Author: Robert Hyatt
Date: 10:26:48 11/12/00
Go up one level in this thread
On November 12, 2000 at 12:26:15, Jeremiah Penery wrote: >On November 12, 2000 at 11:05:30, Robert Hyatt wrote: > >>case 2: target search time is 3 minutes. The opponent takes 3 minutes to >>move. If we take the best two moves he could make and search them equally, >>when he moves, and assuming he makes one of those two moves, we would have >>spent only 1/2 of our target search time on the move he played. We have to >>search another minute and 1/2. If we predict correctly, we save 50%. If >>we predict incorrectly we save nothing. Just using two moves should move >>our correct prediction rate up, but _not_ to 100%. Which means this will never >>save as much time as case 1. If you use 3 moves to increase the accuracy of >>choosing the best move correctly, you only spend 1/3 of the target on each >>move and it gets worse. And worse.. > >It doesn't seem to me that you would have to spend equal time searching each >move. Once you search the first move, either the others can be proved inferior >quickly, or a different move can become the best move, just like in the normal >search. Check out the "lost ply" you get here. In 3 minutes, I can search to depth N assuming the opponent plays a move I think is "best". Or I can search to depth N from his side of the board, which means _my_ move will be made after a depth N-1 search...
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.