frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Show HN: Git for LLMs – A context management interface

https://twigg.ai
58•jborland•10h ago•17 comments

Show HN: OpenSnowcat – A fork of Snowplow to keep open analytics alive

https://opensnowcat.io/
51•joaocorreia•6h ago•12 comments

Show HN: I built a tech news aggregator that works the way my brain does

https://deadstack.net/recent
134•dreadsword•7h ago•75 comments

Show HN: Deta Surf – An open source and local-first AI notebook

https://github.com/deta/surf
116•mxek•13h ago•38 comments

Show HN: Nostr Web – decentralized website hosting on Nostr

https://nweb.shugur.com
86•karihass•11h ago•20 comments

Show HN: Open-source TypeScript SDK for sending and operating iMessages

https://github.com/sg-hq/imessage-kit
2•RyanZhuuuu•2h ago•2 comments

Show HN: ScreenAsk – Free Screen Recording Links for Customer Support

https://screenask.com
13•ladybro•8h ago•0 comments

Show HN: I built Kumi – a typed, array-oriented dataflow compiler in Ruby

https://kumi-play-web.fly.dev/
4•goldenCeasar•3h ago•0 comments

Show HN: FlowLens – MCP server for debugging with Claude Code

https://magentic.ai/flowlens/
4•mzidan101•4h ago•0 comments

Show HN: Hacker News sans AI content

https://tokyo-synth-1243_4mn1lfqabzpz.vibesdiy.app/
4•neom•4h ago•1 comments

Show HN: Tommy – Turn ESP32 devices into through-wall motion sensors

https://www.tommysense.com
68•mike2872•8h ago•54 comments

Show HN: Cuq – Formal Verification of Rust GPU Kernels

https://github.com/neelsomani/cuq
91•nsomani•1d ago•60 comments

Show HN: Silly Morse code chat app using WebSockets

https://noamtamir.github.io/morwse/
73•noamikotamir•5d ago•30 comments

Show HN: Desponsorize – Gray out Amazon sponsored search results

https://github.com/candacelabs/desponsorize
2•kaashmonee•5h ago•0 comments

Show HN: Play abstract strategy board games online with friends or against bots

https://abstractboardgames.com/
171•abstractbg•1w ago•79 comments

Show HN: Coyote – Wildly Real-Time AI

https://getcoyote.app
6•michalwarda•8h ago•10 comments

Show HN: Cadence – A guitar theory app

https://cadenceguitar.com/
191•apizon•1w ago•86 comments

Show HN: BesiegeField – LLM Agents Learn to Build Machines in a Physics Sandbox

https://besiegefield.github.io/
2•zepist•6h ago•0 comments

Show HN: Story Keeper – AI agents with narrative continuity instead of memory

https://github.com/neurobloomai/pact-ax
2•neurobloom•6h ago•0 comments

Show HN: Emojiwhat – Unicode and TikTok emojis to copy and paste

https://emojiwhat.com/
2•kyrylo•7h ago•0 comments

Show HN: Pg_textsearch – BM25 Ranking for Postgres

https://docs.tigerdata.com/use-timescale/latest/extensions/pg-textsearch/
5•tjgreen•8h ago•0 comments

Show HN: Distil-NPC: a family of models for non-playable characters in games

https://github.com/distil-labs/Distil-NPCs
4•party-horse123•8h ago•0 comments

Show HN: Modshim – A new alternative to monkey-patching in Python

https://github.com/joouha/modshim
108•joouha•1w ago•31 comments

Show HN: hist: An overengineered solution to `sort|uniq -c` with 25x throughput

https://github.com/noamteyssier/hist-rs
3•noamteyssier•9h ago•3 comments

Show HN: Katakate – Dozens of VMs per node for safe code exec

https://github.com/Katakate/k7
120•gbxk•2d ago•51 comments

Show HN: 401K Traditional vs. Roth Calculator

https://401k.pages.dev/
2•vjain014•9h ago•1 comments

Show HN: How Software Fails-A book about complex system failures(sample chapter)

2•enginyoyen•9h ago•0 comments

Show HN: A browser for Mac that connects to private web apps over SSH

https://outerloop.sh
2•mrcslws•9h ago•0 comments

Show HN: I built an SVG-generation tool

https://scalablevector.graphics/
6•ninapanickssery•10h ago•1 comments

Show HN: Create interactive diagrams with pop-up content

https://vexlio.com/features/interactive-diagrams-with-popups/
43•ttd•1d ago•5 comments
Open in hackernews

Show HN: Play abstract strategy board games online with friends or against bots

https://abstractboardgames.com/
171•abstractbg•1w ago
Thought I would post in celebration of 1 year of my website being online. I've been working on it on and off and currently the website allows users to play Hex, Tumbleweed, Amazons, and Connect 6 against friends or against practice bots. I've been a long time player of some of these games and I felt for a long time that the world could use a few more popular abstract strategy games compared to Chess or Go.

