Do we really need a chess.com for Go?

[shinuito]

Welcome and thanks for your contributions.

@marinakai

I’m dying for a better native mobile client than using OGS or Go Magic in a web browser, or using one of the two mediocre iOS OGS clients. I’m so annoyed, I suspect I’ll end up building this myself.

The frontend of OGS is open source, and some of the apps might also allow individual contributions, but there’s threads for apps like Sente and Surround so you could make suggestions there.

But if you have a vision for another mobile app/interface go for it

@marinakai

OGS’s UGC puzzles rock and have some extremely high-quality content, but Go Magic’s streamlined unified system (courses + Duolingo skill tree) are so much more coherent for a complete beginner

Yeah it could probably take a good bit of work for someone to be able to match GoMagic’s content and the way the skill tree works. Who knows, maybe they can do some collab or something like they suggest in their slide deck pitch.

@marinakai

Coming from chess, it’s a surprise to me that I can’t always immediately matchmake into a 9x9 game whenever I want on OGS. I get that one solution here is probably “jump up to 19x19”, and another solution is “make Go more popular in the West”, and a third viewpoint expressed in this thread is “that’s a feature, not a bug”

Agreed, though I also play much more niche two player games than Go, and you’d be lucky to find a live game at all that wasn’t prearranged. A lot of niche games only get played in correspondence until they hit a big enough following with people playing regularly

Probably some marketing is needed, and maybe even a way to market 9x9 as a quick fun alternative to 19x19 to get people to play it even if they prefer 19x19, but just don’t have time for a full 19x19 game.

@marinakai

The thing I miss most from chess.com is the detailed AI analysis that doesn’t just say “here’s the move where your win chance really tanked”, but automatically shows you both the bad line implied by your bad move as well as the optimal line that follows from the optimal move (which you can theoretically accomplish even locally with KataGo and your preferred frontend clicking through recommended moves one by one, but not having to do that work yourself makes game review much easier!). They also have (and had pre-LLM, so it’s not all LLM nonsense) nifty AI-generated explanations along the lines of “oh yeah, this move is bad because you hung your bishop”. The latter is a more difficult problem (and probably more difficult for Go than chess), but the former seems purely a UX concern, not a technical limitation of current Go engines.

Maybe such suggestions can be combined into a bigger thread about improving the AI analysis. For example there was a recent thread

Improve AI Feedback

Since some days i enjoy OGS AI analysis and its quite usable (having a smaller but annyoing bug regarding positioning stones reported elsewhere). Previously i was using download and Katrain, which is a much more complicated path - but it has a very handy additional feature. I understand that the selection of “next moves” in a given frame is limited. Now many times i want to compare to other moves and its great that the OGS app allows this and returns immediately with an updated analysis of the…

about improving the AI feedback, so maybe we could collect a few ideas there and see if there’s a few big improvements that could be made at once.

@marinakai

I’d probably be less hungry for something “more like chess.com” if I had the ability to find in-person community and learn that way.

You might be able to find discord groups, that while won’t exactly replicate the club feeling, but can still give a sense of community.

The forums is also a community in one sense, and maybe the main site english chat is another.