Author: Robert Hyatt
Date: 09:39:48 01/07/00
Go up one level in this thread
On January 07, 2000 at 09:16:37, Chris Carson wrote: >On January 06, 2000 at 23:27:31, Robert Hyatt wrote: > >< Big, perhaps relevant, Snip > >> >>Not exactly. The ratings were adjusted to correlate with some human vs computer >>results that were known. But that set the ratings for a couple of programs back >>in 1992-93. Since that time, thousands of new games, computer vs computer were >>played. And as I mentioned, a small change to a program can produce a wide >>gap in Elo performance. With no checks and balances to keep the list in line >>with the old Swedish federation ratings. The ratings were never calibrated to >>FIDE ratings, and after 7 years, there is little doubt that they have drifted >>over 200 points above FIDE, at least. >> >>At least I don't believe any program is a 2700 player, on today's hardware, >>which puts them in the top 10 or so of the worlds best players. I don't >believe >>they are in the top 100 yet... > >Bob, > >Consider the following and let me know where my errors are, >I trust your insights. > >Not an attack, just my own thoughts. :) > >-------------------------- > >I think the following two are in the top-50 (perhaps top-1). :) > >Deeper Blue (not commercial, but hardware does exist and is old >technology, very old by my standards, but I do work in the ASIC >industry). > >Deep Blue jr (same as above) > >NOTE: Neither has a GM norm or enough data to support this statement. > But they do have some very high TPR's. :) > > Deeper Blue has just one good TPR, might just as well produce one > 400 points lower. Especially if a GM was allowed to study games > and prepare for it. :) > >I also think the following program could easily be at 2600 TPR and >in the top-100 easily, but will not be allowed to gain a GM norm. > >Crafty on a 16xPIII-800 (perhaps less MHZ needed). >TPR above 2600 no problem (in tournament play >where all GM's play each other and Crafty, no extra preparation advantage >to the GM's). :) > I don't agree there, yet. Crafty on a 16 cpu 21264 alpha is very fast. And the alpha is way faster than the PIII/800... but I didn't get any sort of feeling of being 'invincible' when I had it running on one for an hour or so last Summer. >Hmmm. Would the above Crafty beat Cray Blitz? :) Doubtful, as that hardware isn't as good as a T932. But the 'styles' are way different too. CB was very tactical, with singular (and other) extensions and no null-move to speak of (r=1, non-recursive). Crafty plays a totally different style of chess, more closely resembling what I would call "anti- human" which is probably not the way to play a tactical monster. But it is the way to keep GMs from rolling it up after a few hundred practice games... > >I also would not be surprised if Roman beat the above crafty on ICC >on a regular basis and if you calculated the rating for Crafty based >on just games from Roman, Crafty might be rated 200 to 400 points lower >than the TRP from a GM tournament. Roman knows where the holes are >in Crafty and knows how to take advantage of them. :) Roman can't beat the current crafty regularly. But it is way clear that he knows so much more than the program, as when he avoids tactical crushes, he finds nice positional plans... > >Yes, I am mixing TPR and MPR. TPR is all I have and may be all we >have. I wish the situation was different, I prefer hard data to >analysis. This is all just speculation. I can never prove any >thing definitively since the computers are not allowed in tournaments. >Alghough, IMHO, small amounts of data support the above speculation. :) > >Just my humble opinion. :) > >Best Regards, >Chris Carson > >< Big Snip>
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.