Skip to content
454: Workshop design with Aji Slater episode artwork

Feb 4, 2025 37 min

454: Workshop design with Aji Slater

from The Bike Shed Hosted by thoughtbot

Joël is joined by fellow thoughtboter Aji Slater as they discuss their previous experiences in designing content for workshops. Learn how to best structure your workshop for an audience, the benefits of a workshop over a talk and vice versa, as well as how to tackle the different hurdles your audience might face when working through your presentation. — Try your hand at Joël’s recommendation of visualising your Git Branching. You can watch Ali’s Enigma Machine workshop here, Or connect with them via LinkedIn Your host for this episode has been Joël Quenneville. 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 - Instagram © 2025 thoughtbot, inc.

Episode details come from the podcast's RSS feed.

Embed this episode

Ready to play

454: Workshop design with Aji Slater

0:00 37:46
of matches

Transcript Auto-generated, may contain errors

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 Kenville, and today I'm joined by fellow ThoughtBotter, Aji Slater. Howdy. And together, we're here to share a little bit of what we've learned along the way. So, Aji, what's new in your world? In my world, I've been getting to play with a concept that was new to me upon joining the client project that I'm with right now. That's one of the things that I really love about consulting is that not only do you get introduced to different businesses and different teams, but you might also get assigned to a project with a new kind of tech or concept that you haven't been introduced to. And this one in particular is to revamp our client's access control system. They've gotten a little bit bigger. They've gotten more sophisticated in their needs. Their hand-rolled system that they started out with is not.

working out for them anymore. And so they're moving on to a relational-based access control system. And that's actually something that I had not really encountered. I've set up role-based access control plenty of times, but the relationship idea was new to me. You familiar with that? No. Explain to me what relational access control is and how that differs from more conventional role-based access. Yeah. So the relational access control, the underlying data structure is a directed graph. So you'll have different entities, resources, and subjects in this particular case.

that have a relationship to one another. So the entities are the nodes in the graph and the relationship being the edge. This way you can connect between two points in that graph, say it's a user and like a post or something that a user might own. If there's any kind of connection along that graph between those two points, then that is a allowed permission. And so it's kind of an easy way to conceptualize what can get pretty complicated if you're going to have different access control needs on particular individual items in a system.

So in this case, basically, my user record just gets a connection to every entity in the system that I have access to. Yeah, there's a couple of different ways to handle it. Because it's a directed graph, you can have a relationship to something like a group or an organization or even a role. You can build role-based access control on top of these systems so that if you have a connection to a role or an organization and then from the other direction, entities have a connection to that organization saying anyone that is related to you can also edit or whatever with me. then you have both fine-grained control of the user directly connected to a particular instance, or you can have that kind of group control of more broad allowances. What is the common use case for something like this over a more traditional role-based system? These came out of social networks and social messaging sites, things like that, where you might want to be able to control off of different comments, different posts.

between different users in really specific ways. And so something like a rule-based system was not going to be fine-grained enough. And so the kind of defining moment in development of relational database or relational-based access control was Google released a paper called Zanzibar that talks about how they had built their access control systems. So across all of the different Google apps, anything that's going to SSO with Google, things like that, that can get really complicated and really hairy in the myriad ways that... these entities might connect. And so they developed this system, wrote a paper about it, and is kind of taken off from there. Now that you've got a chance to play around with it, do you think that if you were building your own project from scratch, would you start with a system like that? Or would you start with a role-based system and only sort of evolve towards a relational-based once you hit certain milestones? Absolutely. I think...

Just because of the way that I like to iterate over software projects, my answer is definitely going to be that I'd want to start with the simplest version that gets the job done first out of the gate. Just like the client did here, they had a role-based system. It met all of their needs until their feature set got more complex. And then they're upgrading into this relational system. The relational system is a lot more work to set up and kind of maintain. So it's not a cost that I would want to pay up front. That's fair. The thing that has me thinking a lot about it is...

is there an opportunity to use the kind of associations that are built in to rails that we're using all of the time to make a relational access control system off of those relationships like they already exist like a user has many posts can we use belongs to kind of relationships to make those connections between the user and an entity the role the group and that sort of thing i haven't really thought a ton about it yet but i think there's really something there the only problem i see is rails being really strongly backed with relational databases they're not great at graph traversal. So there are definitely some interesting challenges to solve there, but I think there's a real opportunity to. I guess that means we'll have to have you back on the show at some point to hear where your thinking's evolved there. Yeah, I'll talk about the new gem that I just wrote or something like that.

