Author: Robert Hyatt
Date: 06:50:41 04/11/02
Go up one level in this thread
On April 10, 2002 at 22:15:01, Terry McCracken wrote: >On April 10, 2002 at 00:21:13, Robert Hyatt wrote: > >>On April 09, 2002 at 15:56:37, Mike S. wrote: >> >>>On April 09, 2002 at 13:01:01, Robert Hyatt wrote: >>> >>>>On April 09, 2002 at 11:29:45, K. Burcham wrote: >>>> >>>>>(...) >>>>>twelve statements from kramnik interview: >>>>>(...) >>>>>4. In almost every position Fritz7 (600 mhz notebook) was suggesting >>>>>objectively better variations (then Deep Blue). >>> >>>>That is a key statement. "in almost every..." If you play 39 moves like >>>>Capablanca, and one move like a fish, you still lose every game... >>> >>>I don't think you mean, a program must play 40 out of 40 Capablanca moves, to be >>>stronger than Deep Blue? >> >> >> >> >> >>What I mean is that if another program can play 39 out of 40 deep blue moves, >>that doesn't mean it is as good as deep blue. That _one_ move might be the >>critical move... >> >>Most _any_ program will find many of the moves played by deep blue. Most >>will find many of the moves played by _any_ GM player in fact. It is the >>ones they don't find that are revealing however... >> >>> >>>I think it's more like, Deep Blue could play, say 30 moves like Capablanca >>>(missing 10), and the strongest PC soft/hardware of today may find 33 moves. >>>Which would indicates it is most probably better. >>> >>>Or what must happen, before we can say, "this computer is better now, than Deep >>>Blue was." What is the criteria? I don't expect that Deep Blue was the best >>>possible chess computer of all times, throughout eternitiy. :o) >>> >>>I think Fritz' search depth on 8 CPU's will be competitive, compared to Deep >>>Blue (the node rate is not a good figure for comparison, also not in this case >>>IMO). >>> >>>Regards, >>>M.Scheidl >> >> >>Node rate is interesting to compare when (a) a program is 100x faster and >>(b) is no "dumber" than the other... >> >>Fritz is not a "genius" in terms of smarts... DB knew much more about lots >of things than it does... > >LOL! I'm sure you're correct Robert, that Fritz is inferior in it's "Chess >Knowledge" compared to Deeper Blue, for obvious reasons, one, if it had much >more knowledge, it would run very slow. > >The advantage of a supercomputer with 488 (?) dedicated "Chess Chips" to work as >software etc. makes for one "Mean Chess Machine"! > >Regardless, FWIW, what top chess engine uses the most "Chess Knowledge", and how >much does it help, as well as how much performance it loses? > >Terry I have a vague "feeling" about this. In a way, search speed is a form of knowledge. And evaluation "knowledge" is a way of handling cases where your search speed is not enough to solve a particular problem. It wouldn't surprise me if we one day discover that a very fast searcher with a modest eval can be just as strong as a very slow sarcher with a very smart evaluation. Two different ways to reach the same endpoint...
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.