If you try it, let me know what you think. I'm always looking for new games or new features to add :)

Comments

zem•1d ago
pentominoes would be a nice one to add (players take turns placing pentominoes on the board until someone cannot make a move). probably one of the variants like callisto or blokus where everyone has the same set of tiles in different colours and there are rules around connectedness
abstractbg•1d ago
Thanks! Will try to figure out a version that I can get permission to add to the website.
quuxplusone•22h ago
You're welcome to "Capture Pentominoes" if you want it!

https://club.cc.cmu.edu/~ajo/free-software/pento/pento.html

abstractbg•22h ago
Perfect, thank you!
zem•19h ago
the usenet thread was a fun nostalgic read - something about the character of discussions was different from the ones you get on web based fora like reddit today. (also i recognise your name from rec.puzzles, which was another nice nostalgia moment :))
Paracompact•1d ago
As a longtime chess and go player, I was just doing some research the other day into what modern abstracts are out there. I was disappointed by how dry I came up.

Even if you expand the search criteria to include video games, there just aren't many deeply strategic discrete-time games that weren't invented centuries ago and have players online at any given time. Here I exclude games that are perpetually changing and/or have strategies locked behind progression systems and paywalls, such as TCGs and virtual deck builders. The very few exceptions I found were niche Discord communities around games like Tak, Hex, or Advanced Wars.

When did we as a society lose the appreciation for these things? I get why including a component of dexterity in strategic video games (e.g. RTS) is to take full advantage of the medium, but all this in conjunction means we are very likely never to see another deeply studied cerebral game like go, chess, shogi, mahjong, etc. arise ever again.

abstractbg•1d ago
I'm very hopeful that the problem is simply lack of general awareness of these games, and that once there's enough content surrounding them, we'll have a healthy population of people playing more abstract strategy games.

Fun story regarding Hex. It nearly reached what I would call a "mainstream" audience with the movie "A Beautiful Mind" about John Nash starring Russel Crowe. Unfortunately, the Hex scene was cut from the movie! You can watch the cut scene at https://www.youtube.com/watch?v=pTZ3nn2Bge4

ixwt•1d ago
I'm not an expert by any means, but there isn't much to draw people to other games aside from curiosity. When it comes to Chess and Go, there is significant money on the line. Chess was also a proxy fight during the Cold War.
Paracompact•23h ago
> I'm very hopeful that the problem is simply lack of general awareness of these games, and that once there's enough content surrounding them, we'll have a healthy population of people playing more abstract strategy games.

Possibly, but my takeaway from COVID is that games like chess and go (which experienced a bump in popularity during the lockdowns, and have since been dwindling back down) are not merely gems waiting to be rediscovered, but instead appeal to outdated tastes in gaming, and are unlikely to be replicated given market realities. You need approachability for game-one beginners, you need vivid and eye-catching visuals, you need progression systems and content drips to keep players hooked, you need monetization to milk the whales, etc.

mrichey-9988•1d ago
The facebook group "abstract nation" has made having a facebook account worth all the bad stuff associated with facebook. Its lead me to find a ton of great abstracts. It also pointed me to this: https://play.abstractplay.com/
Buttons840•23h ago
What about something like competitive Dominion? There are expansions, but the game is symmetrical. All players have the same abilities in an actual game.

Or Spirit Island at high difficulty?

Paracompact•23h ago
I love Spirit Island! Probably played more hours of that game than any other in my collection. Granted, it's purely a cooperative game, and much more artificially complex ("fiddly") than any abstract. It's the immortal simplicity and competition of games like chess and go that I was looking to rekindle, but I guess they aren't well suited to modern gaming tastes.

