Author: Vincent Diepeveen
Date: 12:42:22 08/27/01
Go up one level in this thread
On August 27, 2001 at 13:40:55, Robert Hyatt wrote: >On August 27, 2001 at 12:10:05, Vincent Diepeveen wrote: > >>On August 26, 2001 at 18:36:31, Amir Ban wrote: >> >>>On August 25, 2001 at 20:47:44, Mig Greengard wrote: >>> >>>>Sorry to dredge this up yet again, and ignore this rather than turn it into a >>>>flame war or something worse. I know feelings on this topic can run hot. >>>> >>>>Although we do not have enough of Deep Blue's games to make anywhere near an >>>>accurate assessment of its chess strength, I am requesting a summary of thoughts >>>>on how today's top programs measure up on a science level. In the past I've seen >>>>some admirably objective breakdowns on this topic from Bob Hyatt and a few >>>>others, but did not save them. >>>> >>>>Put Deep Fritz, or other top programs, on the best available platform on which >>>>they can run, and I imagine this is what they will have in Bahrain, and knowing >>>>what we do about DB, what comparisons can we make? >>>> >>>>Subjective arguments (chess knowledge in particular) are also welcome, but >>>>should be concise as opposed to argumentative! >>>> >>>>Thanks, Mig >>>> >>>>Editor-in-chief >>>>http://www.kasparovchess.com >>> >>>I've written on this several times, and to summarize my position, it is that a >>>machine that is much better than all the others must be shown to play an >>>objectively good move that the others don't, or at least the others need much >>>more time to get it. >>> >>>During the debate the supporters of Deep Blue's exceptional strength were asked >>>to name such a move, but failed to show anything convincing. >>> >>>This was the situation even in 1997, when PC's and engines were weaker. >>> >>>If Deep Blue did not play in its entire career a move that shows exceptional >>>depth, then I'm not prepared to believe it had it (and if it had it, it was a >>>wasted career). >>> >>>Amir >> >>Well spoken Amir, but we can add more to this, and that's the huge number >>of bad moves made by the IBM machine which are not made by any of todays >>software, both our programs included. >> >>See journal of ICCA june 1997 with a huge number of moves which Seirawan >>gives a ? or ?! mark where todays software hardly makes any of those >>beginners moves as they are usually covered in basic knowledge. >> >>Best Regards, >>Vincent > > >Vincent, one only has to look at the most recent WMCCC event, at your program >(or any other program there) to find moves at _least_ as ugly as some of the >moves you mention for DB. > >Why can you play lousy moves, game after game, they play one lousy move in >a game, and you criticize them without critizing your own program (or any of >the others there..) Even blindfolded my DIEP plays better as that Deep Blue. Seirawan has only put question marks at really horrible moves. You CAN win a game by playing bad moves if you first get a good positoin. DB blows it chances *before* it gets a good position. In fact even IF it has a good position it blows it. Further in nowadays WMCCCs moves get criticized which back in 1997 would receive a '!' for a program playing them. The level is much higher now from computerchess. Back in 1997 even 1200 rated dudes commented. Now the comment comes from IMs, FMs, and correspondence players like Uri Blass. I remember many beginners putting a '!' behind the Be4 move from deep blue in game 2. Well it is a MISTAKE that move. It deserves a questionmark instead of an exclamation mark. The move Qb6 there wins by force, you get a won opposite bishop endgame at least after Qb6, if black doesn't exchange though he loses even more pawns. After Be4 black can still draw as Seirawan shows. In fact in the GAME it was even possible to draw. After Qb6 it wasn't. However < 2000 rated of course do not pick this up. They think all opposite bishop endgames are a draw and then give Qb6 a questionmark. Of course Kasparov wanted to distract people from how bad he had played that game, and just murmured something about Karpov playing, in fact Kasparov's murmuring and loud complaining worked great, still even today the beginners write over what kasparov said. The Be4 move is NOT played by Karpov. In contradiction. Karpov would play Qb6 there as it gets white at least a won opposite bishop endgame, and Karpov would be of course the first to realize this! Now that type of the level of analysis back in 1997, way lower type of level than nowadays, that's what people forget. Oh the holy past, Fischer would beat nowadays Kramnik easily, that kind of dumb chatter i don't want to join. Of course Kramnik has had better training, better database possibilities, better openingstraining, more examples and better trainers and better technique. The guy would kick with induction on all terrains one of the great hero's of the past of course. Not because such a hero is old now, but simply because Kramnik is way better. It's like this in computerchess too. A nowadays AMD K7 1.2Ghz dual completely on paper is already way faster than any Cray from a year of 15 back. Including your own cray. How comes? Same is true for software. Software from today simply beats old software. Deep Blue is in fact a software program (of course it was put into hardware, let's forget that for a while), but nowadays software of course completely annihilates anything from the past. Like Rebel from nowadays might have beaten diep at the wmcc (congratulations Ed), because of a good bookline played by Rebel and after that rebel kept on playing good moves without hesitation. However if i would run my current program at a dual 1.2Ghz AMD and play it against rebel8 at a 200MMX, with a rebel8 openingsbook, then i of course completely annihilate rebel8. Idem against Nimzo98 even. In fact i recently got back a small match diep - nimzo98. Now that's a pretty fair match, because diep doesn't have learning. With learning every person in this world knows that it wouldn't be a fair match as you keep repeating victories then... Diep won *everything*. Now diep isn't even wmcc world champ, so it still makes some mistakes now and then. Especially its book is not at the current level, but compared to a few years ago it's a complete walkover as todays mistakes are not near the mistake level of a few years ago. Your comparision of crafty versus cray blitz from a few years ago is completely not valid. Crafty 7 ply versus what was it cray blitz 10 ply or something? Take a program doing checks in qsearch like DIEP, or take junior wmcc2001 (way better than junior7 seemingly as it won games and junior7 wins hardly anything). I remember first crafty versions with very little king safety. OF COURSE THAT WAS A WALKOVER FOR A DEEPER SEARCHING PROGRAM. However compare that with a todays program searching 12 ply with hardly forward pruning and a way better evaluation! That's complete suicide then for the deep searching cray. The crays/DB patzermoves now will all be punished at the moment they get made, whereas in the past things like deep blue could walk away with the patzermoves because the opponents ignored them and just allowed the patzer moves getting a real and unavoidable threat which decided the game! Best regards, Vincent
This page took 0.01 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.