Author: Jeremiah Penery
Date: 15:19:58 10/27/99
Go up one level in this thread
On October 25, 1999 at 20:23:28, Bruce Moreland wrote: >On October 25, 1999 at 04:01:32, Tom Kerrigan wrote: > >>You know, there has been a lot of talk about similar strings and data patterns >>in the executables, but this overlooks the most obvious way to tell that a >>program is cloned: how it plays. If you run it for a while on the same >>positions, are the results similar? Same moves? Same depths? Same evaluations? >>Perhaps similar patterns in the evaluations, indicating similar searches or >>evaluation terms? Perhaps similar scores on test suites? If all this stuff is >>different, then it seems like the program has been modified enough for it to be >>interesting. Definitely interesting enough to be allowed at a tournament... >>whether or not it should be sold is a different ethical question. > >If someone takes Crafty and fiddles with the eval, there is a good chance that >they can make it play differently. > >This does not mean that they have created something new and original. I have >not tried to prove it, but I believe you could make a program appear to be >different by changing piece square tables and a few other eval terms, which you >understand would take no effort. > >I argue strongly against this "playing style" method of determining program >originality, because I don't want to be at a tournament where someone >successfully uses this argument on the organizers. It is a bogus argument. > >bruce Exactly. I run a modified Crafty on FICS, and just by changing some of the evaluation code, it sometimes looks like it's not even a computer playing. (I had a version that, because of bugs, played _horrible_ moves. Most of my versions are better. :) There isn't much chance that someone could tell it was Crafty just by watching it, but 99% of it is Crafty. Jeremiah
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.