Blindfold Go Stuff

Played a couple more games (most recent is dexonsmith vs. Amaranthus).

Some thoughts:

  • Be really careful to play “Unranked” games, especially if you’re using a standard board size (non-standard doesn’t let you make a mistake). If you screw that up, you should probably ask for the game to be annulled, since otherwise it will wreak havoc on the ratings system.
  • If you have “double-click to move” or “submit button to move” on, you can click once to see if the square “lights up” with the about-to-move indicator as a way of “sweeping” for stones. This defeats the purpose… I’ve been trying to double-click quickly (since I don’t want to change my default for other games).
  • When there are captures, the player cards tell you, so this can be confirming/disconfirming evidence of the board state (especially if you see it change at an unexpected time). If you’d rather do this in your head, don’t look :).
  • This is really hard! I’ve tried a few 7x7 games but might return to 5x5 for a bit (even though the tiny board is a bit tedious).
  • This feels like a really useful thing for me to practice. The fundamental skill here is tracking board state without seeing it in front of you, which I think must have significant crossovers with reading skills.

That was surprisingly fun !

Tried 3 blindfolded games on 9x9 with Amaranthus.

After thinking the first game’s yose throw-in was atari but doubting that, and thinking I was hallucinating, so tenukiing, I somewhat lost the thread of the game & the tsumego shape in the upper right, which became very interesting.

The second game was much better and I could visualise the shapes somewhat better throughout the game. (although I still played a bit passively on the lower side when I didn’t need to respond, in case I missed or hallucinated anything later, rather than aiming to kill the top right or noticing the tsumego shape immediately :sweat_smile: )

In the third game, I definitely felt an improvement and could hold a lot of the boundaries and shapes, although the odd plays by Amaranthus and clump of stones were a bit hard to hold in my mind as an entirety as the clearly (and oddly) eyeless clump increased, and make sense of. (which made it more fun to try, too)

There was aji in the lower right which I didn’t re-read (a ko) properly when playing D2, (an ''interesting" move leaving aji I wanted to try to see whether I had missed anything!) rather than playing from the other side or descending at C2, and I can notice that practising blindfolded on the relatively small 9x9 would actually have a lot of interesting whole board things to hold in mind.

(one part of the board affecting another, which are probably the types of aji very relevant in real games)

They are intuitively there for me, but weren’t clear when holding the positions in mind and playing relatively quickly in these games, which is very interesting to notice!

I can feel this would definitely help improve visualisation and shape understanding in many ways, and look forward to trying more!

Anyone interested in trying a blindfold game together? (@dexonsmith ?) (smaller board sizes like 5x5 or 7x7 are fine too, I think it would be fun to try with another (human) player :slight_smile: )


I’d love to play sometime, I’m fine with 9x9 or 11x11; 13x13 would be stretching myself, but there’s only one way to learn :slight_smile:

1 Like

There is alternative of how to learn Blindfold Go. 9x9 is not needed. You can start on 19x19.
Just make stones slightly transparent. After each win make them more transparent. After each lose make them less transparent. And so until they are 100% transparent.
Also, last move indicator may be gray for both.

bonus: one color Go without last move indicator.
Make black and white stone slightly more gray…
And so until both are equally gray.

Have people tried this? My intuition is that for the vast majority of the opacity range, there will be little to no difference, and then it will hit a point where you can’t see anything consistently.

1 Like

change as slow as possible

I suppose you could do “the last N moves are visible, older moves are invisible” and increase or decrease N.


That might work

Make older moves gradually less visible, not instantly. And change speed of decrease of visibility.
But thats not available on OGS.


We should prioritize the swap the board in angerness first

That would be a nice QoL improvement to the last N stones idea. Is there a known model for how human foreground-background discrimination deteriorates with respect to opacity? Can Lab be used for this purpose?

Maybe Lab could be used by finding the points of the foreground and background, and interpolating evenly spaced (within the model) points? Would this result in perceptually uniform decrease in opacity?

What is Lab?

My guess is this: CIELAB color space - Wikipedia

(be careful, there is a very deep rabbit hole there)


Yep, what gennan said

1 Like

Another idea:
When opponent makes a move, board appears for short time and then disappears.
That way last move indicator is not needed.
Goal is to make that time shorter.

But it would be hard to not learn Blitz instead.