Author: Robert Hyatt
Date: 06:23:58 01/14/00
Go up one level in this thread
On January 14, 2000 at 02:37:25, Bruce Moreland wrote: >On January 14, 2000 at 00:44:22, Robert Hyatt wrote: > >>On January 13, 2000 at 23:07:07, James Robertson wrote: > >>>If DB plays a match with, say, Leko as a preemptory event and draws or loses, >>>all of the mystique of a super-machine facing the strongest human in the world >>>goes out the window, as obviously it is weaker than many humans. It is better >>>for Hsu to keep his machine under wraps and use its unknown play for publicity >>>and as a weapon. > >>Of course GM players don't go into seclusion for several months before playing >>for the world title, right? > >I have to comment on this, the way I have several times before. It is not a >matter of going into seclusion. It is a matter of being completely unknown, and >that's a different thing. The "program/machine" wasn't totally unknown. Predecessors played in every ACM event, through 1995, 1 year prior to the first match. Then DB Jr was used to do demos all over the place. DB-1 didn't exist until right before the first match, so having any sort of game history would have been impossible. DB-2 didn't exist until right before the second match. But the DB Jr games were certainly available. I saw several of them in person. Hsu's book gives the time-line for these events. It was remarkable, as for both events they _barely_ got the machine together, with very little testing. It took longer than expected (both times) to get working chips out of FAB and into his lab for testing, then into an SP for actual use. > >If someone says that Fritz beat DB, you can say it wasn't DB, and in fact you >yourself did in the part of the above that I snipped. But if someone says that >Kasparov has not seen DB, the argument is that he can go look at the games of >DT. That is the point. As I mentioned, if someone studied games of Crafty from 6 months ago, and found a weakness and planned a match around that weakness, they would likely find that the weakness is gone. Or smaller. Of course they might well find _new_ ones as well. But it doesn't take much of a change in a computer chess program to make it play entirely differently. I believe he was convinced (by others) that playing Fritz was a good simulation for playing DB. It wasn't... > >Well, that is not fair. DT is either DB or it isn't. I don't think that it is. > >I think that it's fair to want to see at least some examples of the play of your >opponent before having to play a match, and since DT is not DB, Kasparov didn't >see any. > \ See above. Without a working DB until about 1 week before the first match, it would have been _hard_ to produce games. >To have to play something that is absolutely unknown is no big deal when you are >eight years old and are playing in a scholastic tournament. But when you are at >the top level, where minor factors matter a great deal because the limits of the >human mind are being reached, I think it has to be a big disadvantage to be >pitted against the mystery box from Mars. > >It's not like he has to see a hundred games in order to try to cook the thing's >book. I think he deserves to see a few representative middlegames and endgames >so he can try to figure out the thing's capacity. During that match for all he >knew the thing was searching forty plies. > >bruce
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.