Author: Uri Blass
Date: 03:15:17 09/29/03
Go up one level in this thread
On September 29, 2003 at 05:04:07, Mike S. wrote: >On September 29, 2003 at 03:35:00, Drexel,Michael wrote: > >>On September 28, 2003 at 21:48:42, Mike S. wrote: >> >>>(...) >>>1 extra ply only will produce results including "wrong" solutions, i.e. when an >>>engine decides for the correct move after 4 plies, keep it during the 5th, but >>>switches back to a wrong move i.e. from ply 6 to 9, eventually coming back at >>>ply 10 after 17 seconds... (or may be never again withing 60 secs!). - It's >>>better to limit the time only, and set extra ply to 99. By that, only solutions >>>should be counted which are kept until the testing time limit. >> >>Then again it does not mean the engine keeps the move beyond 60 secs. >>Important for interactive analysis is IMO to be informed quickly when there >>exists an unobvious tactical possibility. > >An engine can calculate 7+ plies within *one* single second. Test results where >the correct move quickly shows up between i.e. ply 4 and 6 (and maybe never >again), are meaningless. The probability that a correct move is lost after 1 >minute, is *much* smaller than after 1 or even 10 seconds. > >Check your results for it, to see how many wrong quick solutions they include. I think that it may be better to have additional 1 ply and at least 5 seconds but I guess that there is no way to do it automatically if you do not write a special program for that purpose. Uri
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.