Battlesnake With Brad Van Vugt

Yeah, thanks so much for having me. I'm glad we could make it happen.

Yeah, yeah, at its core, Battlesnake is a multiplayer competitive programming game, and so we invite sort of the developers of all skill levels to come and build a BattleSnake, and then you can enter that BattleSnake into leaderboards and tournaments, and/or you could just play with friends or colleagues or classmates. And basically, you're programming an algorithm to control this snake that plays the game competitively, and you're trying to beat out everyone else who has competing algorithms and competing servers. Yeah, I think one of the things that's really unique about how we built BattleSnake is you can use any technology you want. So, it's not, you know, you're not coding in our web editor like some other things are, but it's really the way that you play is you have to actively deploy a web server that runs your algorithm for you. And so, that makes the spectrum, the solution space is very broad. And so we have developers that are doing many different languages, many different algorithms, many different libraries, many different backends and hosts and that sort of thing. And so you see this really broad spectrum in this community that's filled with developers of all shapes and sizes and backgrounds of experiences.

Yeah. So BattleSnake really started as sort of like a local developer community thing. I think that if you look at the broader space of, we'll call it developer education or developer learning, if you are starting to program, there's like a million different ways to do that. And a bunch of other, a bunch of services and platforms will take your money. And try to teach you how to do for loops in Python, which is great and that should exist. But the core problem that we saw was like, if you know how to program, if you're an intermediate or senior or above developer, there's nothing for you to like actually apply those skills and try new things and learn new things. You can take courses, you can watch conference talks, you can do that. But like, there's nothing, you're not applying anything. And developers like to learn by building. So we kind of came up with the idea of BattleSnake of like, okay, how do we want to learn new things? Well, we like to learn new things by building projects and applying them. Okay, let's make a sandbox or a problem space that we can all try new things and learn sort of in our own ways. And so it was sort of like me and a group of friends from the local community up here in Canada, kind of came up with the idea of BattleSnake. And then we held— the first version of it was actually we just held like a live event. And we had— we invited like a bunch of friends and we were expecting like, I don't know, like 20 people to show up. And we had like 150 developers come out and just build, like spend an afternoon building BattleSnakes. And then at the end, this really sort of fun— And this was kind of the creation of BattleSnake was like, We were like, well, what if we just run a tournament at the end of the day? It was like, okay, great. So like someone fired up an Excel sheet and put all of our server URLs into it. And then someone else, we were running through the tournament and someone jumped up on stage and started to like shoutcast the games. And so it was like, the Blue Snake is going up into the left corner and, you know, the Python Snake is going down here and the AWS Snakes are failing for whatever reason. And that made it like really fun and engaging. And that was sort of the birth of BattleSnake. And we were like, this is actually really fun, not just for us as developers, but for other folks as spectators as well. And so that kind of kickstarted a series of live events and then eventually into the sort of live platform that exists now.

Yeah, I think that speaks to like, and we didn't set out when we decided on the game Snake and sort of competitive Snake, we weren't like, oh, this is really interesting from a competitive perspective. We were like, this is an idea, let's see what happens. But it turns out that like the competitive space of competitive Snake play just in general is like actually really complex. And as programmers, there's like, there's so many different aspects of it that you can work on and you see this reflected in the community as well. And the players is like, You can fine-tune algorithms or heuristic trees or AI agents to play and try to win. But you can also go for code optimization. We have some people that just try to win BattleSnake games with the fewest lines of code, which is great. That's awesome. If that's fun for you, then you should do that. But also go for latency. They're like, I'm going to dig into the Rust HTTP stack and I'm going to figure out where I can squeeze more milliseconds out of this so I can try to make more decisions. So it really, in terms of complexity, there's this deep well of game complexity and competitive complexity, but there's also this whole other side of technical complexity that the community really, really likes to dig into.

