically shut them down to avoid paying for excess capacity. How hard could that be? If you've set up Autoscaling yourself, you know it's not that easy, especially using the native Autoscalers on platforms like AWS and Heroku. That's why you need JudoScale.

JudoScale is Autoscaling as a service, and they make Autoscaling simple and easy, as it should be. You can use JudoScale on AWS, Heroku, Render, Fly.io, and more. It's free for low traffic apps, and unlimited plans start at $25 per month. Autoscale on Easy Mode at Judoscale.com. Hello, and welcome to another episode of The Bike Shed, a weekly podcast from your friends at Thoughtbot about developing great software.

I'm Joel Kinville. And I'm Adi Slater, and together we're here to share a bit of what we've learned a long way. So, Adi, what's new in your world? Yeah, I don't know if this is really new in my world, but it's come up again, and I've really been thinking about it a lot. There were a couple of articles, I guess, recently that didn't have a very flattering opinion of Ruby, I guess, and asked if it was a serious language.

And besides that being just a very subjective and hard-to-define question, it brought up one of the moments in Ruby and Ruby on Rails history that I think has been played to death. It's just been brought up every single time, and I think I have an opposite take to what the usual consensus is. Can I guess? Yeah, go for it. Is this the Twitter switch to Scala?

Yes, it is. Of course it is. And I've been hearing this as an argument against Rails for so long, and I think it's time to finally put this to rest. It is, first of all, not a great argument anyway. Second of all, often it's kind of tied up in a lot of the disingenuous marketing or kind of blame shifting of the Twitter leadership at the time, right?

But the fact that Ruby on Rails was too slow for Twitter in that time is not really indicative of how it's going to work for most products. Twitter was the Internet. It was a size that any product and business can only dream of reaching. So Ruby on Rails getting Twitter to that size is a success story. It's not a knock against.

They decided to move off of it, but it got them to being the Internet's, what are they used to call it? Like, town square or whatever, right? Yeah, another front page of the Internet. That's what it's spoken. So this really sticks in my craw every time it gets brought up again, because not only is it totally not true anymore, Ruby has gotten so much faster over the years.

Rails has gotten so much faster over the years. There's been so much work put into performance since then. And of course, machines that they run on are just so much faster too, that it's not really a concern in the way that it maybe might have been in the very beginning. But it's also not a bad thing that happened. It's, yeah, I just keep thinking about it's a success.

If your product gets to Twitter size, wouldn't you be happy to have that problem? I think the sort of subtle argument there is that there are other benefits to Ruby and Rails aside from speed that will help you maybe hit that escape velocity in a way that alternatives don't, such that then you get to worry about your scaling as a problem. So you can iterate faster, you can more easily develop features and speed doesn't matter when you're at, I guess not speed, but at that point I guess scalability doesn't matter when you're just trying to find your fit in the market. And so that speed of iteration is the most important thing and framework like Rails that's more batteries included, more developer friendly is probably giving you that competitive advantage.

And then once you've owned an entire section of the internet, you can invest in doing something custom. Absolutely. And I always reach for that strength of Rails as the first reason as why I would recommend it to a new product team starting up, or even just a new vertical in a business or something like that. It's a little bit like our recommendation typically to use something like Roku for a startup. Yeah.

It's not going to give you the best performance, the cheapest servers, all of that. You're paying a premium and you have limitations you have to work with it. But it means that you don't have to worry about infrastructure. There's a lot of things you don't have to think about and it allows you to move fast. Which is what you want.

And once you're super established and you're trying to scale up, then it might make sense to say, all right, let's make an investment and move over to AWS. That investment might be more expensive to do three, five years from now. But it's also worth it because you get to only do it if you're successful. There's a contingency there. And you get to pay that cost only on success.

Yeah, 100%. I think the counter to that argument would be if there is another language or framework that claims to have just as much speed of iteration, developer friendliness as Rails, but also claims to scale better. Right. So, why not have your cake needed to? Absolutely.

