Author: Don Dailey
Date: 16:12:36 12/25/97
Go up one level in this thread
On December 25, 1997 at 18:33:57, Komputer Korner wrote: >Hold it right there. Guys come to your senses!!!! there is no way to get >5 plies deeper if you are still a full width searcher. I don't have >Junior, but unless it is a selective searcher it is not going to get 4 >more plies than Fritz. >KK I agree on this. Perhaps Junior shows several more plies of PV but this is meaningless, it's real chess strength that counts. Here is an experiment for you programmers: Turn off all your null move searches and always take a cutoff when the static score is above beta. You will see big PV's and they will tend to be quite reasonable, in many cases completely correct. A few well chosen move extensions added to this will give you incredibly long PV's with quiet moves at the end. But then play a few games with it... But the Lang programs have always been good at this too and play very strong. But still, on average Richard Langs programs are not REALLY seeing several ply more than anyone else and never did. There was a time I believed he was about 1 ply ahead (on average) than anyone else and his results showed it. He typically solved most problems 2 - 10 times faster than most of the other guys, this is about 1 ply. This is not to say (lest I be blasted) that Junior is doing anything wrong. But don't expect to see him winning 99.5 percent of all his games against the very best programs because he is seeing 5 ply farther! -- Don
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.