Crafty To 27+ Ply

Discussion of chess software programming and technical issues.

Moderators: hgm, Rebel, chrisw

bob
Posts: 20943
Joined: Mon Feb 27, 2006 7:30 pm
Location: Birmingham, AL

Re: Crafty To 27+ Ply

Post by bob »

Eelco de Groot wrote:Thanks for posting your examples Dann! I still think it is pretty unique to see c4 even if Cray Blitz already could do that a long time before Ivanhoe. I suppose you could write some special rules for a program to play it but this is just with the general search. I don't think Rybka would play it, because it also has a habit of placing Knights in front of pawns, and I do believe there was a discussion at the time that Larry was still tweaking the eval of Rybka, that he could not get the engine to play c4, not without hurting elo I suppose.

It is a pity there are no real supercomputer chess-programs like Cray Blitz active anymore, it would be fun to compare programs on such big hardware with machines that almost everyone can buy these days. Maybe the large number of processors in the big mainframes used for floating point simulations, can not really be effectively programmed to play chess. The transposition tables play such a big role in search efficiency and building up the search trees.

Eelco
Crafty and CB both know to not block the c-pawn if the d-pawn has been pushed first and the e-pawn is not on e4. That doesn't necessarily make them want to play c4, but it can help since c4 also allows Nc3 which most programs really want to play early (develop knights before bishops as most books suggest, etc.)