Author: Jeremiah Penery
Date: 01:02:37 02/25/00
Go up one level in this thread
On February 25, 2000 at 03:14:46, Tom Kerrigan wrote: >Just glancing at this position, it shouldn't be hard for programs to search >15-20 ply, even with blitz time controls. True. My modified Crafty was getting 30+ ply in a few seconds. But even 50 ply searches won't do much in this position. But playing it out, I think the programs will stumble upon the solution. Around move 65-70, mine finally found a large + score for white. If I'd let it search longer (up to the maximum 60 ply), it may have found this around move 60 or slightly earlier. >Hash tables will probably work miracles here... A human can find the entire ~70 move variation in a few seconds, because he realizes that black will run out of moves. You'd think that hash tables would allow the computer to evaluate the position when black has already exhausted his pawn moves (which would be a large + score for white), and then just find a way to reach that position. But a computer can't generalize like the human to say "I need to repeat this sequence x times, where x is the number of black pawn moves". It can calculate itself doing the variation only y times, where y is the search depth/the length of the sequence. Since after this time it doesn't realize what's happening, hash tables don't really show a noticeable affect here until much later down the line, where the search depth will see the answer already. :|
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.