Author: Danniel Corbit
Date: 11:36:17 09/28/98
Go up one level in this thread
On September 28, 1998 at 12:47:01, Danniel Corbit wrote:
>On September 28, 1998 at 01:35:45, Dave Gomboc wrote:
>[snip]
>>I should correct this slightly: even in these rather rare circumstances, there's
>>still a 17-ply or so search from where the program drops out of the opening book
>>to where it is hitting the endgame tables.
>Imagine a system where you have opening books carefully tested to a great depth
>{also analyzed to a high enough degree for confidence}. Then you take the top
>three positions[1] from the end of each book ending and analyze them carefully
>for 20 plies forward {about 3 billion positions per ending}. It's not really
>doable right now, but this is not inconceivable. Eventually, the chess program
>would play less and less calculations and do more and more database lookups
>until it had to explore that rare position not in the calculations.
[1] (1)Best eval (2) best win percent (3) largest volume of wins is one possible
determination.
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.