Arimaa Forum - Print Page


    
      
        Arimaa Forum
        (http://arimaa.com/arimaa/forum/cgi/YaBB.cgi)
      

        Arimaa >> Off Topic Discussion >> Cloning in Computer Chess
        
(Message started by: omar on Feb 21^st, 2011, 11:03am)

Title: Cloning in Computer Chess
Post by omar on Feb 21^st, 2011, 11:03am

There is suspicion in the computer Chess community that Houdini may be a clone of Rybka; and Rybka itself may have been derived from an engine called Fruit. The ICGA is getting involved to investigate allegations of cloning so that entries in the computer Chess championships are original work.

See article on chessVibes by David Levy:
http://www.chessvibes.com/reports/attack-of-the-clones/

Even though the ICGA may be able to prevent clones from being entered in the championship tournaments, there is no way to prevent someone from selling a re-branded clone. A possible solution may be the approach the Rybka developers are taking by making the program available only as a service running on their servers.

http://rybkachess.com.www52.your-server.de/Cluster/

The days of downloading and running a chess engine on your own computer may be coming to an end.

Title: Re: Cloning in Computer Chess
Post by rbarreira on Feb 21^st, 2011, 12:51pm

There will always be open source engines that you can run on your own computer (and the best ones are almost as strong as those controversial engines)...

Title: Re: Cloning in Computer Chess
Post by rbarreira on Jun 29^th, 2011, 4:21am

Wow... They concluded the investigation, and decided to strip Rybka of its computer champion title and ask for all prizes and prize money back, plus ban its author from the computer championship for life.

http://www.chessvibes.com/reports/rybka-disqualified-and-banned-from-world-computer-chess-championships/

Title: Re: Cloning in Computer Chess
Post by Fritzlein on Jun 29^th, 2011, 9:33am

It is sad that Rajlich did not defend himself. Maybe he couldn't share his code because it would have proven him guilty, but I wish we didn't have to infer this. I imagine that if any Arimaa programmer were accused of plagiarism, he would turn over his code to an investigatory committee in order to clear his name.

Title: Re: Cloning in Computer Chess
Post by Hippo on Jul 1^st, 2011, 12:59am

It's problem of "derived from" understanding. At the end anybody using alpha beta or bitboards except the first could be accused by plagiarism. Yes, it is evident Rajlich read the evaluation code of the Fruit and somehow used what he considered good.
Would be everything without problems if he would mention a lot of programmers he read code of ...
So in arimaa anybody using Fame would be accused ... provided you have codded Fame. Would I be accused with Hame, as it was inspired by Fame? ...

I am that much happy here in arimaa comunity we try to cooperate to create good bots rather than playing these "pattent" wars.

And yes, I declare my bot to be derived from clueless, but reorganizing the code to hide it would not be that much problem.

Title: Re: Cloning in Computer Chess
Post by rbarreira on Jul 1^st, 2011, 1:58am

on 07/01/11 at 00:59:43, Hippo wrote:

I think the problem is that they found actual copied code, which he always denied to be the case. Reusing others' ideas is fine, but reusing code without following its license is not (both according to the law and to the ICGA world championship rules).

Yes you based your bot on a version of clueless, but that version was specifically meant for people to start their bots from, plus you never denied this. The programs Vas copied from were not supposed to be copied without authorization (in the case of Crafty) or without releasing the derived program under the GPL (in the case of Fruit).

Title: Re: Cloning in Computer Chess
Post by Migi on Aug 25^th, 2011, 7:57am

I've been reading some of the evidence suggesting that Rybka was cloned from Fruit, and some of the evidence suggesting that it's cloned from crafty. They're both undeniably true, so I think what really happened is that he didn't clone either engine entirely. He just copied major parts of both engines.

I think it's sad, but I can understand why he did it, as you don't expect anyone to analyze the disassembly of your code so deeply. Actually, a lot of GPL libraries have had problems where big companies just use the library in their closed-source commercial projects.

A quick test that some of these GPL libraries have used is to compile your library with the same compiler as they used, then compress both binary executables, once together and once individually. Then, compare the compression rates using some information entropy theory and you have a fairly strong indicator of who has copied your code. Obviously it's not fail-safe, and you can have false positives when you both use the same third party library, but still it can be a useful first indication if any of you are afraid that people are cloning your Arimaa bot ;).