Author: Don Dailey
Date: 07:51:16 08/16/98
Go up one level in this thread
On August 16, 1998 at 08:42:54, Robert Henry Durrett wrote: >On August 15, 1998 at 22:35:16, Ed Schröder wrote: > >>The overall advantage is for the computer in this area. A computer can make >>no mistakes. No holes in his memory. This advantage is by far superior to >>incidental opening preparation by a human on a computer chess program. >> >>Look at Aegon and other Man versus Machine events. Humans are scared >>for the computer books and play unusual openings. >> >>Look at Deep Blue-Kasparov and Rebel-Anand. Even these super players >>try to get the computer out of their books as soon as possible. >> >>- Ed - > >This is the very reason why I believe [and have suggested previously elsewhere] >that Deep Blue had an UNFAIR advantage against Kasparov. Deep Blue had >complete, perfect, and unstantaneous access to it's record of it's entire >opening repertoire whereas Kasparov had to rely on his memory. I feel that the >match would have been more fair is Kasparov had been allowed to use a printout >of his own opening repertoire. After all, IN EFFECT, that's what Deep Blue had! I don't think you can compare, because you are talking about apples and oranges. Who really has a memory advantage? I think it is the humans. We could kill the books, and then can we say it is fair? Human relies on his huge book (Kasparov indeed has awesome book knowledge) and computer gets NO book so it will be fair? >Similarly, I feel that Deep Blue had another UNFAIR advantage. It also had >complete, perfect, and almost unstantaneous access to a very large database of >GM games. Kasparov was denied that access. That was unfair, too. Kasparov was not denied access to any database. I'm not sure what you are refering to but did Deep Blue use a database of games DURING each game? >It's no wonder that Kasparov lost his second match with Deep Blue. The match >did not, in my opinion, demonstrate Deep Blue's superiority at playing chess. >It merely demonstrated that it could unfailingly refer to it's own repertoire >and to the database. > >A more fair contest would have equalized, to the extent practical, the access of >the two contestants to their repertoires and databases. > >But, philosophically, one could write this whole thing off with "Who said life >is fair?" So much for morality and ethics! Money talks. I agree with you completely, I do not think Deep Blue demonstrated superiority at playing chess here. I believe strongly that Kasparov is still the better player. But keep in mind that Deep blue is a well defined chess playing system. You cannot turn off half it's functionality and call this fair, it would be like running a foot race with a car and saying it's an unfair race because the car gets to use wheels. That defines the car. Humans store great amounts of book knowledge in their brains, is it fair to not let computers store book knowledge in their memory systems? I once posted that the ultimate cheat would be to have a 32 man database, or every move would be a lookup in a huge table to find the very best move. This is equivalent to a computer so fast it can compute the information on the fly. The results would be the same, the computer always plays the best move, but the means to achieve this are different. If you played each of these machines against each other would the search based one complain that the memory based was was not playing fair because it did all the work in advance? You just cannot say this, you accept the "chess playing system" for exactly what it is. I think your frustration is that you see the computer having some big advantages in certain areas and I completely agree with you. But the converse is also extremely true, but I don't think you should try to equalize then unless you are viewing it as an interesting experiment, which is really what this is anyway. Just for fun I compiled this partial list of things that are unfair in computer vs human chess matches: . Computer looks at too many nodes, unfair to human . Computer memory more reliable, unfair to human . Human memory too vast, unfair to computer . Human uses pattern recongition, unfair to computer . Computer gets to use endgame databases, unfair to human . Human gets to understand endgame concepts and apply them flexibly to similar positions, unfair to computer. . Human has to go to bathroom during match, unfair to human . Computer invulnerable to psychological strategy, unfair to human . Easier to pull plug on computer, unfair to computer. - Don
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.