It’s a subject I’ve broached before, though largely unheeded, about Stockfish 8 sometimes errant behavior.
Consider this comment by <chessgames> on CG for http://www.chessgames.com/perl/chessgame?gid=1899418:
||chessgames.com: We stuck Stockfish on 21.? for 5 hours to see if it could come up with 21.Bg5!!The answer is “no” — analysis of move 21.?|
First of all, I wish CG would have used the damn round numbers I provided so I could properly refer to which game in the match this was (using the game sequence number in the published paper). That aside, here’s a run on SF8 of about ~ 1minute (or less) where 21.Bg5 is the leading candidate move:
I’m too lazy, at the moment, to do this – but the interested can find an indication of why this could happen in one of my previous Stockfish blogs – just do a quick search on WordPress. I can’t explain at the “molecular” level what is going on, but I’ve been aware of this SF8 “flaw” for awhile.
(mis-)Trust, but verify.