Dominion is also great, and in its simplicity literally invented the deck building genre. But it, too, is too artificially complex to become immortal, even before you get into its 16+ expansions. The proliferation of the deck builder genre also makes it less likely any individual game is going to be deeply studied.

Credit to games like YINSH, anyway, that specifically try to appeal to competitive, deep, and mathematically simple foundations. They just don't have what it takes to thrive in the age of monetized bright flashing lights.

Buttons840•21h ago
I haven't played YINSH, but I haven played some of the other games in that "series". You're aware of the others, right? Which is your favorite (YINSH I assume)?
vintermann•21h ago
Not the OP, but I liked ZERTZ. It's very symmetric, almost, but not quite, an "impartial game" in mathematical terms (where you don't have to know whose move it is to know if they have a winning position). You can set up the most outrageous combinations, going from nothing to having won with a series of forced moves.
Paracompact•16h ago
Only played YINSH, and only a few games. It's nice and I would be eager to play more, but unsure if I would study it deeply myself even if it did become popular. A lot of abstracts tend to blend together in my head as "combinatorial slugfests," where the player with the most RAM in their brain wins out. (YINSH may or may not be like that, don't take my word for it.) It's the primary reason I switched from chess to go, where my propensity to make tactical blunders can at least be offset by larger scale planning and good directional judgement.
dole•20h ago
Dominion is licensed in various online incarnations anyway.

With limits on expansions and other rules, it is possible to get Dominion competitive enough to study games and optimize for turns, the original isotropic had a decent ranking and rating system (RIP and add’l shoutout for their implementation of the Innovation card game):

https://dominion.isotropic.org/leaderboard/

zem•20h ago
I also think the gipf games haven't had enough time and volume of players to see if they are actually as long-term engaging as chess and go. my speculation is that the classic abstracts are the ones that turned out to have an almost accidental emergent depth that kept people playing them even as better and better strategies were devised, because they were never "solved". it is unclear if yinsh will turn out to have an endless stream of better and better strategies emerge, or if it will be fun while people figure it out but plateau when they do.
zahlman•8h ago
> competitive Dominion

Ah, I miss the old Isotropic implementation....

cgreerrun•21h ago
Check out Quoridor
abstractbg•21h ago
Yeah, that's another nice one. I've got a Quoridor set myself.
johnecheck•21h ago
This seems like the time to mention my unreleased board/video game, Star Gambit!

It's a turn-based abstract space fleet battle coming to your browser in 2026. It's already playable over the internet w/ time controls and ratings. If that interests you, join the discord for updates and playtest invites!

https://stargamb.it

https://discord.gg/f3MUSkJjFx

barbs•18h ago
Looks interesting. What engine/framework did you use to make the game?
johnecheck•12h ago
Dart/Flutter, custom engine. I've been very satisfied with it; does everything I need for a 2d game. Once it can natively use shaders/GPU on web it'll be perfect.
BrenBarn•20h ago
I think "have players online at any given time" is the tough part there, but it's also the tough part for many non-abstract games. If you look around on Board Game Arena you can see a number of abstract games available, but with many you'll struggle to find any other players.

There are some cool abstract games out there, but they're not super popular. Abalone was mentioned in another comment here. Octi is another cool one. Some like Azul or Patchwork have a light theme but it doesn't really affect the rules, mostly just an excuse for the piece design, which I think puts them in a similar category to chess.

I'm not sure I'd say we won't get another deeply studied game. I mean, if we're comparing to chess and go, it will take hundreds of years to really know if any modern game has that staying power, let alone remains interesting to analyze. But I do wonder if we'll ever get a game that's both deep and popular.

The popularity of RPGs, TCGs, and expansion-based games suggests to me that a lot of people really like feeling a sense of immersion into a "world" that's constantly revealing new "content", rather than discovering new variety within an existing system. Maybe this is just a stereotype, but I also feel like there's a synergy this and a similar vibe prominent in stuff like fanfic, where people like to engage in this sort of generative building on some core ideas. The "pure" or "cerebral" gamer who is really interested in the ramifications of a fixed ruleset is somewhat more rare. Also there are so many games out there now that even cerebral gamers may be tempted to explore new ones rather than digging deep into familiar ones.

This is just to say that maybe some existing abstract games are actually deep, but in order to know that, we'd need people to take the time to analyze them and explore them. Maybe time will tell.

