´╗┐ rybka vs rybka at full preformance


Back to forum

Joseph Costello    (2008-05-25)
rybka vs rybka at full preformance

Anybody have this program working at its full capacity who has made it play itself and has it recorded you could post ?

Benjamin Block    (2008-05-25 13:44:22)

In playchess.com the computer play against each other the fastest computer+the biggest opningsbook win. But if both computer have the same book and the same fast computer it is draw.

Wolfgang Utesch    (2008-05-25 14:19:47)
Rybka vs Rybka a draw ....

... and what does it verify?

Benjamin Block    (2008-05-25 14:54:33)

I can test with my fritz 11 and it will be draw.

Thibault de Vassal    (2008-05-25 15:14:45)
Rybka vs. Rybka

Not a draw, maybe 50% but it should be about 54%

Benjamin Block    (2008-05-25 15:34:04)

Why should it not be draw? it is like playing agains youre self.

Thibault de Vassal    (2008-05-25 16:18:49)
Engine vs. Engine

A chess engine doesn't know who is its opponent, and playing against yourself does not finish in a draw in all cases.

Benjamin Block    (2008-05-25 16:22:40)
Black don┤t win

But black will not win? Only white?

Thibault de Vassal    (2008-05-25 18:33:27)

Statistics may look like CEGT or SSDF ones (ie. Rybka 2.3.2 vs. Rybka 2.3.1, with probably more draws as the engines versions are the same) but whatever the engine, it will win as White and Black sometimes.

Wayne Lowrance    (2008-05-26 05:14:25)
Rybka vs Rybka

Thibalt your wrong !. If enough games are played to get rid of the noise and is statistical relevent then it will be a draw, period Wayne

Gino Figlio    (2008-05-26 07:03:03)
not always a draw

It all depends on what the start position is and the engine settings otherwise the randomizer mode in Rybka would not work.

Thibault de Vassal    (2008-05-26 12:19:28)
Rybka vs Rybka

Ok, we all know that engines are not free will. With no randomizer, the regular start position, "normal" settings (do not avoid a draw at any price, regular values) and a large opening book [which is a quite good randomizer itself], I'm quite sure statistics after 500,000 games will look like 50% if Rybka 3 #1 & #2 play as both White & Black, 51 to 53% if Rybka 3 #1 play as White only while Rybka 3 #2 play as Black only, about the same (but not exactly) for Fritz 11 or other engines.