I do like that you said claim in there. If another framework claims to be able to do both of those, obviously I don't have any cyclopaedic knowledge of all web frameworks that are out there. But I haven't come across another one that has the 20 year proven history in a way that Rails does. But, you know, this is thought about maybe we're a little biased, but I think we're biased for a good reason. But that's what's going on in my head.

What's new in your world? So, as was mentioned in the previous episode with Sally, we're recording a few episodes during about the Raffa Palooza, which is our end of the year hackathon, two days, step away from client work, get together, form small teams and just do interesting work. Cool things we can convince other people to do. And, of course, this is a year where there's a lot of interest in AI related things. And so, a lot of teams are working on AI stuff.

And I think in 2025, when we talk about AI, we immediately think about MLMs. But AI is a much broader field than that. So, there's an interesting conversation that happened this morning. There's a group that we're trying to write an AI to basically an AI opponent to play Tic-Tac-To. And they're not using an LM, they're doing old school classic AI techniques.

And one technique that got brought up in conversation was the minimax algorithm, which I find is a really fascinating way to think about things. Yeah, I think that's the way when I have implemented Tic-Tac-To opponent in a terminal version or something like that. That's the algorithm that I use. And that's the one that you reach forward with Tic-Tac-To. Yeah, I think it is the classic solution to Tic-Tac-To.

And also, Tic-Tac-To is the classic problem to demonstrate. When somebody wants to talk about the minimax algorithm, they typically show Tic-Tac-To because it's a smaller bounded game. There's an interesting article that went around when we're talking about some of this stuff. I'll link it in the show notes. But yeah, I think it's useful to think about AI beyond just what LMs can do.

There's a whole field out there, some really interesting algorithms. If you're excited about AI and want to do some things, yes, the current hype and the cutting edge is in LMs. But also just have fun with some of these other ones. So maybe check out that article and dig into that. Absolutely.

I think it's done a little bit of a disservice to the field that AI has become so synonymous with LMs in the general public point of view. Because like you're saying, there are so many other approaches and algorithms and things that have already been solved or improved by AI that people might just now think, oh, an LLM did that. And it's like, no, it was a specific new way of learning that was developed for this one problem and just so many different pathways to success in that kind of field that it kind of limited in the way that the general population perceives them. Have you ever tried to debug an issue in production while switching between your air tracker, your old viewer and APM tool?

Maybe you know how painful it is but you can see the error but can't find the relevant logs, or when you see slow performance, but it can't connect to the root cause. Scout monitoring is an all-in-one platform built to make your life easier, with log management, performance tracking and error monitoring coming soon. All of your data will be together at last. You'll go from bug to fix in one place. The result?

Less context switching, faster debugging and monitoring that actually helps you ship code instead of distracting from it. No DevOps team? No problem. Scout works with your existing stack and gives you insights in five minutes, not hours. And bug to fix in one place at ScoutAPM.com.

On the topic of fun and games and trying to get clever with your algorithm, I'm remembering a time where you and I got into a game or an AI competition, RailsConf. That's right. Which Rails Conf isn't this Portland? I think it was Portland. It was the first one back after virtual RailsConfs.

I think it was one of the vendors that was putting this on and I forget which vendor it was but this was really cool. The idea was that your game board was effectively the memory of a computer and you got to write a program in their sort of pseudo assembly. So you got to, you know, they had a, here is like a set of five instruction types that you get to call and then you get to write a program that is going to be a sequence of maybe up to 10 instructions. You were limited. You and your opponent each write a program and your program will be dropped at a random point in memory.

And then at the beginning of your program, you each have an execute cursor that will just sort of walk through your program or just actually it doesn't walk through your program. It just sequentially jumps to the next step in memory and executes the next instruction, keeps going. First program to crash loses. Now the instruction set a few interesting quirks to them as you could do with a regular assembly. You don't have to execute your instruction sequentially.

