Author: Uri Blass
Date: 02:03:48 11/09/03
Go up one level in this thread
On November 08, 2003 at 16:07:53, Peter Berger wrote: >On November 08, 2003 at 10:48:13, Drexel,Michael wrote: > >>On November 08, 2003 at 06:34:14, Peter Berger wrote: >> >>>This is the second match in (planned at least ;) ) a little series between >>>modern middle-class amateur engines and old professionals. >>>(Movei- Genius 2 1-6, 1 draw). >> >>This is a very good idea, Peter. You should continue these tests until Uri >>finally removes those unsound lines from his book. >>Poor Movei... >> >>Michael > >Hi Michael, > >I am not doing these matches to prove anything, I do it for fun. > >There is the danger that it could be mistaken as anti-movei bickering ( I know >you didn't suggest it, and Uri knows it isn't). But it could be misleading to >see several match results when movei might lose most of them. > >Movei is a reasonably strong middle-class amateur engine these days and it was >just chosen as an example for this group of engines. I thought about sending >results to Uri only, but I know some people enjoy looking at these games. > >Having a relatively weak opening book is a common problem for this group of >engines. Btw, Uri's approach has its advantages too. In the last match against >Genius it got one point for free when Genius's book had a long line that led to >an immediate loss (not because the line was bad itself ,but because it was too >complicated to find the moves once it was out of book), while Movei had no >problems to find the moves to react. > >I agree the line with Qe7 isn't to be recommended with the short line movei has >in its book, but for example in the games with white several of the unusual >lines worked quite well for it. I sent my latest movei to a beta tester who is going to generate a better book for it. I prefer not to waste time on building a book for it and only to concentrate on the engine and I also have not too much time to work on the engine because I have other work to do. I believe that genius2 is clearly better than chess system tal at this time control. It is also possible that movei's time management is better at x minutes/40 moves and I do not know if it is the case. I am slightly disappointed by the results so far because my feeling is that a previous version got better results in Leo tournament including losing 2.5-1.5 against Deep Sjeng1.5 and losing 3-1 against Aristarch4.21 and 2-2 against GreenLightchess3.0. It is possible that the book that it has was better for the version of Leo tournament and it is also possible that it does relatively better at longer time control against strong programs and it could avoid part of the losses against genius2 at 40/40. One of the problem of it is the endgame and at long time control search can help to avoid part of the mistakes. Uri
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.