Author: blass uri
Date: 08:34:47 06/22/00
Go up one level in this thread
On June 22, 2000 at 00:48:20, Dann Corbit wrote: >On June 21, 2000 at 23:04:56, stuart taylor wrote: > >> With a minimum of positional knowledge? Of course, brute force at 300 ply would >>require absolutely no positional knowledge at all, to be an ultimate authority >>on perfect positional chess moves. No human would ever beat it, and it wouldn't >>require any opening knowledge either. >> But maybe 200 ply also? or how about even 60 perhaps with just basic assesment >>knowledge? >> >> At any rate, I would be much more intersted in a program that guranteed perfect >>chess even if it required a few hours per move, than a super strong program like >>fritz 6a which displays most of its strength after minutes, if not seconds. >> I don't even mind if it made big blunders at 5-minute games,or even longer >>games. > >Calculate how many nodes are needed for just 20 plies. Then you will quickly >see that any program which can make 20 plies is not examining all the nodes. >That will be true for any program which manages the feat this century. Hence, >they will all be doing some kind of pruning. Therefore, there will always be >loopholes in the algorithms used to exploit. > >sqrt(pow(35, 60)) is still a "rather large number" and I would not worry about >any programs coughing up 60 accurate plies any time soon. This number assumes no hash tables. The theoretical number can be slightly smaller because hash tables can help program to be faster but even if I assume only pow(32,60)>pow(10,90)=10^90 and if computers become twice faster every 1.5 year then in order to be more than 10^80 times faster you need to wait more than 400 years. I also do not believe that the improvement in the hardware will never end and I expect it to end in the next 100 years. Uri
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.