Author: Fernando Villegas
Date: 14:02:15 04/25/01
Go up one level in this thread
On April 25, 2001 at 14:46:14, Dann Corbit wrote: >On April 25, 2001 at 14:40:53, Fernando Villegas wrote: > >>Hi all: >>It seems that after all Chris W., the official villain of this site according to >>many, was the father of a real good idea: that no matter how much you search, >>you cannot outsearch the power of positional awareness when speculative attacks >>are the issue. With just or mainly searching capabilities, the engine can see >>maybe 16 plies, really a lot, but an accurate positionallly based attacking shot >>can go a lot deeper, some times from middle game to the very end. >> >>If something has been probed -at least until now- by the incredibles wins of DJ >>is precisely that. Look the way it has won. Look the evals of Fritz. Look at >>that of DJ. Look at the lethal sacrifices of DJ againts all purely searching >>criteria. >> >>Luck? Just a casual output of "only" five games? Or maybe six by now? I cannot >>believe that. I see that Mr Ban has got his lessons from Gambit as Theron got >>his lessons from Wittington. By "lesson" I understand not neccesarily an >>specific learning of some witty tricks, but the fruitfull development of an >>idea. >> >>I cannot way the day this Super DJ will be in the shells. Or even better, the >>next program inspired, at its turn, by the perfomance of DJ. A new era of chess >>programming maybe - well, almost surely- is beginning. > >5 games is way to early to start proclaiming victory. Let's see what happens >from here on out. Right. Mine is only a guess. > >Positional evaluation terms are included in all the good chess engines. I think >that Shredder is very good at it. > >Tactics are also very important. > >It seems that sacrifices are working. But are you sure that they *are* >sacrifices. Perhaps these engines are actually seeing far enough that it is >only deep tactics. If the program can see the compensation, it isn't a >sacrifice. Well, it depends. My feeling is that in some of the games DJ did not see exactly what was going to happen in searching terms. At least not fully. Perhaps some extensions let DJ goes far beyond 16 plies, but then, being extensions, are in eseence speculatives as much they are heavily prunned lines in any case. > >I doubt if Amir has copied anything that Christophe or Chris have done. After >all, he can't see their source code. I did not say "copying" I have tried to say that it is a matter of an idea worked in his own manner for each programmer. > >My guess is that there are lots of brilliant things in Amir's code. Brilliant >search, brilliant eval, brilliant tactics, brilliant strategy, brillant opening >books. > Of course. >The same can be said of *any* program which has made it to the top of the SSDF >list. Of course also, but even so something more seem to be happening with DJ. Regards Fernando
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.