cross-posted from: https://piefed.zip/c/machinelearning/p/1092955/looking-for-ml-coders-for-help-with-open-source-creative-commons-board-game-ai-player-logi

I know this is probably a long shot, but I’m not sure where else to ask so I’m going to take a shot.

I’ve designed and abstract board game (think chess, shogi, go, etc) and have completed coding the rules for play against an AI player, however getting the actual AI to be good is a whole other problem.

I would love if someone who is experienced in ML would be interested in collaborating on this open source project.

The game is strictly a hobby project, with absolutely no plans for monitization or anything. Currently it’s playable in the browser against AI (no multiplayer yet set up) at: https://greenants.github.io/Amalgam_Webgame/

GitHub Repo: https://github.com/GreenAnts/Amalgam_Webgame

Disclaimer: I’ve mostly used AI to code this project, as I’m a pretty novice programmer. Obviously that’s controversial, so I want to make that clear - but remember this is simply a hobby project, and is a way for me to get my board game design digitized and actually played by others. The code will likely be a bit on the messy side, but I think for the most part the ML coder would only be interacting with the controller - so shouldn’t be too much of a factor.

From my limited understanding, the actual search depth and complexity of the game is quite high, far higher than chess, so it’s been quite hard for me to try and get this set up even with the help of AI coding with hueristics.

If you are interested in in the project at all, I’m always looking for help to farther this project - as I’ve been working on the board game itself (on and off) for more than 10 years.

The GitHub Repo listed above (in the README.md) has a graphical rulebook as well as a video tutorial linked for you to learn the rules and get an idea of the game complexity if you are interested.

Like I said, I know this is a long shot, and unlikely anyone will be interested, but I figured I’d give it a shot :)

  • Tabitha ☢️[she/her]@hexbear.net
    link
    fedilink
    English
    arrow-up
    2
    ·
    21 hours ago

    No offense, your How to Play page is full of Lorem Ipsum, I think you could get more out of search/heuristics before you need to burn $10,000 - $100,000 on H100s.

    I’ve mostly used AI to code this project

    first, optimize the FUCK out of your How to Play document. I read the first document I could find and I wasn’t sure if you move all your dudes in one turn or if you move only one dude each turn. You need a “cheatsheet” type reference document for mid-to-experienced players, a document designed for people who might not even know how to play chess/checkers/go, and you need a youtube video, and you need an FAQ. I didn’t think what you had was bad, just incomplete and not battle tested against hordes of new players. If you game’s GUI is really well designed, this could alleviate the burden for solid documentation, but you probably can’t send that to an AI. Then have clear documentation on the APIs and data structures a bot would have access to in order to read the game state it was access to (probably simple, as I did not see fog of war elements), and the APIs it would use to place moves.

    Give all of this (minus the youtube video) to both chatGPT and claude. see if chatgpt/claude have advice on how to make the interface for potential bots better. If possible, spend time optimizing this first.

    Give all of this (minus the youtube video) to both chatGPT and claude. Tell you it you want a design document markdown file for building V2 of your game bot code. I suggest allowing the user to pick if they face against V1 or V2 of your game bot. You’ll need to include that in the scope of the design document. Ask it to ask you a shitload of clarifying questions. You should eventually acquire a single design document. You should spend a lot of time on this document before moving forward, even if you’re just vibe chating chatgpt the whole time. If chatGPT/claude doesn’t ask you upgrade from free to the first paid tier at least once, you might need to spend more time on the design document. Read and fact check the final document. Commit this to your repo’s docs folder.

    Then ask chatGPT and/or claude for a task list markdown file optimized for zencoder/kiro’s spec driver design feature. Open a new account in only one of those (zencoder: you have one week to use 770/day premium AI API calls per day, Kiro: you have 30 days to use 1000 credits). Last time I checked, neither require a credit card to get started. If you’re already paying for claude code and don’t want to open an account for free AI trials, try claude code’s “planning mode” feature. Only pick one tool, you should be able to accomplish this using only one of these. After you picked one, make sure to open them up to the spec-driven-design feature. give them the design document markdown file and task list markdown file. Kiro/Zencoder should translate the documents into it’s own preferred format, then be able to work on the tasks sequentially under your supervision.

    You should have enough remaining credits to fix broken tests (or add tests), make amendments, or you can blow them on unrelated new features.

    • UnfinishedProjects@piefed.zipOP
      link
      fedilink
      English
      arrow-up
      2
      ·
      21 hours ago

      Thanks for the detailed response.

      One of the major issues is that I am not paying at all for any LLM, and only using free Claude projects (Claude code I think requires a paid API)

      So this vastly limits my context and capabilities.

      Regardless - this is some solid advice I will definitely try to apply.

      As for the rules:

      1. Main website’s rule page - yeah this isn’t yet implemented.
      2. Are you talking about the actual rulebook: https://github.com/GreenAnts/Amalgam_Webgame/tree/main/assets/Rulebook . Because if so, these 5 pages I thought were very well outlined if you actually read the pages in order and in their entirety. If these are the ones you actually read, and are still confused, than that is good to know and means I need to revamp my rulebook.
      3. The YouTube rules walkthrough exists, and is linked in the main GitHub Repo.

      Other than that though, thanks so much for all the advice and tips - I will try to optimize an API/documentation of the rules.