It would be nice to create studies based on a specific position for the game Go.
FEN is not really "native" to go, however, and I am not quite sure how to understand it.

Trying to "decode" how to write a valid FEN this afternoon left me puzzled and perplexed:

Example from a game:

empty board - FEN
9/9/9/9/9/9/9/9/9[SSSSSSSSSSssssssssss] b - 0 865 0 0 55 0 1

after first black move (black moves first, for the uninitiated):
9/9/9/9/5S3/9/9/9/9[SSSSSSSSSSssssssssss] w - 810 55 0 0 55 0 1

after first white move:
9/9/9/9/3s1S3/9/9/9/9[SSSSSSSSSSssssssssss] b - 10 65 0 0 55 0 2

next: 9/9/2S6/9/3s1S3/9/9/9/9[SSSSSSSSSSssssssssss] w - 20 65 0 0 55 0 2

next: 9/3s5/2S6/9/3s1S3/9/9/9/9[SSSSSSSSSSssssssssss] b - 20 75 0 0 55 0 3

next: 9/3s5/2SS5/9/3s1S3/9/9/9/9[SSSSSSSSSSssssssssss] w - 30 75 0 0 55 0 3

By trial and error, I see that the part [SSSSSSSSSSssssssssss] seems optional.

Certain numbers (the first after the -) and the last number seem to correlate with the number of full moves.

I can't quite grasp those 65, 75, etc that appear as second and fifth number

I wouldn't want that only a "valid state" is accepted, that is, can only be reached by playing an equal number of stones with each colour (plus or minus one), and even then given that players may pass, any arbitrary difference in number of stones should be allowed.

Is there any documentation? Otherwise very difficult to create example studies with only partial board position.

1

We are hoping to support setup from position for all games and allow certain input formats for the analysis tool etc.
For now info about how the fen is generated is coded within our strategygames repo (github.com/Mind-Sports-Games/strategygames/).

For Go specifically the fields are:
s"${board}${pocket} ${turn} ${ko} ${p1FenScore} ${p2FenScore} {p1Captures} {p2Captures} ${fenKomi} ${passCount} ${fullMoveStr}"

The scores/komi are multiplied by 10 so that there is no decimal.

1

Ah, I see! Thanks for your reply.

If the Analysis board then gave the option to pass (Pass button? maybe I could not find it) we would have this immediately solved. I can open a FR to the strategygames repo if it can help?

With regards to the solution of the study/chapter, I have an additional observation. In the study tool/octopus, I see also that you are immediately rewarded with a "Congratulations!" if you guess the correct move (which is nice!), but you can enter a longer sequence, and the player has to keep guessing at every step. In other words, it seems more geared towards a "1-move answer". For Go, that is only educational for the simplest beginner-level "capture the stone" type of problems.

This is because individual stones are more like words in a sentence, and while a "yes/no" refutation is sometimes possible, a sequence/sentence is more common.

If it's not too much to ask, is it possible to have a guided, or longer, solution sequence? Maybe this is already possible, and I am merely showing my inexperience with the tool!

In books, sequences are typically represented "kifu style" with a single "result diagram" that essentially shows a position 5-10 moves later - the trick being that the move sequence is written inside the stone, a bit like this, sometimes with even additional symbols like circles, triangles, or squares, as well as letters and numbers:
senseis.xmp.net/diagrams/23/ba5361062673765f0ff32ac61eb88f65.png

1

Ah I now see that one can add to the Octopus both exposition and entire sequences, e.g. playstrategy.org/study/FMcKsiE2, so I'll try to do something similar.

Please disregard the second part of the message above!

It would still be neat to have a Pass button in the Analysis, though...

1

Thanks for the messages.
Yes we want to fully support go in the analysis board, this requires adding a pass button, and deadstone calculation/removal. This is on our list of items to fix but unfortunately isn't that high up on our priorities.

1

I think I am getting the hang of it! I made some progress - the trick seems to be "balancing the equation", for example, I can add 3 white stones - as long as I remember adding 30 points to the ${p2FenScore}. Or remove 1 black stone as long as I remove 10 points from ${p1FenScore} as well! A bit like balancing a chemical formula!

1