Author: Ed Schröder
Date: 23:01:22 09/23/98
Go up one level in this thread
>My position, in short, was that I thought that these guys should send >auto232-enabled versions to the Swedish folks, but not to the public. This >would prevent them from autotesting against each other, and would prevent >end-users from autotesting, but it would also keep people from getting an >advantage by preventing others from autotesting against them while allowing >themselves to autotest against others. >bruce I see the following disadvantages: #1. Progress on many chess engines will slow down. Auto232 in an easy way produces hundreds of games. These games are studied and weak points (patterns) of an engine become visible ready for improvement. I know it helped Rebel a lot to make progress. #2. Progress on the opening book will slow down. Auto232 and its easy way to produce hundreds of automatic games shows the many and many holes in opening books which then can be fixed, so opening books improve. IMO Auto232 through the years has been responsible for good progress on many engines but also on opening books. Whatever your thoughts are on "book cooking" or "book tuning" (which I consider as 2 different things) Auto232 has forced programmers to use wide and good opening books, next "learners" and "book learners" were added to deal with these kind of things, meaning progress. I know you have ICC as your source for improvement. But not all of us are able or willing to join ICC as a new source of information. But if auto232 is going to die because of the ongoing discussion I will certainly re-order my thoughts about ICC because I need input to improve my chess program and the more input the better. - Ed -
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.