Well, that's what's been going on for me. What's up in your world? I've got a chance to try out a new diagram type with mermaid.js, which is a sort of GraphViz-like system for a text-based representation of graphs, but then it has a fancy renderer built on top of it. It's built into things like Obsidian and GitHub, so you can get fancy graphs just embedded directly in a PR comment or something. Something that I've got a chance to try is they have just regular sort of directed graphs.

But they also have a git graph type. So you can have different branches with commits on them that merge into each other and things like that when you're trying to demonstrate a particular git flow or something. And so I got a chance to try that out this week, and it's pretty fun. Nice. I know your love for Mermaid.js and all of the different visual styles it affords. When you're saying a git graph, is this, I'm trying to picture it, the sort of typical thing that you might see with lines that represent branches, little dots for nodes, and new lines will kind of sprout from those nodes to represent where new branches have started? Yes, that's exactly correct. Nice.

How have you found that? It works pretty well. What's interesting is that the syntax for it feels a little bit like an actual history of Git commands that you would run. So it's all things like branch, branch name, commit, commit, commit, check out other branch name, commit, commit, check out a different branch, merge other branch. So it reads a little bit like a history of commands that you'd run on the terminal, but it produces this beautiful visual of what the Git history would look like. That's really nice. I know I've used Mermaid a few times and I've found it a little...

kind of awkward to jump into a new syntax for the different graphs. So that's really nice that it sort of matches something that you're already doing. Are there particular concepts that you've been able to get across better with this style of graph? What have you used it for? I've mostly used it as a tool to talk about iterative development, and particularly the value of breaking up larger chunks of work into smaller pieces and showing particularly how it fits in a Git history, how that might look nicer, how that might speed up your workflow. An example I used recently was talking about you're doing some feature work, and as part of it, maybe you're doing a little bit of refactoring, but that refactoring might have some pieces that are controversial. If you put both of those up in a PR on the same branch, you're going to get a lot of pushback from the team.

and your feature is going to be held up from being merged for a long time. Instead, if you put the controversial bits into a separate branch, PR that separately, they can build off of each other, and that's fine. The uncontroversial feature work will probably get approved, and you can merge that and go live pretty quickly. The controversial parts can sit in PR, and you can have a big conversation with your colleagues, or you can even build alternative solutions for it. You might reject it altogether, and in the end, something else gets merged a week later, but you didn't hold up the feature work. And so that helps you to move faster, not...

to be as blocked on things just by organizing your work a little bit better. And the Git graph is a great way to demonstrate that. Yeah, I love what you're saying there about separating out something that's going to hold up the feature work. I think that's a fantastic way to handle that. And I can definitely see how having a little visual representation of that would help it get in my head a little bit quicker than I'm trying to do the visualization myself. When you think about the scenario I described, do you think of it in terms of visualizing a commit line in your head? Or how do you tend to visualize that? Hmm, that's a good question.

I generally do have to draw out relationships and things for more complicated concepts that I'm working through. But I think because of the way that I interact with Git completely in the terminal and like a text-based application called lazy, lazy Git, that those sort of representations are just sort of what's stuck in my head about how Git works. So in this particular case, I don't usually think of the lines. Although that is how I learned. There's a Git visualizing tool that I linked to people. I don't have the name of the website right at hand, but we'll put it in the show notes where you basically just type Git commands and see them visually what happens. That's great for learning Git when you're trying to figure out, oh, what does it mean when I rebase one branch over another? Or what does a merge look like? And it just builds out this whole visual thing. So that's a great resource. And I think that's similar to you. I typically do my Git work on the command line. So I don't have the visual, but I think of the visual in my head. It was, yeah, definitely easier to learn.

