Friday 11 February 2011

Send in the clones

Last month I reported on a new program Houdini. Via Chessvibes comes news that it has just defeated Rybka 4 in a match 23.5-16.5. However this result has re-started a huge debate concerning the origin of Houdini and other programs like Strelka, Ippolit and even Rybka itself.
The point of contention is that Houdini author acknowledges using ideas from open source programs like Ippolit, Stockfish and Crafty, and the Rybka team claim that Ippolit is (illegally) derived from a decompile of Rybka (something the Ippolit team denies btw)
Interestingly the author of Fruit, Fabien Letouzy has become involved in the debate, in a round about way. Letouzy has been away from Computer Chess for the last 5 years but was recently asked to check of the program Strelka had been based on Fruit. He concluded that it was likely that Strelka was a 'bit-board' rewrite of Fruit. What he wasn't aware of until after he checked Strelka was that previously the Rybka team had claimed that Strelka was just a decompile of Rybka 1.0. So following the principal of 'If a=b and a=c then ...' there are now questions being asked about the birth of Rybka (and whether it is a rewrite of Fruit).
This may have serious repercussions for the Rybka team in that Fruit was released under the GNU Public License, which requires that programs that use GPL licensed code must themselves adhere to the GPL conditions (ie make their source code available, although the GPL does not preclude such programs being sold for profit).
This portion of the debate is now happening at Talkchess, although be prepared to wade through a huge number of posts if you want to read the entire discussion.

No comments: