Author: Komputer Korner
Date: 21:01:19 05/28/98
Go up one level in this thread
On May 28, 1998 at 23:46:49, Komputer Korner wrote: >On May 28, 1998 at 14:25:19, Robert Hyatt wrote: > >>On May 28, 1998 at 10:57:11, Komputer Korner wrote: >> >>>On May 28, 1998 at 09:22:31, Steven J. Edwards wrote: >>> >>>> >>>>There is the question also on how big the library has to be. Having a >>>>million position library will cover just about all of modern GM praxis >>>>well into the middlegame. >>>> >>>>Scanning PGN files is the way to go. Years ago I had tried typing in >>>>chunks of ECO, but it's kind of a drag. The question is how deep to go >>>>in a game while recording the positions; I have this set to 48 ply in >>>>OCD which seems to be a reasonable number. >>> >>>One million is not enough to cover all of opening theory. The Fritz >>>powerbooks are 10,000,000 positions. Genius GM powerbooks are 6,000,000. >>> >>>-- >>>Komputer Korner >> >> >>that has *nothing* to do with opening theory. They are storing >>*complete* >>games... Opening theory doesn't generally extend from 1. e4 to mate. >>It >>stops after 20 move or so. the "minplay" in crafty will take an opening >>book made from 330,000 games and reduce it to 7.5 megabytes of stuff, >>because over 90% of the moves in the stuff was only played 1 time in one >>game... >> >>Most programs don't want to follow "singular" lines because you put too >>much faith in the moves played by a single player in a single game... >>and >>you get burned as a result... and with so many games, it would take a >>century of learning to disable the lemon moves one by one... > >You are right about those books but opening theory actually is much more >than 1,000,000 positions. I would say that theory is any move/line that >has been played by GM/IM titled players up to the recognized mistake. >Even if a line goes 35 moves/70 ply of theory, and it has been played >only once, if the losing move has been identified on move 36, then the >first 35 moves actually become part of opening theory because GM's need >to know the position up until the bad move because it will be played >again with another move substituted on the 36 th move. This happens time >and time again at top GM level and especially in postal chess. So >opening theory is much greater than the lines found in ECO and much >greater than most of us realize. -- Komputer Korner Some further clarification. The above 2 Powerbooks referred to are not complete games. One ends up at 29 moves and the other one ends at 31 moves. Also opening theory, it could be argued is the bad moves as well as the good ones. Seen in this light, 1,000,000 moves/positions does not cover all the theory. Sure you could make a pretty good repertoire book from this but that is all.
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.