Ask HN: Which programming games helped you become a better programmer?

Also, I don't particularly love the act of writing code itself, I like the results (problems getting solved or fun things happening) of which there are none in most of these games.

Interestingly it seems it's the exact opposite for games simulating other kinds of jobs. I hate driving, I do it to get somewhere and nothing else. But trucking simulators? Endless fun and relaxation. Same thing with chopping onions, which I only do to make a nice meal. But in a burger joint simulator? I'll gladly slice tomatoes for two hours!

Playing Battlesnake has helped me become a better programmer. I've used it to gain experience in multiple new (to me) programming languages as a freeform way to practice without deadlines, I've learned new algorithms and how they apply to game solving, and I am beginning to understand how many game-solving algorithms can be applied to more practical applications since "games" in the game-theoretic sense can just as easily be models of real-world situations where the players are rational people instead of "game" participants. For example, "pathfinding" and "decision making" has many parallels.

Learning game tree searches such as minimax, MCTS and CFR is teaching me about tree/graph algorithms and data structures, how to work with, reason about, and test explosive search spaces where the entire tree can't be feasibly completely exhausted, in the process of researching these things I've learned to read whitepapers in the literature that would otherwise be scary to me as a developer and actually care to understand the math, it inspired me to tinker with Formal Methods including TLA+, and all the while having fun and making friends.

Learning Goal Oriented Action Planning, or GOAP (which I actually didn't apply to Battlesnake in the end, but I never would have found it if I wasn't looking) helped me optimize my real-world productivity by translating the concepts from game AI to personal task planning. In my head there is an analogy now between handling backlogs according to the current working context, and a GOAP-like stack based finite state machine. Learning a simple game algorithm helped me procrastinate less.

For newer developers, you'll also out of necessity learn many real-world web development and ops concepts because the way Battlensnake works is that you run your own web server conforming to their move API. So you have to keep your snake AI up and reachable, responding to each request within the allotted timeout, and know the basics of HTTP on day one. Games can also run concurrently so you learn how to deal with concurrent games and whether to make them stateful or stateless between turns, which can be a different experience on different web frameworks/languages and might teach you about things like threading or the actor model or distributed systems.

For advanced developers, there's always somewhere to go next. Your snake AI can be anything from hard-coded rules to tree searches (as mentioned above) to deep reinforcement learning.

https://play.battlesnake.com/

Both are somewhat obliquely related to programming, but I really enjoyed incredible.pm and (the demo of) vim-adventures.com

You're absolutely right, it is extremely hard to emulate the programming experience via such games. If you break down the skills you need in most (with obvious caveats) programming related jobs - reading others' code, explaining your code to others, designing, documenting, brainstorming, breaking down problems, etc, etc, it is difficult to train for those. In the abstract, it makes sense to practice those skills since for many programmers there isn't anything exciting happening day-to-day, and there is a lot of grunt work maintaining existing code bases.

In most physical fields (athletics being an easy example) there is an amazing amount of carry-over from training into performance. Training being where you build up skills and performance being when you apply said learned skill to the performance. But in more abstract and cognition-heavy fields its muddy and vague.

I've played a couple of these games (Opus Magnum, EXAPUNKS). What I liked about them is that they distill the experience of software engineering to "just the fun problem solving" part.

In these games, you don't write RFC, set up CI for your project, do code reviews or battle old dependencies. It's just you, a clear goal and all the right tools to reach it - somewhat similar to assembling Lego (or Ikea furniture) but with the fun of problem solving added.

Making something real would definitely net you a more satisfying end result, but maybe wouldn't be as entertaining. Obviously this depends a lot on what you find fun in programming and how smooth your real projects go!