Author: Oliver Y.
Date: 14:52:18 04/11/99
Go up one level in this thread
On April 09, 1999 at 14:00:08, Dann Corbit wrote: >On April 09, 1999 at 06:30:32, Mark Young wrote: >[snip] >>At one time I was firmly in the camp that chess is chess, but as I have tested >>programs more against humans. It is much harder to deny that some programs play >>much better against humans then they do against other chess programs and vice >>versa. >> >>I will give one examples for each case, much better against humans, and much >>better against computers and see what others think. >> >>Much better at playing computers then humans. >> >>Junior 4.6 >> >>Much better at playing humans then computers. >> >>Crafty >> >>I found that each one of the example programs played +100 rating points better >>when playing against computers for Junior 4.6, and humans for Crafty. >That's a very interesting proposition. If you have actual data, it may be worth >while to run calculations and produce the mathematical ELO scores against both >types of opponents. Bruce made a very sound post that showed -> most of the >time it's pure conjecture. You seem to have made some measurements. A careful >measurement of the data should prove interesting for a number of reasons. >1. We can examine those programs and find out what is it about the way that >they play that makes them superior against a particular type of opponent. >2. We can use this information to *possibly* write algorithms that are better >against humans or machines as we see fit. >3. We can look at the "favorite strategies" tried by these programs and ponder >'why' are they successful against a particular type of opponent. > >When I think of what I have seen regarding these programs, I have a fuzzy >emperical feeling that you may be right about both of them. Say, what's the difference between "fuzzy empirical" and "heuristic"?
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.