Paracompact•16h ago
Yeah, Azul stands out as a great exception to my rant. I get similar feelings playing it as I do playing riichi mahjong (a fabulously deep emergent blend of luck, planning, and player psychology). It's also managed a decent popularity not just as a quasi-abstract but as a board game, period.

There is another aspect to the thing I didn't mention but you did: It may be that the sheer amount of interest in board and video games at large is more to blame for the lack of deep abstracts than is the inherent lack of appreciation for them. In sum total, there are probably games out there currently that could be mechanically fit as a successor to chess or go, but with so many competing games out there (both cerebral and otherwise), even the cerebral gamer is unlikely to pursue them singlemindedly.

mylesp•20h ago
A fun and fairly popular one is Hive. It is a physical game akin to chess but has an active community online on https://hivegame.com/
bondarchuk•15h ago
It's only correspondence play, but this site has 209 games at the moment and a reasonably active player base https://play.abstractplay.com/games

Most games are quite modern, including some designed this year. And they have Homeworlds!

AtlasBarfed•12h ago
Civilization?

Master of Orion?

Here's a wealth of turn-based strategy out there, especially in the retro days.

hermitcrab•7h ago
I loved the first versions of Civilization and Masters of Orion. I tried one of the more recent version of Civilization and didn't like it anything like as much. But I am a lot older - so it could just be me. How do more recent versions of Masters of Orion hold up?
kkukshtel•11h ago
The space is being actively developed but it also requires you know where to look.

This is a journal that tracks a lot of new abstract stuff: https://www.abstractgames.org/

Notable recent releases:

https://boardgamegeek.com/boardgame/352238/turncoats

https://boardgamegeek.com/boardgame/2655/hive

https://boardgamegeek.com/boardgame/272380/shobu

https://boardgamegeek.com/boardgame/430875/high-tide

baruchel•8h ago
May I suggest the Greenchess website? https://greenchess.net/ It certainly has some huge potential: nice minimalist interface and probably about 50 or 60 chess variants for all tastes. Some of them are obviously more or less "recent". You can easily start a new game or accept a pending invitation. I would call the set of players quite "active" and certainly motivated. But it is true that the pool of players remains small and you generally play against the same players again and again.
abstractbg•8h ago
Very cool website! Also, check out 5D Chess with Multiverse Time Travel XD
7373737373•1d ago
https://en.wikipedia.org/wiki/Abalone_%28board_game%29 would be cool
abstractbg•1d ago
Thanks!

I haven't added a commercial game before, but I will reach out to the owners of the game and see what I can do.

ixwt•1d ago
I'm personally a big fan of asymmetrical games. A game I've wanted to play but have never had the board to play it on is Unlur [0]. Arimaa [1] is another one with some history behind it that is uncommon.

It is very much appreciated that I don't have to make an account to play. That is one of the most annoying thing on sites like these to play games.

[0]: https://www.iggamecenter.com/en/rules/unlur

[1]: https://www.iggamecenter.com/en/rules/arimaa

abstractbg•1d ago
Thanks for the suggestions! Guest accounts are definitely a necessity. I had been looking at some asymmetrical games like Adugo and Viking Chess, myself.
sloum•23h ago
I also really like asymmetrical games. In particular the various Tafl[0] reconstructions. Some are unbalanced, but some are very balanced and fun to play as either attacker or defender. There are various versions with rule variations to accommodate various board sizes too.

I have not played Unlur. Looks like a cool hex variant. I like the initial phase where who plays white is decided. It is a neat way of working that out.

[0]: https://en.wikipedia.org/wiki/Tafl_games

dgrin91•23h ago
As a lay boardgamer, what makes a board game abstract?
vkou•23h ago
Chess is an abstract proxy for a wargame, but is not in itself a wargame.

'Abstract' is somewhere on the chess side of the spectrum between Go and moving miniature battle tanks around and flipping to page 237 of Appendix E to look up how much water the average Italian soldier needed to boil his pasta in the Tobruk campaign.

mNovak•8h ago
As an aside, I have a slightly masochistic desire to implement an online-playable version of Campaign for North Africa. It bothers me that they can credibly claim there is no documented case of someone finishing the game.
quuxplusone•22h ago
No text, no chance, perfect information.

If you have to read things, roll things, or hide things, it's not an abstract.

(This fails to include backgammon and Parcheesi when maybe it should, and includes Zark City when somehow I feel it shouldn't, but it's not a bad starting point.)

