What Is Freestyle Chess?
Freestyle chess is a chess variant, which just means a different way of playing a chess game, where pieces usually move the same way as normal, but other aspects of the game are changed. It goes by two other, older names: Fischer Random, after its inventor GM Bobby Fischer, and Chess960, after the 960 possible starting positions in the variant.
The pieces are placed randomly on the first and eighth ranks, Black’s pieces mirroring White’s, with only two restrictions: the kingmust be placed between both rooks, and the bishops must be on opposite-color squares. The queen and knights can occupy any starting square, and pawns are placed on the second and seventh ranks as in a normal game.
In Freestyle Chess, the starting position is random but still symmetric, i.e. both sides are working with the same position.
Perhaps this principle could also work for the go equivalent? I.e. generate a random position on one half of the board, and mirror it to the other half (inverting the colours).
That might prevent starting positions with a high advantage for one side, reducing the need for AI komi.
I recently played a game like this. Began with five randomly placed stones, radially mirrored. It was great:
ChatGPT prompt:
Generate five random pairs of numbers between 1 and 19, then replace the first number in each pair with the corresponding letter of the alphabet, skipping ‘I’ so that after ‘H’, ‘J’ is used for 9 and continuing accordingly.
You can believe in the fairness of a not symetrical position evaluated as fair by a 9d bot.
A big fun is that those first moves chosen by the bot may be not optimum, as long as the final position is evaluated as fair, so we could start with very different opening as we use to have. ( Like stones away from the corners, on the second line or somewhere more in the center)
Shouldn’t be difficult to ask the bot to create a fair board position with 10 moves never put on the third and 4th lines for example.
Considering that even if being fair a position may seem to be harder for one side to play with in a human perspective, it could be interesting to reintroduce a betted komi (one fix a komi, the other one chose the color)
I also suggested something like this before and I think this should be interesting. The stones don’t need to be totally random and can follow certain logic like that in the Fischer random chess.
If you take the suggested procedure (forbid 3d and 4th line for some moves) and still ask the bot to play optimally, you’ll surely get some logic in its choices (not saying it’s the best way to get something most interesting, should be tested)
That would look just too much like a normal game. Would be interesting to see how you take care of the still opened corners in accordance to the stones already on the board, or going to some advantages on sides or center … By breaking the priority corners/sides/center that could offer a very new experience.
I m already very curious to see how a bot would play. Will he prepare for some huge corner enclosure, build something oriented to the center, consider some fortress on the sides, or something else? From which point will it not just jump into a corner after getting the possibility and before, how will it proceed? This looks really entertaining to me (and fruitful as a way of exploring the game)
I think that’s also possible that’s really random random. My idea was just to use a random opening that is less explored and see how the modern players will play it out. Similar to how the Fischer Random Chess is still using a configuration that is considered normal to both sides.
, although the second quote is a strong argument.