Git, which can be really difficult to wrap your mind around when you're first encountering it. Those visualizations really help. The thing that really cracked Git open for me was I took a workshop by former host of the bike shed, Chris Toomey, back at RailsConf. It must have been 2017, 2018. And he broke down the data model of Git really well. And I just sort of learned it from the other direction, from the bottom up, instead of seeing the bigger picture. And that's maybe another reason why I don't picture that kind of visual representation as much anymore. Fascinating.

As someone who has both participated in workshops and given some of your own, how do you feel about approaching workshop content from the bottom up rather than the top down? Yeah, I think it's going to depend on the subject, really. There are going to be some concepts that are easier to get into if you know the whole story and then break it down. And there are going to be some that are easier to get comfortable with if you're taking your group along step by step and sort of building up to the big picture. So like everything, the answer is it depends.

I feel like the more sort of extreme ends of things that I've seen have been sort of how the Haskell community approaches their pedagogy of how they like to teach new people coming into the community versus how the Elm community does that. Both languages are similar syntactically. There's a lot of similarities between them. But the Elm community is very sort of practical focused. They're not trying to teach you the big ideas of the language. They're trying to get you productive as soon as possible. And the idea is that as you start building experience with the language, you will sort of start seeing the larger patterns and you'll eventually pick up on the bigger ideas. Whereas Haskell takes the opposite and says, hey, let's teach about the fundamental ideas here, applicatives, monads. And once you get a sense of these big patterns, then we can sort of look at building code using those patterns. How do you think you tend to be in terms of maybe your personal way of engaging with new content?

Do you tend to seek out the big ideas and the patterns first? Or do you try to just as minimum amount of learning you can get to be productive and then figure it out later? I think probably the second one. The way that I really like to dig into a concept to get it sort of into my DNA and like muscle memory of how to think about it is to get something end to end working and then just start to break it in a lot of different ways. Like, I don't expect this will work. How will it fail? And kind of go top down from the larger concept.

seeing how and where it breaks apart to come up with those more individual concepts. It's really interesting that the differences in the community and the way that they are viewing their language and how they work with it can influence the way that they teach it so much. I see that definitely in the Elm community, the desire to get to productivity really quickly. So you gave a workshop at RubyConf last fall. Yeah. Did you sort of follow a particular pedagogical paradigm or how do you structure the content to try to make it accessible to the people who participated? For my workshop, it was actually a workshop version of a talk that I had given the RubyConf before. It was building the unbreakable code who's breaking one World War II. And it's about the Enigma machine.

We didn't get into cracking the Enigma machine because just recreating the thing was complicated enough. And so I was lucky enough to have a talk that kind of walked through the process that I can then turn into a workshop. In that instance, I did follow a very small pieces added together, building up to larger components approach, because not only was I trying to get across how the Enigma machine worked, but we were also talking about object-oriented principles, using the Enigma as a way to get into those ideas. And it was the couple of concepts that I had picked out really fit together well if they were given in a particular order. I didn't know that order starting out, but as I was kind of working through and building it myself, thinking about how these things work together, that structure sort of came for free almost.

Do you feel like the concepts you were trying to teach really built on each other? Or were they sort of independent enough that you could do them in any order that you wanted? And maybe that allowed you to work in a narrative order rather than a teaching order. I want to say maybe there's a third way there because it seemed almost like, I guess it's closer to the building on each other, but not quite. Because they weren't necessarily dependent on each other, but sort of functionally it made sense as a building to go through in that order. Maybe a little bit of both now that I'm thinking about it. It's not the third way, it's both at the same time. Have you ever...

Given a workshop? I have, yes. I did a sort of intro to Elm workshop because I'm introing sort of a lot of new topics because I don't expect anybody to know the language yet. I go in like very bite-sized pieces and I did it in such a way that every, you're able to sort of like do a complete little mini arc. You have a motivating problem that you're trying to add some kind of new feature that's going to, oh no, require you to learn something new. But only enough new stuff that you can like learn a single concept, use it to solve your problem, have a working app and commit.

