Some Mysterious Bug in Stockfish

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

abgursu
Posts: 92
Joined: Thu May 14, 2020 3:34 pm
Full name: A. B. Gursu

Re: Some Mysterious Bug in Stockfish

Post by abgursu »

Marek Soszynski wrote: Thu May 11, 2023 8:07 am I think one or two of us misunderstood your point, which isn't why doesn't Stockfish choose 37.Bxe6 sooner but why doesn't Stockfish reject 36...Qa7 sooner.
Exactly!
syzygy
Posts: 5805
Joined: Tue Feb 28, 2012 11:56 pm

Re: Some Mysterious Bug in Stockfish

Post by syzygy »

abgursu wrote: Thu May 11, 2023 9:38 am
Marek Soszynski wrote: Thu May 11, 2023 8:07 am I think one or two of us misunderstood your point, which isn't why doesn't Stockfish choose 37.Bxe6 sooner but why doesn't Stockfish reject 36...Qa7 sooner.
Exactly!
And the reason is because it thinks the pawns on the a and b files supported by the queen will be strong enough. But there it miscalculates.

In many other positions SF will get it right and win the game by playing a move that a human would consider too risky.

SF does not look at the position and think "I can have no more than a draw anyway, and Qa7 might lead to a variation that I overlooked, so I should not play it and instead pick a safer move". It would be nice if it did, but alpha-beta search does not work that way.