Author: Robert Hyatt
Date: 11:40:31 05/02/01
Go up one level in this thread
On May 01, 2001 at 16:06:55, John Merlino wrote: >On May 01, 2001 at 09:38:05, Robert Hyatt wrote: > >>On May 01, 2001 at 02:10:40, Alois Ganter wrote: >> >>>> >>>> >>>>This is why we don't autoplay _important_ matches. >>> >>>Doesn't sound encouraging for AI development: Recommending human interference. >> >>Sorry, but if you take out "human interference" you can throw all the chess >>programs right into the can. Guess how they were "developed"? >> >>Chess programs don't understand "match" yet. They will. But at present, >>there is no facility in the protocol to keep an engine up to date on a match >>and its progress. As a result, engines don't worry about it. When this >>becomes an issue, then it will be trivial to develop the code to handle it. > >I would say that, as of Game 22 in the DF-DJ match, it HAS become an issue. But, >yes, it is trivial to code. > >Most programs, at least, have an adjustable "contempt for draw" setting. But, >sadly, according to Amir, DJ's value for this was hard-coded at 0 specifically >for this match. Very bad oversight, and very sad.... > >jm I am going to propose a fix for this in the next winboard protocol version, so that the interface will tell the program the current match score in wins, losses and draws, and the number of games remaining. Then a program could easily adjust the draw score as needed to avoid draws when wins are needed, or try for draws when that will ensure a match victory...
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.