Author: Robert Hyatt
Date: 08:11:22 01/13/99
Go up one level in this thread
On January 13, 1999 at 00:30:07, Howard Exner wrote: >On January 12, 1999 at 09:14:24, Robert Hyatt wrote: > > >>As a note, deep thought running at under 2M nodes per second produced a true >>GM rating over 24 consecutive games to claim the fredkin level 2 prize. I >>don't know of any program that could do that today. IE fritz only needs to go >>3-4x faster to reach this speed. Yet I haven't seen it win a game vs a GM at >>tournament time controls, much less perform at 2600+ over 24 consecutive games >>at 40/2 games... > >This observation of Fritz vs GM's at 40/2 can be countered with your own words. >You've reprimanded folks for them saying, Deep Thought/Blue did not >win all the ACM events by correctly pointing out that how could >they have won them all when they were not entered in them all. (Recall >your argument to the statement "Deep Blue has not won a recent computer >event") Well ... ? > >How can Fritz achieve this if it is not entered in 40/2 GM encounters? > >Yes I agree that DT's GM rating based on the 24 consecutive games is a great >accomplishment. Why the comparison to Fritz ? I didn't bring fritz up. It came from the poster I responded to that made the statement that fritz at 2M nodes per second might do better than DB. I pointed out that DT at 2M nodes per second produced a true GM rating over a long series of games vs GM players, with all games 40/2 or longer. We've not seen a commercial program anywhere, much less Fritz specifically, do this. We have seen lots of good results at action-chess speeds like game/30. However, fritz at 2M would only be 4x faster. Somehow I don't get the sense, from watching Fritz play, that 4x faster is going to help with the positional mistakes it makes when playing strong humans. And against a GM, I don't think its wonder- ful tactics are going to be enough to let it escape from making these mistakes over and over. Of course, I could easily be wrong. Since there is no good data to go on. So we are on opposite ends of the see-saw here. :) Why do you think Fritz hasn't played many dozens of 40/2 games vs GM's to see how it would do? I have an idea. It comes from the old saying "it is better to remain silent and be thought a fool than to open your mouth and remove all doubt." In this case, "it is better to remain silent and be imagined to be a GM, than to play a bunch of them and prove you are not." The programs are good. They are _not_ anywhere close to DB however. And IMHO they are not anywhere near playing like GM players either. But I must add a big *YET*, of course. They are getting closer. And I still maintain that there is a *HUGE* jump from just becoming a GM, to becoming a Kasparov-like GM. That's _many_ years away for computers (general purpose computers anyway). Just an opinion, of course.
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.