Author: Don Dailey
Date: 12:33:54 06/15/98
Go up one level in this thread
On June 15, 1998 at 14:05:12, Dan Homan wrote: >On June 15, 1998 at 08:24:44, Thorsten Czub wrote: > >> >>Any society will learn that conservatism is no value at all. >> > >It is funny you would say that. It seems to me that you have >very conservative views about computer chess. > >The knowledge based approach that you advocate is quite old and >conservative. Many of the strong PC programs of the past used >this approach. The strong PC programs of right now and the future >seem to mostly use a faster, search based approach. Demanding >that they all use the older, knowledge based approach seems >quite conservative to me. > >Conservatism is quite useful, particularly in the sciences. It >keeps us from embracing every half-baked idea that comes along; >new ideas must prove their worth before being generally accepted. >But I think that the speed based approach has proved its worth. >I am not saying it is the only way to play good chess, just that >it is has proved itself enough that you should give it some >consideration. > > - Dan I agree that knowledge based approach is more conservative. I also think some degree of it is needed and am happy to see lot's of progress these days, current programs are much smarter than the old ones. I remember people thinking the Spraklen program was quite remarkable to "know" a Bishop vs King was a draw. This was considered unusual at the time. I also think knowledge is more brute force oriented than search. I somehow believe Thorsten envisions a tiny smart slow program as the ultimate chess program, and yet a knowledge based approach just trades speed for huge memory requirments. If you want to simulate the powerful pattern recognition associative techniques of the human mind you had better get ready for some HUGE memory requirements. - 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.