Tumbleweed: this game weirdly reminds me of Go

Tumbleweed is an abstract strategy game, designed by Polish mathematician and game designer Michał Zapała. Somehow, the rules are totally different than Go (placements are based on “lines of sight”), but somehow, a territorial game with captures emerges from the very simple ruleset. In this game, however, groups don’t always live and die together - so different notions of good/bad shape are used. I have played it now over a hundred times at the iggamecenter and I’m totally hooked. Probably, a strong Go player could learn the game quickly and become the champion of this growing community. I love this game, and I hope to see more Go players discover its riches!

Tumbleweed rules explained here:

Iggamecenter home page, to play Tumbleweed, Go, and other games here:

Try it out! There is also a facebook page and Discord for scheduling games.


This is very interesting, every piece is like a pole, and any other pole “seen” is like added a designated outline with an “invisible” thread between them. Every pole added in between makes the line stronger, and harder to break.

This is like circling land plots and claiming ownership. I wonder if it would work on grids. Or even on a board with a web of connections. This idea of marking lands by borders gives me some idea what might be the missing link between ancient shape rearranging and custodian capture involves moving pieces and chances became “settled” and territory oriented simply due to safe zone naturally arise out of “encirclement and capture” from afar.

1 Like

I think you’re right. It would be really interesting to get your perspective on strategy after playing a few games.

Since a triangle is important, I feel the six corners would be like 4-4 points in Go, while the sight offset tiles next to them would be like the 3-4 to make a “remote” connection to the opposite side of the edge tiles (like encloser but with a big triangle). The key would be to make moyo (big triangle) as fast and far apart as possible, but not too far to make the opponents can invade with their line of sights to break them. The one that can surround and cover the center neutral tile would win the game if everything else is equal and got the edges covered. Mirror-tactic could work to a degree? Not sure about the local fight around the center would evolve into, and some type of komi would probably be needed to offset the advantage.

1 Like

Hi Claire, would you like to try the game on http://www.iggamecenter.com/ ?
We would really love to have such a strong Go player like you try Tumbleweed!
Just let us know when you are free and we will arrange a game :smiley:

Claire , we have discussed komi, but currently the game is being balanced using a 2-stone pie rule:

The idea is that the two players are the Guest and the Host.
The initial board has a neutral 2-stack on the center cell.
Then, the Host places a Red and a White 1-stacks on the board.
The Guest chooses whether to play first with Red, or second with White.

Using the pie rule, the Host is motivated to place Red’s first stone in a suboptimal location, so that the choice whether to play Red or White is not clear to the Guest.

Right now, the community is not strong enough for issues of balance to matter yet, statistics show that 50% of games are won by Red and 50% by White on iggamecenter.

Please let me know when you might have time to try playing a game of Tumbleweed!

I still think it is not enough for a 2-stock center cell to off-set the first pick priority.

Consider if any square is slightly better or worse, the best way to place stones at the start to prevent your opponent to gain an advantage from the start is to pick exactly mirror with each other, sort of like the ancient Chinese fixed stone placement. And if they all place next to the center, then the first one to pick (2nd player) would have the advantage to take the center tile, like a liberty race. Maybe making the center tile open with komi (or up to 3 stacks, hence the surround wouldn’t be able to take it, and would require further line of sight to take it?)

And If a tile has 6-stack it means it would never get taken right? The true occupation of a tile (A minimum living group). I was trying to find the 2-eyes living group equivalent in this game, and what’s the fighting sequence, and contact first seems to gain an advantage every time when it has a higher line-of-sight first.

And from an end of game positions point of view, working backward, where every tile is at its maximum height, Any stack can only be reinforced to up to the same light-of-sight tiles right? Any convex corner will eventually be taken, only parallel borders are truly “settled”. I am not sure if some of the end-games I saw are truly ended. Must be a way to estimate the true end-of-game tiles.


Hi Claire, you may be right, and you may be wrong, but in my experience after playing many games of Tumbleweed, simple mirror strategy and “get the center” approach does not really work to win the game. There are too many complexities arising during the middle game involving fighting. Winning the center might lead one to have a strong influence on the board, but such configurations can also get attacked, surrounded, and become weak.

A 6-stack can never be captured, that is correct. But you are incorrect that convex corners “will eventually be taken”. A 3-stack on a blunt-ended (120 degree angle) corner is safe and alive. A 4-stack on a sharp (60 degree angle) convex corner is alive. a 2-stack on a parallel border is alive.

Many new Tumbleweed players think that capturing the center piece is the sure way to victory, but change their minds after playing a few actual games. I encourage you to try out the game for yourself to experience the emergent complexity!

I made a video to teach people the rules of this game.



The second bi-annual Tumbleweed open tournament will happen in April 2021.
Anyone reading this who is interested to join, dm me here on OGS!
The tournament will be played on iggamecenter.com
Prizes - $100 for best new player, $50 for second best new player, $20 for a random new player
(new player meaning anybody who didn’t play in the first tournament)
4 round Swiss system tournament + playoffs to tiebreak winners.
Board size 8 cells per side.
Time control 30 min + 30 sec/move
Registration date deadline 1 of April.


Considering getting into this game besides merely joining the server (I’m not even on discord now).

I just watched the championship final so that got me a bit energised~

My main non-Go non-chess game project is to improve my absolutely atrocious level of Arimaa, though.

  • Tumbleweed
  • ACNOS (Alek’s own game)
  • some music

L19 discussion of “Stigmergy”, a game based on Tumbleweed


Thanks Bugcat, I just posted this over at L19 and let’s see how people react!

1 Like

Did @le_4TC make some tumbleweed tool?

I did! :slightly_smiling_face:


I recognised the domain on the tool they were using for the finals commentary in the link @bugcat posted.

It looks cool :slight_smile:

1 Like

Tumbleweed is a very young game and already heavily associated with the Go community and OGS.

That, happily, makes all of us very well-placed to get into Tumbleweed if we want.


I could probably play a few games of it if you want?

I lost to a bot fairly easily, it only plays up to 6x6x6 it says.