You can have jump instructions that move your execution cursor somewhere else. But there's also a crash instruction which you know, put that crash instruction executed you automatically lose. So you don't want that in your program. But an easy way to get a win condition was to somehow put the crash instruction in a part of memory that you thought that your opponent cursor might be in. And if they ever executed that instruction, then they would lose.

Yeah. There were a couple of other instructions that sort of helped you determine where you wanted to place that crash instruction. You could see, I forget if it was look at a specific point and tell if your opponent is before that in the block of memory or after that, there were some very coarse-grained tools at your disposal to try and locate or read the environment a little bit. But to do with any kind of certainty, you would have to iterate on your data that you got back and your feedback from those kind of questions that you were asking about the memory. And the longer you wanted to iterate over and have some algorithm, the longer your cursor was just happily rolling through memory, waiting for the opponent to insert a crash statement right in front of you.

Yeah, it was a real fun challenge. Yeah, because initially I think we did some really naive stuff. And then we realized we could really clever and effectively add a rudimentary binary search. Yeah. And then we started splitting the memory space in half progressively to get a sense of where the opponent was.

And I don't know if we did it. Maybe we only did it once, maybe we did it twice. And then we just started like blanketing that zone of memory with crash instructions. And the tricky thing was that, again, you only have so many, you have max size on your program. And when all you have is a jump instruction and some really rudimentary conditional, it takes a lot of instructions to create a binary search.

It really does, yeah. And I think we were trying to like take it like one level deeper or maybe we're trying to like a quadtree approach or something. And the program size is balloon so much we had to give up on it. Yeah, yeah, absolutely. The thing that was so like mind twisting for me was the fact that your instructions and your program was kind of spatially bound.

And you had to think of the playing field, this series of addresses and memory as not only where your program was going, but also sort of the place on the board that you were, and how those two things interrelated and how much of the board were you traversing? Because every step further is another opportunity to lose. Yeah, it was really cool because it would display this visually as a grid. And you could see, some of what they would do is you would upload your little program and then they would load it and another random program that you would face off against. And they would drop you into random places on this grid of memory.

And then you could actually see the head sort of executing jumping around. You could see where the crash instructions were placed on the board. And as the programs were searching, it was actually really well all done. Yeah. And they just were running all of these on a giant TV screen at their booth to randomly so people could upload programs when they wanted.

But then you could just all day just watch them play against each other and see them go up and down the rankings. The best thing was, is once you uploaded, you could still edit your program. So you could see how it worked, how it failed, what succeeded, and then make changes and re-upload them and see how that was fairing against the other opponents after making change. And they would make a change and you would make a change. And it was really kind of evolutionary in a way.

Yeah. And I think one thing that we noticed was a little bit what you were saying was that the longer programs added sort of more cycles before we could start placing crash instructions, which then left us vulnerable to other people, like as we're doing a thing that requires 10 instructions to just search the space and we get really precise. But that's a lot of cycles where you're not placing crash instructions and where you might get crashed yourself. And I think that's similar to when you are doing some really low level code of keeping track of how many cycles it takes to do certain operations and reducing the number of cycles through clever algorithm design or clever shortcuts or sometimes through hardware is a key to making things faster.

My understanding is that the, some of the older, I'm talking really old CPU chips are the, I guess, ALUs technically had hardware circuits for addition, but not for multiplication. Right. And so you would have to in software loop addition to get a multiplication. But that's a lot of cycles instead of just sending things through a circuit. And so all of a sudden you're not going to get a lot faster if you added a dedicated multiplication circuit to your, your ALU.

And as I understand that's kind of where the big difference between like a CPU and a GPU stems, right? They have different dedicated circuits for specific kinds of computation. Yeah. I think a lot of like linear algebra kind of matrix transformation things that sort of, you have a chip that's all circuits dedicated to those, that kind of mathematical operation. Right.

So one thing that you mentioned, I think also speaks to computer architecture. That's really interesting is the idea that, you know, we designed this program, it's in memory, it's moving around, it also gets to modify the state of its own board. And that might sound like an actual sort of, oh, it's a game decision, whatever, but that's, that's how real computers work. And that's kind of a, I don't know, it's probably maybe a genius decision and then has its own issues as well. I think they call this the von Neumann architecture, but the idea that the program lives in its own memory and that it can be modified.