And then I'm going to throw the next feature at you. And oh no, your current knowledge set isn't quite big enough. You're going to need to learn one more thing. and then we're going to introduce one more concept. You're going to learn that. You're going to get this feature working, commit, and so on, until you've learned a pretty big chunk of things. But some of these steps are really, really small. They're just like, hey, I can display text on the screen. But it's a new language. You've got to learn a couple things to get to that. Yeah, absolutely. And then it's like, OK, well, let's make a dynamic. Let's introduce background requests. Let's do all these things. But it's one at a time. And I think that forced me to really think about how I could break things down. I noticed, as you started to describe your methodology there, that

I was thinking at first that it sounded like the Haskell approach that you had talked about. You were talking about small pieces and breaking them apart, but then really quickly you said that it was in service of creating a feature and being more practical in the approach. And I find that fascinating that you used both techniques at the same time. Yeah, so the breaking down into small pieces, definitely, but they're practical pieces. So I'm not leading with theory. I'm leading with practice. I think that's the difference between them. Yeah, that makes sense.

A thing that's nice that you can do with something that has a sort of web front end is that it can be very visual. And I think that helps with a little bit of that dopamine rush of getting a thing to show on the screen, which is fun with something more back-end. You're like, oh, we're going to do a terminal-based program that's going to do some things, and then we're going to get something out to the terminal. But there's something about being able to have things show up in a browser and be interactive that I find particularly satisfying. And so trying to bring in as many of these little loops that will say, hey, what's the minimum amount of work I can do to get you to another one of these little wins that will get you excited and then motivate you to learn the next thing so you can get that next win?

try to build that in. That was a part of the way I was thinking in terms of the design. Nice. One of the things that I sort of struggled with defining about how I wanted to set it up when I was building out my workshop was where to sort of hand over control of the moment to the group, to let them play with the concepts or let them try and figure out the next piece. As you were thinking about these small steps, were you planning in where you might let the attendees sort of drive? No, it was relatively scripted.

These steps were really small, so there was not a whole lot of room to do things differently. Maybe you wanted to split something into two functions or something, but for the most part, everybody's going to end up doing about the same thing, and there was less room to explore. And I think maybe it was different because for me, I'm teaching a sort of intro to a language rather than showing patterns and then allowing people to explore how those patterns might interact. It was also relatively fast-paced rather than, hey, here's a problem. Explore it and see where you can get. Nice. Was yours a little bit more open-ended for...

audience? Yeah, it was actually. I wanted to present a little bit of a solution, talk about the concept that led me to that solution, propose the next problem, the next step, the next feature, whatever it was, and let them apply what we had just learned a little bit and maybe run into some blockers, maybe do something a little bit different and have that kinesthetic experience with the project. And I thought that the interesting part, because I was thinking so much about the object-oriented concepts and what I wanted to teach, I had broken up the pauses around those concepts rather than the functionality of the emulator that we were building. And I found that during the workshop, people were having more fun and they had more freedom to solve the next problem. And then we got to talk about the different ways that it happened and the approach that I took. And so rather than thinking about, oh, I don't know.

encapsulation or something like that. They were thinking about how to make the next cog work with the other thing. And then we could kind of step back and take a look at the concepts through there. So it was really one of those moments of you can't tell the audience what's going to be interesting. They will tell you what's going to be interesting to them during. I like that. That's a more open-ended nature. Did that feel a little more risky to you? I think for sure. It definitely did. My talks are usually very scripted. I know every word that I'm going to say. And this was a big departure from that for me.

A little bit, I was worried about filling as much time as I had, but that ended up not being a problem at all. And I worried a little bit that people might go in just completely different ways and take steps that kind of got off script for me that wouldn't let me get back to the sort of flow that I was trying to do. But I think it was broken up at good points and they sort of steered me. I sort of steered them that it all worked out pretty well. The way I had structured mine is that I had a sample repo people would pull, but I also had commits in the repo where I had like solved the exercise for each of these. I think it was...

some like 15 different save points over the course of an hour and a half. So they're relatively small chunks of work. And so people could always sort of look at the answers, if you will, or even check out to that commit if they wanted to, like if they were feeling stuck and didn't want to fall behind. And so I sort of used that as a tool to help keep the pacing on track. Did you sort of think a little bit about how you wanted to deal with pacing or, because the classic thing in a workshop is that somebody gets stuck and everybody else is ready to move on and this person has a weird local machine issue and that derails the whole thing. How did you think about dealing with that situation?