I think that speaks exactly to the crux of what makes BattleSnake so fun. It's like you can solve it in a million different ways with a million different technologies and whatever makes fun for you. This is like when we talk about education, it's not really education. It's like we're giving— I want to give developers an excuse to build something with the technology they've wanted to build with. And that's really what it is. The keyboard-controlled Snake, there's a fun story there. I think this was like the second live event that we ever held. And we were like, let's put, I think it's like a $100 prize up for first place or something. And currently in the current version of the game, you have, I think you have 500 milliseconds to respond, but that originally wasn't the case. We were like, great, we'll give you 5 seconds to respond and you can run all this compute, whatever you want. But in that second tournament that we ran, there was a team that figured out that in 5 seconds, you could WebSocket a game board out to a browser and then do a keyboard input. So it was actually, it was like some, one of the competitors was actually just playing live on their laptop the whole time. And I think they got second place in the tournament. And we were like, great, like good on you. That's, if you figured that out, that's awesome. And then we lowered the timeout to try to limit that. But yeah, I love— hearing of like crazy solutions like that.

Yeah. So we kind of, we originally planned this for, the original version of BattleSnake was like a monolithic Django app. It was bad. And then we very quickly realized that we needed to rethink that. Now it's kind of split up into essentially 3 pieces. There's a game engine, which is open source, and it's written in Go. It's super highly concurrent. And that's what powers all the request concurrency. And that runs— it's running in GCP right now on like a Kubernetes cluster, and it's auto scaling as it needs to. So that's like the high performance piece of it. That was written by Go developers that are a lot smarter than I am, frankly, but it's done really well. And then the other 2 pieces that exist are sort of like, there's a primary Django app that kind of glues everything together and schedules tournaments and schedules leaderboards and does all the scoring algorithms. That's a lot— the real-time demands on that are a lot lower. And then the 3rd piece that we have is actually the game board. So the game board is like a whole separate repo. I think it's done in SvelteKit. because we rebuilt it a couple of— we rebuilt it like a year ago. But the whole visualization that you see in the real-time visualization is completely open source. And so we have developers that run variations on that, and you can run the game board locally, and you can replay tournament games or leaderboard games locally if you wanted to put a different UI on top of it for your your friend group or your classmates or whoever you're playing with. And so we see lots of really cool variations there. So a lot of it is open source. Most of it is open source. And the other important thing there in how it's built is the— I talked about the game engine. The rules for deciding who wins a game is also open source. And this was a goal of ours because early on we get a lot of, well, this didn't resolve the right way, or I'm not sure that— There's lots of questions from developers questioning, is the game game engine working properly or not? Or how do I understand this complex scenario and how it played out? And so we realized right away that if we build this to scale, all of that can just be open source. And so we don't answer technicalities over the rules anymore because we just say, hey, go look at the code and you can trace it through and see exactly what happened.

We have, but Like 95% of the time it's their own, it's their own bug. Um, but we have, we have had people that have found and fixed bugs in the, in the, in the game engine. Uh, it's usually around like very strange edge cases and sort of timings.

I think it also, yeah, like 100%. And I think it also was designed a little bit, or like some of our intent with doing that was to, this is more of a far-fetched dream, but like, we want the sort of, you know, BattleSnake players start playing the game, start building a snake, and then eventually they graduate to contributors or some of them graduate to contributors. And not all of them are like, you know, deep Go concurrency developers. Some of them develop in JavaScript and build UIs, and that's great. And some of them build backend apps, and that's great. And so we intentionally wanted to like create different pockets and different contribution points. Again, speaking to the idea of like our developer community is so diverse, how do we serve a contribution model that kind of can speak to everybody?

Yeah, definitely. Definitely.

Uh, yeah, I mean, definitely if you're interested in, in playing BattleSnake, it's at play.battlesnake.com. Or if you just search for BattleSnake, you'll find it. The one thing I will mention is the project itself is run entirely off GitHub sponsorship. So we rely on the community to— and our community is wonderful and lovely and giving and incredibly supportive. And so if you're looking for a good GitHub project to sponsor, BattleSnake's not a bad one. And then we also have a Discord. So if you want to check out what people are doing with their snakes or how you know, what the current meta is, uh, you're welcome to, to join the BattleSync Discord and hang out with us there.

Yeah. Thanks so much for having me.