That's really the leap that brought computing to general computing, right? Instead of purpose built machines that would run through a single algorithm because they were hardwired into that. But once they could edit or be reprogrammed by a user is when they became that kind of new class of machine, the things that we're working with all the time nowadays. But you could conceivably have a separate memory set for the instructions and for the sort of general purpose memory. And so you say, there's a set of instructions hard coded and then you have regular memory that you can read and write from.

But what you actually have in most computers is, and this game is a single set of memory that includes both the instructions and the other data that you work with. And you can read and write from that memory. You can edit your own instruction. A program can edit itself sort of assembly level metaprogramming. That's terrifying.

But that's what we were doing in this game, right? It was a assembly level metaprogramming and sometimes it's metaprogramming your opponents program to crash. Right, right. Yeah, it was a very different approach. And I think we first started out trying to do our programs the way that we would think of programs in Ruby or something.

We were using the tools that we had at hand. And it took a few iterations before we were kind of thinking in the right way that this play field, that this sandbox required. I do remember not to tune our own horns, but we won the competition on the first day of Rails Comp there. And while that was pretty great, that felt awesome and it was a lot of fun, that did mean that we were disqualified from competing in day two. And we had so many more ideas that we wanted to try out after seeing all the different algorithms and talking to the other competitors and of course thinking about it over the night.

But we were barred from reentry. But I also feel like that was one of the one of the cooler things I've ever done in a vendor hall. Oh, absolutely. Absolutely. It got and it was I don't know if it was great for the business.

I'm sure it must have worked out somehow, but it got people crowded around their booth all day long and interacting with them all day long. I'm going to try and find out who that was. But 2022, that is lost to the midst of time. It was a long time ago. Do you feel like the experience of working through that taught you something about software computing?

Yeah, I think it gave me a bit of an understanding for lower level systems and things that I had never really thought about. I didn't get a computer science degree, so that's that kind of theory and working at that level is not really something that I do. I work at the web application level and get lower as I need to from there. Was this your first time doing something that was like assembly-esque? The closest I think you could say was I had done like basic.

Okay. What, 30 years ago now? And so that had like line numbers and go-tos, but yeah, really assembly language level. This was the very first time I had encountered it. And while I don't know if there's anything specific that I applied forward, I think it was a great exercise in that it exposed me something new.

It made me try to solve a problem in a way that I hadn't done before. And it was just so engaging and fun that it was one of the things that brought an energy back to work. And that's what I love that conferences do so often. But this one really was just playing in the world where we work and showing that all of this can be fun. I think the other aspect that made this especially fun was that this was not like a solo project.

And I paired on this together and this sort of like bouncing ideas off each other. Oh, what if we did this? And what if we took a binary search approach? What if we can we take this like assumption and be like, yeah, it's going to be wrong some of the time, but as long as it's true more than half the time, it gives us an edge or yeah. It's one of those like magical like you hope this is going to happen at a conference if you like take the hallway track, you're finding someone cool, you're like someone's going to pull out a laptop and you have this amazing pairing session and like sometimes it's going to be a little bit of a fantasy like it's not a thing that happens all the time, but it was absolutely real.

Yeah. I remember we kept looking at the schedule being like, oh, there's that I really want to go to that talk. I think we can eat out a little bit more performance and we stayed in the vendor room pretty much all day. Well, the something is once we started realizing we were actually on the leaderboard, we're like, oh, we're like in like third position, we could just accept this and go to the next talk or we have some ideas and we think we could probably make it up to two or one. Right.

And we did. We did. It was great. Yeah. I was a little surprised when the folks running the booth said that it wasn't very often or maybe we were the first that they had seen attempt this as a pair that it was it was not how the game was usually played.

