Author: Ron Langeveld
Date: 15:20:50 07/01/01
Go up one level in this thread
On July 01, 2001 at 18:14:11, Uri Blass wrote: >On July 01, 2001 at 17:08:08, Ron Langeveld wrote: > >>On July 01, 2001 at 15:16:38, K. Burcham wrote: >> >>> >>> >>>not what i was talking about at all. >> >>If you're talking about the average number of "blunders" that a GM of a certain >>level is allowed to make, then this also hints the gap that computers cannot >>fill to become GM. Notwithstanding the official regulations regarding GM norms, >>real GMs exhibit the behaviour to acknowledge a "blunder" and learn from it. >>They have the knowledge to see the error, if only afterwards. They can admit it >>was a mistake because the know it was one. Computers however cannot admit to a >>mistake, for they will play the same blunder again and again, just because of >>lack of knowledge. They simply don't know better. This imho sets apart the >>programs from the flesh. >> >>Ron > >There are programs that do not repeat the same opening that they lose >so they will not repeat the same blunder again and again. > >Some programs have a very small book and after 1.c3 they are out of book but >this problem can be solved by a bigger book when you need hundreds of games and >a lot of wins in order to repeat the same opening twice. > >Even if they repeat the same opening, learning by position can help them not to >repeat every blunder again and again. > >Uri Uri, please explain what you mean with "learn by position" and provide the names of the programs that have this implemented. Of course I want to see that an engine is able not to just avoid a bad opening line :) Ron
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.