Author: Bob Durrett
Date: 18:08:09 11/08/00
Go up one level in this thread
On November 08, 2000 at 19:10:12, Simon Waters wrote: <snip> >Of course it is possible that the program has learnt from it's own experience, >or had it's weekly dose of latest GM games injected into it. <snip> Following this logic . . . So, if some of the computer programs, such as Fritz, can learn from experience in the middlegame [as opposed to during the phase where the program is still using it's opening book], then it may be that no two Fritz programs are the same, after they have been used awhile. [Assuming "learning in the middlegame" has not been disabled somehow.] In other words, if 100,000 copies of Fritz 6.0 are out there being used by chessplayers, then it would be likely that no two of these computers would be exactly alike. This would follow from the improbability that any two users would have played exactly the same lines, taking exactly the same amount of time at each move, etc. Is that a fair assessment? If it is, then it might be impossible for a GM to prepare for a match with the computer program since what was "learned" in the middlegame would, in all likelihood, be different from what that GM's home computer program would have "learned in the middlegame." Do some of the programs out there learn in the middlegame [after out of opening book]??????? And, if so, is this a big determinant of the program's playing style and strength? If so, maybe we should try to purchase copies of the programs which had been used awhile in the home of one of the top GMs. They might have learned something not in the original program!
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.