Author: Robert Hyatt
Date: 11:29:16 09/17/99
Go up one level in this thread
On September 17, 1999 at 11:03:32, walter irvin wrote: >On September 17, 1999 at 05:52:27, Andrew Williams wrote: > >>On September 17, 1999 at 02:56:38, Terry Ripple wrote: >> >>>On September 16, 1999 at 12:05:28, Timothy J. Frohlick wrote: >>> >>>>On September 15, 1999 at 17:57:08, Joshua Lee wrote: >>>> >>>>>i think it's time to start playing the best PC software vs the Mainframes and >>>>>Dedicated Comp programs. What do you think would happen?? >>>> >>>> >>>>The Dedicated units would get trounced but manage to win no more than 30 percent >>>>of the time. Deeper Blue would beat all the PC software more than 90 percent of >>>>the time. >>>> >>>>I always wanted to own Ken Thompsens' Belle machine and now I own a computer >>>>that with Hiarcs 7.32 is far superior. The mainframes are largely >>>>decommissioned after limited use and neither I nor you have the millions of >>>>dollars to get them up and running (maybe you do). >>>> >>>>It is like the progress in our society. People have more things and can do more >>>>travelling but they are becoming less and less powerful with respect to each >>>>other. >>>> >>>>Tim Frohlick >>>------------ >>> >>>I doubt "very" much that Deeper Blue could beat the top programs 90% of the >>>time, like Hiarcs7.32 and Fritz5.32!!! That win percentage would mean that its >>>about 400 ELO points stronger than the top programs and there is no way that it >>>is going to beat Hiarcs7.32 or Fritz5.32> 45 out of 50 games!!! Guaranteed!!! >>> >>>Terry Ripple >> >>Have you read Hsu's article in IEEE Micro (March-April 1999)? It's about the >>chess processors in Deep Blue. In it he gives a tantalizing glimpse of what >>a one-chip version of Deep Blue could do to a couple of the top programs in >>1997. He suggests that the programs were very easily beaten. It's a very >>interesting article. >> >>Andrew >its possible that deep blue jr could win 100 out of 100 vs hiarcs fritz cm rebel >ect ect. and still not actually be that much stronger ,just seeing that much >deeper .but may not score that much better vs humans further ,much further test >are needed.but if pc catch up with deep blue jr. then it would be a much more >even playing field .when DB jr comes out it would be in high demand i think. You are correct of course. I have reported this observation many times, that tiny changes to a program make it play drastically different. And having one important piece of knowledge that your opponent doesn't have (King Safety, in the case of deep blue) means that most games are going to swing around that particular concept. Hsu mentioned that King Safety was _the_ issue in the DB (crippled chip) vs commercial programs... So yes it could win every game, but when you play them both against a group of human GM players they might be very close (doubtful, but absolutely possible).
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.