Author: Tony Hedlund
Date: 07:58:54 02/12/99
Go up one level in this thread
On February 12, 1999 at 10:19:27, Robert Hyatt wrote: >On February 12, 1999 at 02:27:13, David Blackman wrote: > >>On February 12, 1999 at 01:13:44, Stuart Cracraft wrote: >> >>>Fellow Chesspians, >>> >>>I am curious what your result was in implementing >>>thinking on opponent's time in your programs. >>> >>>Any comments on what you saw in terms of performance/strength >>>as a result of, if any, would be helpful. >>> >>>For example, how often did it guess the opponent's move? >> >>About 50% for any reasonable opponent. >> >>>How much "free" time did it get with this feature? >> >>I guess it's about an extra 50%. More if you have to operate >>the program by hand against a human opponent. >> >>>How much more strongly, if at all, do you think it played? >> >>Objectively, not a lot. Better i think, but not really >>enough difference to notice much. But against a human opponent >>it has a great practical advantage. Making reasonably good >>moves instantly reduces the opponents thinking time and >>"upsets their rhythm". >> > >it has a definite strength advantage. Because if you predict well, and >50% of your moves are 'zero time moves' then the program has become 2x >faster, because it gets twice as long to search. And 2x is a measurable >speed increase... About 70 ELO points. Tony >>>Thanks, >>> >>>--Stuart >>> >>>P.S. Yes, I have already implemented it. No, I have not tested it >>>in practical play against others.
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.