Author: Dave Gomboc
Date: 16:03:13 02/01/00
Go up one level in this thread
On February 01, 2000 at 18:58:16, Dann Corbit wrote: >On February 01, 2000 at 18:32:11, Côme wrote: >>On February 01, 2000 at 18:14:08, Dann Corbit wrote: >>>On February 01, 2000 at 18:02:16, Côme wrote: >>>[snip] >>>>But It doesn't explain why crafty don't find this winning move! >>> >>>The others found it by accident, or by a different positional evaluation. >>>For positions where the answer is not clear, which programs solve it is sort of >>>a random number function. The next iteration of the same program may choose a >>>different move. >>> >>>I am not convinced by any answer that does not have an evaluation which shows >>>that it knows why it picked the move that it did. >>> >>>I would be very curious to find a complete list of programs that do solve it >>>*AND* a comment by the programmer to explain how it was chosen. >>> >>>Actually, the list has no value. We already know what the answer is. But the >>>comments would have great value. >> >>I agree with you Dann but Junior 6 show a great advantage for white ! > >I would not be surprised if Junior 6 chose the move for all the right reasons. >However, a +1 pawn evaluation is hardly convincing. In fact, I see computers >think that they are up one pawn in gambits all the time, only do discover they >are about to get the tar smacked out of themselves. > >Since Junior seems to be at or among the strongest computer programs in the >world at tournament level time controls against other computers, that indicates >(to me) enormous tactical ability. Therefore, I suspect that Junior "knows what >it is doing." > >On the other hand, it may be accidental. We will never know unless Amir cares >to tell us what is going on with this eval. What is the expected score? Free piece or something? If Junior is evalling Bg5 at +1.0, and other programs are evalling it at 0.0, I doubt it has much to do with accident. Dave
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.