Author: Darrel Briley
Date: 20:14:57 05/16/05
Go up one level in this thread
On May 16, 2005 at 13:45:52, stuart taylor wrote: >On May 16, 2005 at 13:26:40, Uri Blass wrote: > >>On May 16, 2005 at 13:06:16, stuart taylor wrote: >> >>>On May 15, 2005 at 16:04:25, gerold daniels wrote: >>> >>>>On May 15, 2005 at 11:25:12, Uri Blass wrote: >>>> >>>>>On May 15, 2005 at 10:44:49, stuart taylor wrote: >>>>> >>>>>>On May 15, 2005 at 09:13:24, Klaus Wlotzka wrote: >>>>>> >>>>>>How can this be? Is this really true? What's the catch? >>>>>>Maybe we are speaking only of Blitz? Maybe some other top engines were absent? >>>>>>Or what? >>>>>>S.Taylor >>>>> >>>>>Time control is 10 minutes per game+10 seconds per move and the hardware is >>>>>more than twice faster than ssdf hardware. >>>>> >>>>>It seems to be about 3 times faster time control than the ssdf games. >>>>> >>>>>I do not know the rating of Shredder9 at 120/40 and I cannot say that it is >>>>>impossible that it is also 100 elo better at 120/40 time control. >>>>> >>>>>The fact that it had problems against Junior proves nothing because a program >>>>>may have problems against one opponent. >>>>> >>>>>Uri >>>> >>>>good afternoon Uri. Sedat has shredder 9 104 points above junior 9 in his latest >>>>tourney. 60 min.10 sec. >>>> >>>>gerold. >>> >>>THAT wasn't being disputed! >>>An actual match between the two is hard for Shredder 9. (Even though Shredder >>>7.04 crushes Junior 9). >>>S.Taylor >> >>The results of the ssdf do not prove that Shredder7.04 does better against >>Junior reltive to Shredder9. >>The main problem is that a match is not undependent games. >> >>I did not analyze what happened. >>The problem is that it is possible that shredder7.04 was lucky to find some >>weakness in Junior in the first games and also found weakness in the learning of >>Junior so it could beat Junior easily when Shredder9 may be unlucky. >> >>Without looking at the game and understanding why Shredder7.04 did better you >>can say nothing. >> >> >>Here is a possible explanation(I did not look at the games and did not try to >>analyze the reason for the better result of Shredder7.04): >> >> >>It is possible that shredder9 started with 1.f4 in the first game with white and >>lost so it never repeated 1.f4 >>It is possible that the reason was some line that Junior play normally with >>probability of 1%. >>It is possible that Shredder7.04 started with 9 wins in with 1.f4 and one >>loss(in the same line that Shredder9 also lost) and the single loss did not >>convince shredder7.04 to avoid 1.f4 and it simply chose a different line with >>1.f4 but a different move later. >> >>Uri > >I looked very carefully, not at the games themselves, but at many sets of games >between the two (or three!), and many different testings in different formats, >which have been posted ever since Junior 9 came out. You will find this ro be >ALWAYS (or practically always) the case. I asked many times how this is, and I >never got answers, except perhaps the usual stuff....."not enough games". I >think most people ignored this interesting fact. >S.Taylor Despite the results of the SSDF Shredder9/Junior9 match, testing done on my own machine (Athlon 2600+ Barton Core, 1GB RAM, Hash 256 for each engine, own books 120/40 30 minutes for remaining moves, so far a total of 137 games, shows Shredder9 UCI to be clearly stronger than Junior 9. I don't know what to attribute the difference to, other than the hardware, and the larger number of games.
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.