Does Stockfish really need syzygy rtbz-files at all?

Discussion of chess software programming and technical issues.

Moderators: hgm, Rebel, chrisw

Jouni
Posts: 3291
Joined: Wed Mar 08, 2006 8:15 pm

Re: Does Stockfish really need syzygy rtbz-files at all?

Post by Jouni »

Yes but I have set no 50 move syzygy draw flag. It doesn't help if GUI decides the result?
Jouni
syzygy
Posts: 5566
Joined: Tue Feb 28, 2012 11:56 pm

Re: Does Stockfish really need syzygy rtbz-files at all?

Post by syzygy »

Jouni wrote:Yes but I have set no 50 move syzygy draw flag. It doesn't help if GUI decides the result?
As you say, the GUI decides the result. If the engine ignores the 50-move rule, it fools itself into thinking the rules of chess are different than they are, but that does not change the rules. It is like telling yourself that traffic signs have different meanings than the law says they have and then finding out that the police does not agree.

But even if the GUI ignored the 50-move rule, there is a good chance that the engines would stop moving once the 50-move rule has been reached. I'm not entirely sure what would happen. Syzygy50MoveRule only affects TB probing, not the rest of the engine.

Enabling the Syzygy50MoveRule option will make SF internally evaluate TB positions that are "cursed wins" as wins. So if SF's search finds a forced way to reach such a 5-piece position from, say, a 7-piece position, then SF will do that. But there is no guarantee whatsoever that SF will manage to win it. (There is a guarantee that SF won't win it if the opponent has WDL+DTZ.)

The same happens if the engine uses Nalimov DTM tables: it will think a mate in 127 moves is winning and then draw by the 50-move rule.