Author: Louis Fagliano
Date: 08:31:36 02/06/03
Go up one level in this thread
On February 06, 2003 at 10:21:20, Uri Blass wrote: >On February 06, 2003 at 10:06:26, Louis Fagliano wrote: > >>Junior made a what was thought to be a spectacular attacking sacrifice but which >>in the end turned out to be merely a drawing sacrifice. However, it was playing >>Black. Thus, achieving a short draw with the Black pieces in a match is >>considered good since having White in a match is equivalent to having the serve >>in tennis and drawing with Black might be considered breaking your opponents >>serve. >> >>But what if Junior had White played a similar sacrifice only to end up with a >>short draw? It would have been considered stupid, right? Throwing away a >>chance to win with White? Yet I don’t know if the program distinguishes between >>having White or Black when considering a speculative sacrifice such as that one. >> >>Is there an algorithm in the program to make it value lees an early, just out of >>the opening drawing sacrifice when it has White then when it has Black? Should >>such a factor be added to it? And are there any other programs that have that >>feature? > >No need for a new algorithm > >If the computer see negative evaluation it will go for a draw and if it is >positive it will not go for a draw. >If the computer is black it will see negative evaluation in more cases. > >Uri Ah, but was Junior really "going for a draw" when playing the sacrifice on h2 or did it play it because it thought that was objectively the best move regardless of whether it was Black or White? It all depends on how much time it has to analyize. If you let Junior run for 24 hours on the 10th move it maybe would have enough time to analyize deeply enough to find out that it's only a draw after 10... Bxh2+ and then choose something else. With the limited time avalable on a 40/120 time control it probably couldn't analyize conclusively but whatever current depth it was on it showed that that move (10... Bxh2+) was the best it could have objectively played. The point is in a similar situation with the White pieces it would play, let us say, a sacrifice on h7 (without being able to conclusively see the draw because it didn't get deep enough into the position due to time constraints) only to find out it had stumbled into a draw and threw away a chance with White. That's why it think such an algorithm as described above may be necessary.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.