Author: Vincent Diepeveen
Date: 15:18:41 08/02/02
Go up one level in this thread
On July 31, 2002 at 18:51:51, Dann Corbit wrote: >On July 31, 2002 at 18:35:54, Vincent Diepeveen wrote: > >>On July 31, 2002 at 16:16:32, Dann Corbit wrote: >> >>how can you get only 14 ply with crafty after an hour. this is >>a version with SE? > >No, just a slow machine with only one CPU. > >If it is a huge passed pawn bonus, then I would expect Junior to make strange >moves at times. How is it that the general play is excellent? There must be Yes it does. Also it misses wins. Take for example the weird Ba8? move against DIEP. That was real bad. Also the winning of the queen. They could have won material. Also it took the passer of diep just too soon on c6. I'm not so sure the position was a draw at all. Only after taking i got some hope. My score jumped up 2 pawns. They make horrible mistakes allright. How they win world titles is always a major mystery to me! Usually they play in a way that in the position that gets on the board, that opponents make even bigger mistakes there, that's the only possible explanation i have! >some heuristic to decide when to employ the bonus and when it is a bad idea or >the overall play would be damaged. It definitely is damaging, no doubts here. >In any case, I think it is the most impressive thing I ever saw a computer do. I saw Chess System Tal give away a rook in world champs 1997 against Nimzo (if i remember well). Very impressive. Regrettably it didn't work so a few moves later CST resigned being a rook down. >I can make crafty solve WAC.230 by giving a huge passed pawn bonus. But then it >will play badly for many other positions and over all be weakened by a large >amount. This was (apparently) the production version of Junior -- at least they >were confident enough to run it with the settings that they chose. You exactly mention why everyone is always so amazed that junior wins games anyway. It plays in a very risky way, sometimes it loses real badly. If it loses it looks worlds worst program. If it wins, it looks worlds best program. 2 championships they played consequent now and won all games by major swindling. Somewhere this is real cool, because it shows that computerchess has a long way to go still, that dubious moves can win major points. >For sure, a wonderful achievement. That game (not that I have reviewed it) is >the very sort of reason why I like computer chess a lot. They can do >breathtaking things, like GMs do, from time to time. Oh in this position any FM would take on c4, but not all would also realize they easily win with it here. Best regards, Vincent
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.