Equipment helps athletes hit harder, swim faster, etc. Even more than physical sports, chess has been revolutionised by technology. Engines can routinely beat world champions and the best sell in large volumes though most are free, or low-cost. Chess programming is highly competitive with both academic credibility and money involved. Programmers sweat to develop search, evaluation and pruning functions that work better.
Since June 2011, allegations of computer plagiarism have snowballed into a long-running controversy. Somebody writes new algorithms, somebody else reverse-engineers those, and adds some twists. Normal programming, and code-plagiarism as well, routinely involves translation across platforms and languages, which makes it difficult to prove charges.
In analogy, suppose Hundred Years of Solitude is plagiarised as follows: It is machine translated from Spanish to Malayalam, and then to German. Names, places are altered and then it's translated back to Spanish. This end-product will be quite different from the original.
Between 2006 and 2010, the engine Rybka won four world computer championships. Rybka is written by IM Vasik Rajlich, an MIT-trained Czech-American. Rajlich admitted he had developed ideas from an open-source engine, Fruit, written by Fabien Letouzey in 2005.
In 2011, 34 persons were asked by the International Computer Games Association (ICGA) to investigate charges that Rajlich had plagiarised Fruit. Many commission members were rival engine developers. But the ICGA had to balance peer review considerations versus conflicts of interest. The ICGA found Rajlich guilty. Rybka was stripped of its titles. Rajlich refused to release code (for fear of plagiarism!). The ICGA had to reverse-engineer, which means mapping to compare machine translations, rather than exact code.
The proof cited revolved around an extra decimal point. Rajlich used "0.0" in one place where "0" fits better with Rybka’s logic structure. (It make no difference to efficiency). Was this a Rajlich typo? Or was it due to copy pasting (Fruit uses "0.0")? Or was it error in reverse-engineered maps?
The ICGA has other ongoing plagiarism investigations. It will have to evolve more transparent processes since these charges will multiply, not evaporate. Rajlich continues developing new Rybka versions and he has made counter-accusations of plagiarism against other programs.
The diagram, WHITE TO PLAY (Rybka Vs Deep Sjeng, World Computer Chess Championships, Pamplona 2009) shows Rybka’s fantastic endgame prowess. White played 34.a7 Rc8 35.b4!! Bxb4 36.Ra4 Be7 37.Rb3 Nc7?!. The best defensive try is 37. - Nb4 38. Raxb4 Bxb4 39. Rxb4 Rc7 40. Kg2 R8xa7.
Now white penetrates with 38.Rb7 Nd5. A pretty variation is 38. - c5 39. Bg3 Bd8 40. Ra6! Kf8 41. Rxe6! Nxe6 42. Bd6+ etc. 39.Kg2 Nc3 40.Ra5 Nb5 41.Bxb5 cxb5 42.Raxb5 Rc2 43.Rc5! Ra2 44.Rcc7 Re8 45.Rb8 e5 46.Rxe8 Kxe8 47.Rc8+ Kf7 48.a8Q Rxa8 49.Rxa8 (1-0).
Devangshu Datta is an internationally rated chess and correspondence chess player
