If Go Were Going To Evolve, How Would It, And Why?

The one problem I have with 3d and triangle go is that they both create too many liberties. It’s probably something I can eventually adjust to, but right now it feels like there aren’t any major attacks, like you can’t even do a ladder with trigo.

Although I do admit I was thinking of cubic for 3d go, if you can do a sort of tetrahedron go where there are only four adjacent spaces, it might work.



From the Wikipedia go variants page, here is another picture of diamond go (along the lines of what @mekriff and @Eugene mentioned):

1 Like

go would only have opening+middle game, endgame would be simulated by AI
endgame is played but once everything is pretty much settled, AI would automatically scores for you instead of asking both sides if they want scoring


Lol, less than a week ago I offered to resign a game in endgame because I thought I was behind by 10 – 15 points. I ended up winning by a point + komi. That was at 6k.


Haha yeah been there :stuck_out_tongue: always best to count :wink:

In the time since this thread was active, @seequ has set up a variants server at https://go.kahv.io/, the Variant Go Server (VGS).

In October of last year, he ran – in association with the Nordic Go Dojo – a tournament of pixel Go, a variant developed by Mikkgo and Antti Törmänen in which one places 2x2 dango instead of single stones (see NGD Pixel Go Tournament is tomorrow!).

As I’m discussing variants, I’ve noticed that one-colour Go seems to have become more popular, mainly due to its integration on CGS (the Color Go Server). Interest in Thue-Morse Go also continues. Let’s also not forget Alex’s 25x25 correspondence tournament, Hunting at Night (Hunting at Night: The 1st 25x25 Tournament).

There are also interesting ideas in Features everyone secretly want on OGS but will never be implemented.

If we examine modern Go culture, there seems to be widespread acceptance of three variants, which is to say “standard Go” on the board sizes 19x19, 13x13, and 9x9. Historically, of course, the traditional rules of Chinese, Tibetan, and Korean Go were locally popular, but have now been largely supplanted by Japanese or “standard” Go.

An important part of the evolution of Go thus far seems to have been first a development, and then a rejection, of set stone patterns. It’s possible, as I said earlier in the thread, that set stones might return to the game. This might well be in a semi-randomised form analogous to Chess960. There was a thread about this in 2020, Semi-random starting position, but it was a bit too technical for me to follow.

Of course, we’re seeing an evolution of both local and global play since the AI revolution, with emphasis shifting to territorial, precise, and confrontational play relying more extensively on reading; we’ve the normalisation of the early 3-3 invasion and the abandonment of some popular 20th-century moves like the slide.

I don’t really envisage Go evolving in what could be called a “chronospecific” sense, which is to say with the greater mass of the game shifting all at once. Rather, I expect that what will happen is that variants will continue to branch from the trunk of 19x19 standard Go – which will itself remain healthy for an indefinite time – and will rise or fall in popularity over time, proportionally to how interested or bored players are with the “trunk game”.

It’s possible that variants like one-colour Go, random-opening Go, Thue-Morse Go, 21x21 etc. will gradually grow in popularity over the course of this century, and it’s also possible that they might not. At this point, of course, we cannot say.


I love one color go its so good for the brain. Learning how to memorize a full game and play them out entirely should honestly be a goal made early on by everyone. One color go can really help flex that muscle that helps you memorize full games and playing them out from move 1 to finish.


Chess makes a good analogy to the “trunk / branch game” model.

The trunk game is “chess”, aka Western chess or, most formally, international chess.

Some of the branch games can be observed at Top Rated TV: IM Abik02 vs FM FlamingFM • lichess.org – Chess960, king of the hill, three-check, atomic chess, horde etc. The trunk game is the central point of the site, in this case Lichess, and the site provides a space for the branch games.

Chess itself is not being replaced by another game, but its popularity is allowing player drift into chess variants, which are evolutions of chess.


A lot of posts were about busywork in the endgame.

If you get bored replay old games with a small margin victory. It turns to be really impressive much beyond the basic knowledge we share. Like playing a pretty gote move letting the opponent squeezing a bunch of points in sente, but then getting wonderful sente from that gote move before. Or not answering a 50 points threat because you have one of 51… Search variants, what if… For one side it’s easier to understand but for the other side it still can be very deep.

1 Like

Instead of placing a stone, you could choose to remove (not take as a prisoner) any of your opponent’s stones, except the most recently played one. Your opponent could not immediately replay that stone because it would cause a repeated board position. This could create entirely new types of ko fights, and groups with two or more eyes may no longer be unconditionally alive. Ending the game might be a little tricky and require some additional rules.


Thanks to computers (for convenience), we can imagine in a extended way that we start from a board fully covered of stones and play the whole game by removing stones.

I like this idea!

Since the games are likely to be long, you might want to play on a smaller board, like 15x15.

Neat idea but I do have a question: What would prevent players from repeatedly removing opponents stones and stalling the game indefinetely?



  • 17x17 → 19x19
  • No komi → komi
  • Empty board → Chinese opening → empty board
  • Stone scoring → area scoring + territory scoring divergence

The only thing I think is extremely likely is standardized scoring and that people will completely abandon either territory or area scoring.

The game is better on 23x23, but people have less time and are more stressed, so I think it is more likely that the board will be reduced in size.

More than 2 players seems like a good idea but is probably too chaotic.

Maybe a pie rule like in Hex could automatically balance komi.

I think future rule changes will be subtle and that they were more radical historically because people didn’t have as much contact with each other across different cultures as today. So there will be less divergence and more standardization.


What is your reasoning for skipping over 21x21 straight onto 23x23?


Exactly, and why does everybody skip the even sized boards? Is Tengen so important to you?

Maybe they like an odd number of points on the board :slight_smile:

Hm but for what reason? Isn’t it because people are accustomed to the star points? I remember this post:


1 Like

Well, I really think 27x27 is the next logical step, but 23x23 is inbetween and the star points become sort of uneven on 21x21 (and 25x25). 21x21 is also not that different from 19x19.

Yes, tengen is that important. Need the Great Wall Fuseki, which is the only real way to play the game. Great Wall at Sensei's Library This is also the next obvious step in the evolution of fuseki, AlphaGo just needs to become a little smarter and realize it.