Cameron Reilly wrote:
> I've been re-reading Hofstadter's rationale for why computers had not (at
> the time of writing GEB) been able to surpass a chess grandmaster.
GEB is over 20 years old now. How much faster are computers in that interval?
> specifically refers to the ability of chess grandmasters to view the board
> positions in "chunks", compared to the computer's ability to "see further
> ahead", and the superiority of the former.
As highly as I regard Hofstadter, I did not consider his speculations on computer chess as all that insightful.
> And I got to thinking, what did
> the Deep Blue dev team do differently that allowed it to beat Kasparov? Did
> it merely crunch the numbers harder than its predecessors?
This is still being debated by those who concern themselves with these things, but it appears to be related to the old Russian saying regarding military equipment: Quantity has a quality all its own.
The stunning speed of Deep Blue evidently made some difference. IBM immediately retired the machine after the successful match when the development costs had already been spent. Another match would have cost them almost nothing. Some of us chessers guessed that IBM was not so worried about being defeated by Kasparov but rather they were worried about the massive mainframe being defeated by... Apple. Or some other microcomputer, thus calling into question the purchase of high performance mainframes.
> I've been to IBM's site but all of the information seems pretty lightweight.
Perhaps intentionally so.
> Does anyone know of a source for a deeper explanation of the Deep Blue
> strategy?
The debate continues, but as evidence of my previous contention, it should be mentioned that in that one critical game, where Kasparov resigned in a drawn position, a microcomputer running a 100 dollar software package found the draw inside of 2 minutes and played it. If Deep Blue were to be allowed to play the micros, they would occasionally lift its scalp. spike