I said very similarly to the way that you just described yours. I had a repo that people could pull down or they could start with a blank text file. I wanted to avoid that. Everybody's trying to get on the conference Wi-Fi. Nobody can pull the repo down effect. So I made sure that.

If you needed to, you could follow along just with a text file. But I had the same sort of thing of like save points and thought that if anyone had fallen behind or if they took it in a really different direction, they could just check out the next commit. We'd all do that together and move forward. But of course, even with all of that kind of like setup and planning for contingencies, there were some environment and setup issues that had come out of completely left field. And that took a few of the folks, their sort of first or second hands-on break to solve for themselves.

No amount of planning will save you for all of your different situations that might come up. But my group banded together and actually helped each other out getting through those setup issues. It was actually kind of a fun little bonding exercise at the beginning of the workshop. Nice. For you, how do you think that workshops differ from maybe less in terms of your process, but more in terms of a medium? So you're thinking of you have an idea and you're wondering, oh, how might I give that idea as a talk versus as a workshop? It'll probably develop in very different ways just because of the constraints and availabilities of the medium.

How would you contrast those two? A very interesting perspective on this because this same project that became a workshop started from a talk. And so a lot of the kind of groundwork had been laid by myself in the past with the talk. I think the thing that I was focusing on making different was having that opportunity for hands-on time, having that opportunity to experiment and explore. Because my talk version really walked through the building of this emulator very step-by-step, the workshop version kind of...

allowed for that room to play. And so that's why I really wanted to have moments to let people try out the concepts that we had just learned and we had just gone through. Try it out, see how far it can get you, and then maybe run into a bit of a block that then I'll throw back to me up on stage and be able to show the next concept and move through. So I think for me, the difference between a talk and a workshop is sort of... maybe not attention, but like the focus of control almost. I feel like in a talk should be the speaker and in a workshop should have more opportunity for the participants to kind of get their hands dirty. How does that contrast with your approach to the Elm workshop? I think a workshop takes up more space time-wise, but anything that you're doing with multiple people is going to be dramatically slowed down. So the amount of code or concepts you're gonna be able to work through in a workshop are a third of what you think you're going to be able to do.

and you have much less control over that than you would over a talk. Because a talk, you can rehearse, and you can get that content to fit just right, and you cut and you make it fit. A workshop, I guess you could rehearse with an actual set of people, but it's harder to do solo. And so I think you have to be... almost more intentional in a workshop about the content that you keep versus the content that you cut. Yeah, for sure. I definitely resonate with that, especially if things are going to get more amorphous from time to time, knowing what are the key concepts that you can kind of come back to as things are meandering around is super important. And I actually found that even the difference between the talk version and the workshop version, we got the exact same amount of work done, even though the talk was 40 minutes and the workshop was two hours. So I absolutely understand what you mean about the work.

kind of spreading to fill the given space. The advantage with a talk, right, is you can just put a code slide up on the screen and be like, yeah, we did the thing. We wrote the code. Next. Whereas those like 5 or 10 or 20, that scary 20 lines of code on a slide, that can take a while for the people in your workshop to write those down. Absolutely. Especially if you're not the best typist or you're not a touch typist, you're going to have to split your focus like that between the screen, your computer, the speaker. You're actually asking a lot of the participants to...

really keep up and and stay along with the content and that's a big responsibility to make sure that as much as you can you don't leave people behind yeah i feel like that's almost more of a risk with a workshop i feel like you've got you almost have to have a narrower sense of who your audience is for a workshop because if you miss just one little piece it's much easier to fall behind in a workshop as opposed to a talk you can sort of like oh you know there's a section where i like maybe didn't fully understand but then the next sort of chapter in the talk i got back in and then that made sense and i followed through the end of the talk Maybe I'll go back and rewatch that first section that I didn't quite get, or I'll ask a follow-up question to the speaker. But if everybody's in a group session is writing code and you didn't get that and things build on it, you might be out of luck. Yeah, I think that's why the approach that it seems we both took to have something that the participants could step forward to, almost give them a little tool that is kind of like getting caught up in a talk, could be super helpful.