I was really surprised by that because pairing is such a I don't know, like a table stakes kind of a first order way to work that it just wasn't even a question. It's like, you want to pair on this? Great. Yeah, let's do it to find out that it was so rare to have a team competing, but they let us do it. Yeah.

I mean, we had so much more fun for doing it that way. Oh, 100%. I feel like this is the sort of thing that often maybe is what a hackathon would like to create in terms of experience. Yeah. Have you ever been to another sort of hackathon type event that you felt gave you something similar and it might not have been a competitive game.

Maybe it's something more collaborative or just do something. But have you ever been in a hackathon that you felt like captured some of the same feelings? Yeah, I have actually. There is I don't know if they're still running this, but for a long time there was a hackathon at the opening of the Major League Baseball season. What?

Yeah, the near to opening day, the MLB would make its API packages free for anyone that was competing in this hackathon. And you would come and basically pitch an idea of a project to build with all of this baseball data and find teammates and work on a thing. And you had like six to eight hours. And at the end of the day, you show everybody what you've got. And I think the combination of something that everybody there was already really excited about and passionate about being baseball sort of lent a little bit of energy into the projects as well.

Have you participated in one of these? I did. Yeah. I think what my team did was something to the effect of comparing the amenities at different ballparks so that you could say like, I want this and this and this and it would give you a rank of which ballparks are going to be best for what you're looking for in the ballpark experience. And it was mostly coded in R. So I didn't really contribute too much to that aspect of it.

I remember it ran in a browser and so I worked trying to like get through the JavaScript and manipulate it from the outside in order to make some such feature work. Yeah, that was a lot of fun. Have you had a similar hackathon experience? I participated way back in the day in I think it was called the Rails Rumble. It was a I think it was like a weekend remote hackathon thing where you would get a team and you'd build a Rails app.

Some cool product or whatever and put it up for review at the end of the 48 hour period and there was some kind of vote and you might win some prizes. It's fun. Do you remember what any projects that you did for it? The project that I did was a thing where you could upload photos and it would extract the main colors out of them and give you a color palette based off your photo and maybe would generate a website theme for you or something. That's exactly what it did after that.

It was pretty fun. That's nifty because that's also I imagine playing in a world that maybe your typical Rails app doesn't do. It's not a crud application. You're doing some image analysis or whatever aspect of it was. Yeah, there's definitely some crud involved but there's some other stuff as well.

That made it pretty fun. You're having a constant because it is a hackathon. But you want it to look really pretty because it's going to get judged. So you're sort of always balancing that ship fast versus how much polish to put on things. You're also really having a balanced scope because what can you get done?

Absolutely. It's really really important. So it's all the things you have to do as an early stage startup but like compress in two days. Yeah, that's a really good exercise for learning how to dramatically reduce scope but still accomplish something meaningful. I think is something that comes out of that kind of pressure cooker situation.

I was thinking maybe I wouldn't do this but it wouldn't be me if I didn't mention a Rails comp keynote. Of course. The closing keynote of this last Rails comp, Aaron Patterson made an impassion plea to Ruby guests, to Rails developers to not lose the fun and the joy in programming and encouraged us to do stupid things. Like to do things that had no meaning. Just for fun, just for exploration.

He said that he believed that that aspect is something that can and has led to innovation in the field because you encountered a hiccup or something while you were doing this other thing and it turned into a gem that turned out to be really useful for a lot of people. That kind of thing. I think as sort of the clown prince of the Ruby world at the moment, he is the perfect person to make that call for silliness. I think what I'm hearing is that maybe it's a good thing if Ruby is not a serious language. 100% I agree.

On that note, shall we wrap up? I don't know how we could do any better.

Frequently asked questions

How long is this episode of The Bike Shed?

This episode is 30 minutes long.

When was this The Bike Shed episode published?

This episode was published on January 6, 2026.

Is there a transcript available for this episode?

Yes, a full transcript is available for this episode. You can read the complete transcript on the episode page.

Can I download this The Bike Shed episode?

Yes. Use the download control on the episode player to save the publisher-provided media file.