Author: Harald Faber
Date: 21:25:05 05/14/02
Go up one level in this thread
On May 14, 2002 at 20:38:16, P. Massie wrote: >Since I have a copy of Hiarcs8, I decided to review the opening book a bit. >There is definitely a problem of some sort in the book. The entries in the book >look quite reasonable, as do the results. The problem lies with the "N" column, >which represents the number of games played with that move. This number >sometimes looks quite reasonable, and at other times is clearly quite wrong. > >In the following case, the book follows approximately 4,000 entries to the Black >response to move 12. 12 ... Rd8 gets the majority of those, but once you go >down another level after 12 ... hg5 you find the number is again back up to >4,000 for that branch alone. And that is obviously wrong. I went through Mega >2002 + MegaCorr and could only find four cases where this move (12 ... hg5) had >been played, so clearly the "N" column in the Hiarcs book is wrong. And since >that provides a weighting factor, the moves played frequently become moves that >should never be played. > >Paul If you take a look at the Hiarcs 7.32 opening book you will see the same. There you have values/n like 1024, 4096 etc. This is due to convertion from the old DOS book format and no bug.
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.