Not every group is going to be the same. They're going to go off in different places. They're going to get stuck in different places. They're going to breeze through things that some groups took really long with. So there's always that. I guess it's a risk, but it's also kind of where the good stuff happens in a workshop. That's what you get. It's individualized to the group instead of the talk going to be pretty much the same. This reminds me of the idea of catch-up mechanics in video games and board games, where to keep a game interesting for everyone.

People who are further behind, there are often mechanics that might benefit them. They kind of push them back towards the front so that you don't have a situation where you find out halfway through the game, oh, I've fallen behind. There's no way I'm ever going to win, so why play the game to completion? And that's no fun for the person who's ahead who would like to play the game to completion and actually get that real win. Yeah, that's a big component of Euro-style games, I think, as a category of a board game. Some of my favorite games have mechanics like that that...

the winner is a complete surprise because it's the person who was behind the whole time and they did something just right that fell into place and can bring them along. Just like in a workshop, leaving someone behind is something that you're trying to avoid. And in a game, having people lose interest and not play, they're not participating anymore. It's the goal of some games, but those are not my favorite kind of games. Did you have any other catch-up mechanics built into your workshop? Not really. I was lucky that most people were actually able to get the repo, and so I didn't need to worry about too many other ways to bring people along. But I was certainly ready to get derailed. I had notes and everything. I knew where I wanted to go, but I also was open to the possibility that that would all go out the window as soon as I met the participants. But luckily, we kept pretty close together, and I didn't have to fall back on that. Or not fall back on that, but I didn't have to go to there. Were there other catch-up mechanics that you had?

sort of taken into account or brought into your workshop. I had a couple, I had one TA who was available. So if anybody got stuck or was unable to sort of keep along, I could sort of keep moving the instruction forward and the TA person could go and give you individualized attention and help you figure out the situation where you were stuck. That's really great. I'm curious for you, how do you feel about workshops as a way to gain familiarity with a concept versus something like a talk? When you go to the RailsConf, do you tend to find that you like talks or workshops better? What makes you pick one versus another when they're head to head against one another in the schedule?

Yeah. How do you decide what to go see? That's always the challenge at a conference, isn't it? When there's multiple tracks, I think I would probably gravitate towards talks, but in saying that and thinking about the Git workshop that I had mentioned before. that really changed the way that i work so fundamentally and was such a great like workshop and experience that maybe i've been doing it all wrong maybe i should be looking for these workshops a little more often than i do do you find that you gravitate towards one or the other or is it based on the topics i tend to avoid workshops just because they're a really big time investment and i feel like i often don't get that much out of them and it probably varies right if you get like a really exceptional one that's like better than the two or three talks that you would have done if you had not gotten to them. But my sort of default is oftentimes to focus on the talks rather than the workshop. I know many people feel the opposite in that a workshop, you have to be there. And so you get the value of the workshop in person and you can always listen to the talks at home when they're put on YouTube. So I know a lot of people will sort of take the opposite approach, but I tend to focus on the talks. Yeah, that's a really interesting way to look at it. I feel...

As I've gone to more conferences, I find the hallway track actually to be the most rewarding, satisfying part of a conference to me. And so in a little way, I am taking the approach of the people that are going to workshops because those are happening there and you can't recreate that experience and that the talks are always going to be around later. So I think I kind of resonate with that, even if I don't gravitate towards the workshops. You mentioned that Chris Tooney's workshop on Git fundamentally changed the way that you think about Git in a really good way. I know Chris also did a whole series for Upcase on mastering Git. Have you seen that content at all? And do you know if that resembles what was in that workshop? I'm wondering for listeners who might be thinking, oh yeah, maybe I could do for an update in Git and learning the fundamental model sounds like a great idea. Where can I get this? I'm wondering maybe this content is the closest thing that they can get. I don't know for sure, but I would have to imagine that they are related. Chris had said that it was the first time he had given that workshop, but everything...

was very polished and fit together. So it felt like material that he had spent some time with. I don't know for certain, but I do know that those trails on Upcase that I have done have always been really solid. We'll put a link to it in the show notes for any listeners who are curious about digging more to get with some of this content that Chris put together. On that note, shall we wrap up? Yeah, sounds good.

Frequently asked questions

How long is this episode of The Bike Shed?

This episode is 37 minutes long.

When was this The Bike Shed episode published?

This episode was published on February 4, 2025.

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.