Author: Graham Laight
Date: 01:27:16 10/15/04
Go up one level in this thread
On October 14, 2004 at 13:41:33, Robert Hyatt wrote: >On October 14, 2004 at 06:24:52, Graham Laight wrote: > >>On October 13, 2004 at 18:22:48, Robert Hyatt wrote: >> >>>On October 13, 2004 at 17:35:57, Graham Laight wrote: >>> >>>>Your "truth" is that the top computers are worse than top humans. "Think lower", >>>>you told me. >>> >>>Not quite. You seemed to have an exaggerated expectation for how the computers >>>would perform. I said "think lower". That doesn't mean that I believe the >>>computers are way worse than the humans. I just don't believe they are clearly >>>better yet... >> >>Good - this is a step in the right direction! ;-) >> >>>>Well - I know you're not going to give me an estimate as to the win/draw/lose >>>>probabilities of top computers v GMs, so I won't bother to ask - but after the >>>>work I've done with my simulator today (have you tried it? It's quick and easy >>>>to run - just follow the 4 easy steps), then if GMs are significantly better >>>>than computers at chess, I can tell you that Fritz and Hydra getting 7/8 was a >>>>sensational result. Let me give you reasonably accurate analogies from other >>>>sports with which you have some familiarity: >>>> >>>>1. it's like the 6 stone weakling who has never had a fight before flooring the >>>>national karate champion >>>> >>>>2. it's like a donkey and cart winning the regional drag race evening >>> >>>No. In drag racing or karate participants are very "steady". Not so in >>>computers vs humans at chess. Odd book lines. Good book preparation by the >>>humans. All serve to significantly skew final results in odd ways... >> >>This "steadiness" strongly implies to me that there are big differences in the >>participants' ability levels (be that engine power, skill, or whatever else) - >>this is what produces "steadiness". If competitor A has a 95% probability of >>beating competitor B, the results of games between them will look "steady" to >>the casual onlooker - wheras if that probability were, say, 50%, the results >>would look distinctly "unsteady". >> >>-g > >If you run in an x-second class, you can't run faster than x seconds or you >break out and lose. So most cars are the same as far as performance goes. Now >it comes down to the human driver in heads-up racing. Reaction time, making >sure you beat your opponent but not breaking out, etc. Those become pretty >variable thanks to human nature... So presumably the results are not so "steady" in that case. -g >>>>-g
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.