Anti-chess

Is there any computer game in which both engines tried to play the worst possible move? I wonder how would this game look like.

Also, could anti-chess become another variant, like trying somehow to force the opponent to mate our king?

I think anti-chess would be very rich tactically, because both opponents will try to expose the their king and not capture material, so the game would always have enough complexity, and probably end with an anti-zugzwang.

I’m actually excellent at this variant of chess. I’m glad to see it gaining recognition within the chess community!

@1

Anti-chess has been solved
https://magma.maths.usyd.edu.au/~watkins/LOSING_CHESS/LCsolved.pdf 

Amazing, thank you

Acually tho, that's a different version in which captures are compulsory, that makes it much simpler and it's no wonder that it has been solved.

I watched a funny video on Gotham Chess about a game between two ELO 100s. They were obviously trying to make the worst moves possible. Very funny game. A real blunderfest.

This is an interesting concept. As a software developer I need to look into this.
It would really be interesting if different developers wrote separate engines which made worst possible moves, then see the outcome of the engines playing each other

I would think this would have to end in stalemate, unless there were rules about no repeat moves. But, Of course, could be completely wrong. Very interesting

Since stalemate is counted as a draw, and the objective is to lose, i.e., “win” 0 points instead of 0.5, I think the anti chess engine would work diligently to avoid stalemate.

My guess is that the games would be very long, upwards of 80-100+ moves. Also, I think the engine would keep an eye out for and avoid three fold repetition, etc.

I don't know how likely a stalemate can be in a game where both players try to not capture material. With a lot of material on the board there will always be legal moves. A stalemate might be the outcome of a long game, in which players manages to force each other into captures.

Are new engines actually needed?

What normal engines do is they evaluate the position and make the best move. The evaluation process might be very similar if not the same, the difference is that they chose the worst move instead of the best

That seems cool 😎

“Are new engines actually needed?”

If we can convince AlphaZero, Stockfish, etc., to get involved with this concept, then no, we wouldn’t need new engines. Otherwise it’s up to us.

“What normal engines do is they evaluate the position and make the best move. The evaluation process might be very similar if not the same, the difference is that they chose the worst move instead of the best”

Correct. I’ve started a code project implementing this. I’m making it expose the Universal Chess Interface (UCI) and building it as a DLL to facilitate pairing it with other engines within a common graphic interface.

Also, I’m thinking the experiment would be a lot more fun on Chess960 rather than classic chess. This would make it more practical to pair two engines against each other in a match and not see the same game over and over (depending on which plays white or black, etc.)

 

Maybe, but that wouldn't develop the anti-chess opening theory