Additionally: No dexterity (which is kind of a special case of "no chance").

Paracompact•21h ago
The most important subjective criterion, IMO, and the one from which all other objective criteria are just approximations, is that the game be as rules simple and as strategically deep as possible. If there are any superficial criteria, it's that the theming be "austere" (which is what makes Zark City stand out slightly).
fsckboy•21h ago
is there a game that is an abstract version of D&D?

i've always been curious about the gameplay, but have absolutely no interest in reading past words like "wizard", "mage", or "eldritch", and i won't look at anything where i have to see or imagine that somebody is wearing a cape, or even play with gamers who are willing to imagine they are wearing capes in other games.

krapp•21h ago
Why do you have such an aversion to capes? There's no rule saying you have to play as a character in a cape.

If you're asking if there's a version of D&D that's pure math and no roleplay, the closest thing I can imagine would be Dragon Dice, or something with the mechanics of a very basic roguelike. The math in D&D itself such as in combat is affected by your character's stats and the encounters created by the DM, so the mechanics and the role playing are kind of tied together.

vkazanov•18h ago
The thing is that dragons and magic and Harry Potter all become irrelevant at some point in life. Power fantasies do have an expiration date.

Now, the genre itself, all the storytelling involved, can easily be adapted to more serious, or even abstract, thinking.

In fact, it was, and there are plenty of alternative rpg universes. But, similar to how serious non-marvel movies are a niche, serious rpgs are also less popular.

atakan_gurkan•14h ago
I understand where you are coming from, but I hope you realize this is very subjective. The things you mentioned and other elements of the fantasy genre do not become irrelevant for many people through their life, well into adult life as well. In fact, they may become irrelevant and become relevant again.

Just to make it clear (and perhaps to state the obvious), you are not believing in these when you play these games or read these books, you are voluntarily suspending disbelief.

krapp•11h ago
They're all games. They're all escapist power fantasies. No one wants to role play a character who doesn't matter to the story, regardless of the genre.
vkazanov•6h ago
Well, this depends on a story. Escapist? Yes. All of them, by the very definition of, ehm, role play.

Power? Not necessarily. Numerous rpgs have nothing to do with power. Be a Cat has nothing to do with power, and everything with escapism (and cats).

I mean, there is nothing wrong with power and fantasies. Or powerless fantasies, as in horror stories, or whatever.

vkazanov•18h ago
There are literally 1000s of non-DnD RPGS!

I am also past the age of wanting to believe in magic. And also past the age of having time for long plays. Most of my friends are busy doing serious adult stuff.

So outside of baroque dnd-like systems, my personal favourite right now is a subgenre called journalling solo RPG, which are often minimalist, but covering all kinds of themes and topics.

gwd•14h ago
I mean, the most important part of a role-playing game like D&D is... the role-playing. You're basically telling a shared story, with each player controlling one character, and the DM controlling everyone else. The fighting mechanics are sort of a mini-game inside the larger game.

If that's all you want, there are loads of combat games with miniatures. I've never played Warhammer, but that's obviously the biggest one; I've enjoyed the Star Wars X-wing combat game as well.

jader201•21h ago
To me, when I think “abstract”, I think “no theme”. Games that aren’t abstract have a theme that helps make sense of the rules of the game.

There are also games with themes that don’t really help make sense of the rules — the theme is just pasted on. These are still considered abstract, despite technically having a theme.

The “ameritrash” genre are known for having strong themes that tie to the rules of the game. E.g. a lot of co-op games with plastic figures.

Old school euro games often have a pasted on theme, and are more abstract.

bix6•21h ago
I don’t quite understand this tumbleweed game but it seems fun so I’ll look up a video.
abstractbg•20h ago
Here's the rules video used by the Tumbleweed community https://www.youtube.com/watch?v=mjA_g3nwYW4

Also, in case you are curious, Tumbleweed has a discord https://discord.com/invite/wu6Xdtt497

They are currently playing through the 2025 Tumbleweed World Championship. Lots of strong players there!

bix6•5h ago
You should put that video in the info link, super helpful. I am getting crushed by the bot lol. Fun game!
mylesp•20h ago
Hive is my favourite for anyone looking for another abstract strategy boardgame. It has been very fun and feels very deep.
workingonit3•19h ago
Hive is a really nice game.

