Author: Richard Pijl
Date: 00:59:07 11/25/03
Go up one level in this thread
I can understand the wish of banning openingbooks if you're looking for artificial intelligence in chess. I think having prepared openingbooks and endgame databases could be considered kind of cheating then. Instead of solving the puzzle on its own, the engine looks up the answer in a database. This would also be true for position 'learning' mechanisms that in fact is storing precomputed results for (opening) positions in a database. In fact, these learning mechanisms also create a sort of opening books. So multiple mechanisms commonly used today should be illegal when following that reasoning. Another way of looking at is would be to reduce the human influence and allow precomputed tables (i.e. openingbooks, position learning, endgame tablebases) as long as they are generated by either public available means (egtb's) or are generated by the engine itself, as that should put the engine knowledge in a central spot. Unfortunately this type of rules would be very vulnarable to cheating as is could be hard to check whether a precomputed result is really generated by the engine, or by human intervention to select the results to store in the precomputed tables. An engine could have a preference to reach a position that is in the precomputed table. Selecting those positions by hand would effectively create an opening book. In fact, the opening book of my engine the Baron (hand-cooked) is already resembling that format. It would be very easy to make it appear like a position learning file ... So, you should probably go all the way (forbid all precomputed results in any form, also the hardcoded variants without external files) or allow everything (that is legal, e.g. considering copyrights), like it is now. bye, Richard.
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.