This year we picked up Homeworlds, there are more rules, but scratches a deeper itch. Plus you feel like a space general while playing lol.

https://youtu.be/Nz16s6oCIlQ?si=lzSYIZZG4LtSLZUf

nenadalm•19h ago
nice website, Thanks for sharing. I like board games, so I'll definitely try some (or all) of them.

> I'm always looking for new games or new features to add

Nice feature to add would be single device multiplayer for the games.

I play board games mostly in person. I know that there are travel versions of many games... but I don't like to carry too much stuff. So I've created 2 games that I sometimes play with other people (e.g. on a train, bus, or anywhere else if there's nothing better to do).

Both games have only 2 player mode on single device. They're PWAs with offline support, so you can install them on your phone and don't need internet to play.

These are:

- backgammon: https://nenadalm.github.io/backgammon/ (link to the rules at the bottom in menu)

- virus wars: https://nenadalm.github.io/virus-wars/ (link to the rules at the bottom)

abstractbg•19h ago
Thanks, single device multiplayer is a great idea.

Someone else recently mentioned to me that Virus Wars is their favorite game! I'm glad to see it getting some love.

aidenn0•18h ago
The name of my favorite abstract strategy game escapes me, but it is played on a solitaire[1] shaped board, with lines connecting the dots indicating valid moves (all lateral moves are allowed, diagonal are somewhat restricted; I can draw the board from memory). It starts with the offense occupying every spot except for the nine that make up one limb of the cross (the "fortress"). Defense has two pieces only, placed anywhere inside the fortress.

Offense can move any piece by one dot, following the lines, and cannot jump. Defense can move or jumping the pieces from the offense, in any direction, following the lines.

Jumping is obligatory, consecutive jumps are allowed (and also obligatory, e.g. you can't not take a double jump).

Game ends when:

1. The offense occupies all 9 squares of the fortress (offense wins)

2. There is no legal move for a player on their turn (that player loses)

3. The offense has fewer than 9 pieces left (defense wins)

For practiced players #1 is the most common end to a game and the offense gets a number of points equal to the remaining pieces; players then switch sides and the player with the most points wins.

1: https://en.wikipedia.org/wiki/Peg_solitaire

Paracompact•16h ago
Asalto? https://en.wikipedia.org/wiki/Asalto
aidenn0•11h ago
Bingo!

That looks exactly like the game I'm thinking of, but it was not named that. What I played was in one of those books that came with checkers and had different board-game printed on each page, and I'm guessing "Asalto" is trademarked or something.

quuxplusone•6h ago
The Book of Classic Board Games called it "The Dalmatian Pirates and the Volga Bulgars."

https://boardgamegeek.com/image/926606/the-book-of-classic-b...

aidenn0•2h ago
That image looks exactly like what I remember; thanks!
animal531•17h ago
I love games and once back in the day had the idea of using some form of genetic algorithm to combine game elements to make new games (Today we'd probably replace the genetic part with some form of NN).

I spent a lot of time collecting and breaking down game elements from all board games I could find, but as things go with ADD I then ran out of steam before I had any kind of functional prototype.

As always I highly recommend The Player of Games by Iain M. Banks, and although its not his best book I really enjoyed the game aspects in it.

Paracompact•16h ago
Are you familiar with The Glass Bead Game by Hermann Hesse?
pred_•16h ago
How about 2 players playing on the same device, i.e. offline play?

https://en.wikipedia.org/wiki/K%C5%8Dnane might be a fit.

tasuki•16h ago
What do I think? It's amazing! Got crushed by a 300 bot in Hex, then beat a 100 bot, then again lost against 200. It's great how one can see the review immediately afterwards!

Feature requests:

- TwixT - just TwixT PP like on LittleGolem is fine (much easier to implement).

- Quoridor - a delightfully incomprehensible game.

- Larger board sizes. Hex starts being really fun at 19x19!

Questions and suggestions:

- On reddit you mention you "used AlphaZero-style methods to train the bots" - I suppose the networks are size-dependent? You could look into the many KataGo improvements[0].

- You mention the source code isn't released. If you released it, people could help add games.

Again, very well done, thank you!

[0]: https://github.com/lightvector/KataGo#training-history-and-r...

almostdeadguy•13h ago
I love TwixT, I discovered the Bookshelf Games version of this recently (a friend of mine's parents had an old copy at their beach house). Also a big fan of Alex Randolph's Ricochet Robots.

Dieter Stein's games are also supposed to be wonderful abstracts (Urbino, Fendo, Tintas) though I haven't had a chance to play those yet.

abstractbg•11h ago
Wow, beating a 100 bot is still impressive in my book.

Thanks for the feedback and suggestions!

Yes, the networks are size dependent right now. It's a great idea to copy-paste and then adapt the KataGo network architecture since it isn't size dependent and has been proven to reach superhuman strengths.

tasuki•9h ago
I'm about 1950 on littlegolem, but trying again, I just lost a couple of games in a row against the 100 bot.

A couple more questions/remarks:

- It seems everything is happening on the client? My CPU (I'm on a laptop without a real GPU) goes wild during analysis. But also I don't notice any big bag of neural-net-weights being downloaded. Mind sharing how it works?

- Care to share more about the networks? How long did you train the networks for and on what GPU? Circa how many params? Any and all details you'd like to share :)

- This Tumbleweed game is fun! This game seems somewhat inspired by both Hex and Go, and the author lives in Warsaw. Interestingly, I lived in Warsaw in 2020, am very active in the Go scene (4 dan), and at least know the names of the people who play Hex, and yet somehow never heard about Michał or Tumbleweed before...

abstractbg•8h ago
The analysis happens on the AI server.

Sans proper profiling, I would guess that the CPU going wild during analysis is due to a combination of 1. analysis is streamed live to the client in 20 simulation intervals 2. some post-processing on the client side 3. the fact that I am using a global context and reducer in React which causes the entire page to re-render each time an update happens.

The networks are simple Resnets with a value and policy head. It's 20 layers with 128 channels per layer. I trained for several days on 2x 4090s. However, recently I trained a few networks (Hex 14x14, Amazons 10x10, Breakthrough 8x8) on a GH200 and it was 2x faster, roughly 100 ckpts per 24 hours for Hex 14x14. I'm not sure about the number of parameters but the .pt and .ts files are on the order of 30-90 MB. There's definitely room for improvement using tricks like quantization during selfplay inference.

I'm very happy you like Tumbleweed! If you're curious there's a Tumbleweed community run by Michał (the creator) https://discord.com/invite/wu6Xdtt497 They are currently playing through their 2025 World Championship.

anentropic•15h ago
> Hex is a two-player abstract strategy game invented independently by Piet Hein in 1942 and John Nash in 1948.

This sounded very intriguing - what are the chances of two people inventing the same abstract strategy game!? At the least there might be some interesting story about how they arrived at same idea.

Wikipedia phrases it differently though:

> Hex was invented by mathematician and poet Piet Hein in 1942 and later rediscovered and popularized by John Nash.

Ok less amazing

abstractbg•12h ago
Sorry, thanks for pointing that out! Yes Nash rediscovered it. Will update once things calm down
robobro•9h ago
Reminds me of boardspace, though the UI is much cleaner.

You should try to add the Gipf games at some point, they are very wonderful!

fhcbix•2h ago
I was just going to recommend boardspace.net here! It also somewhat recently went open source: https://github.com/ddyer0/boardspace.net

It clearly is stuck in the Java Applet era where it started but Dave eventually made an Android port and modern Java has no issues running it!

hermitcrab•8h ago
Have you come across the Mancala family of board games? I learned to play the Hus variation in Namibia:

https://mancala.fandom.com/wiki/Hus

It is fun how it is deterministic in theory, but it is hard to predict more than a few moves ahead - so seems random in practice. Also that you can go from being way ahead to losing, very quickly. I implemented a game engine for it. Can discuss if you are interested.

zahlman•8h ago
> deterministic in theory, but it is hard to predict more than a few moves ahead - so seems random in practice. Also that you can go from being way ahead to losing, very quickly.

How does one become skilled in something like that?

hermitcrab•6h ago
I guess similar to chess. You have to be able to see more moves ahead than your competitor. The game tree is quite narrow compared to chess, but complexity is added by the cascading nature of moves. It is a fun game. I recommend trying it. Good for kids as well.
kej•6h ago
This reminds me of the game engine "Zillions of Games" from the '90s. It was a basic system for describing abstract strategy games and then an AI that could play them with you. They're still online but I don't know how active the community is: https://www.zillions-of-games.com/