More Computer Chess Controversy
Monday, January 24, 2011 at 9:22PM Somewhere around a year ago there was controversy about a series of programs that allegedly stole pieces of code from Rybka; now there's some controversy (or renewed controversy - I remember seeing something about this last year) as to the degree to which Rybka, at least in its earlier insilicarnation, depended on the chess engine Fruit. This forwarded letter by Fruit's programmer seems to be the instigator of the latest salvo.
I have no idea who, if anyone, is guilty of anything in this matter, and as a legal matter all the accusees should be considered innocent until proven guilty. I am curious about some points, however, like what counts as learning from a program as opposed to stealing intellectual property. If one uses someone's algorithm but expresses it in different code or in a different computer language, is it considered theft from a legal point of view? What about the moral point of view, from within the software community? And as a practical matter for those who want to take the highest possible road, are there any chess engines out there that are competitive and don't involve some "borrowing"?
computer chess
Reader Comments (9)
Dennis, The practice of studying a software application to learn how it works is called reverse engineering, and whether it is permissible depends on the license that you agree to when you use the software. That's right, the license--those 20 paragraphs or so that almost no one reads! So you'd have to check the license to make a determination.
[DM: Fruit was open-source, so my questions remain relevant.]
The Fruit programmer is claiming that Rybka source code (version 1.0) was largely, though not entirely, plagiarized from Fruit. In other words, Rybka programmers were looking at Fruit source, and typing code that was almost, but not exactly, identical. This is pretty much the same thing as when a student copies a few paragraphs from a website, but then substitutes a few words here and there and maybe switches a semicolon to a period.
I have no clue whether the accusation is true or not, though, without reading the Rybka and Fruit source code, which I am not likely to do in this lifetime.
[DM: I have no idea about the accusation's truth either (though I did understand what Letouzey was protesting), and since Rybka's source code isn't open, almost no one else will either. That's why I asked purely general questions in my second paragraph.
Dennis,
it is perfectly legal to reuse code from open source programs, but with the condition that the new code that was built on the open source code should also be released to the public. Rybka does not release its source code, and hence cannot reuse Fruit's code directly.
As to using the algorithms in Fruit and coding them differently or in a different language, the legal issues depend on whether there is a patent or a defensive publication on the algorithms involved. If there are none, it should be legal to refactor the algorithms in some other way without breaking the law. However, the ethical thing to do in such cases is to acknowledge the source from which the algorithm was borrowed.
However, the ethical thing to do in such cases is to acknowledge the source from which the algorithm was borrowed.
This is exactly what Houdart does with Houdini, explicitly saying on the front of his webpage that Houdini is based on Stockfish, Ippolito, and Crafty(?).
Hm, you'd have to find out which law text is applicable here first. The various intellectual ownership laws are not at all up-to-date when confronted with questions concerning software, but in any good legal text on the matter there will be some form of umbrella article that forbids blatant copy-pasting (even if altered slightly). My experience with law tells me it's one of those 'up to the judge' calls, probably only guided by a handful of cases as jurisprudence.
Little bit off topic, but not too much.
Dennis, when you checking your analysis of chess games, which engine you use (thrust the most)? According to the http://computerchess.org.uk/ccrl/4040/ Rybka and Stockfish (and also many other engines) are pretty close when it comes to ELO rating, but still, evaluation they gave for the same position may differ a lot.
[DM: At the moment, Houdini 1.5.]
Code that is distributed as source code or as a binary can be protected using patent law, copyright law, trade secret law, and/or contract law. In the US these areas of the law for software are quite mature. I'm not familiar with the facts of this particular case to determine whether or not the actions were legal.
I think the moral dimension is more interesting. We don't seem to feel the same way about stealing tangible property as we do about making illegal copies of a copyrighted work. A person may never think of stealing tangible property, yet may happily make copies of music CDs. Part of the reason might be that when you "steal" IP, you don't "take it away" from someone. You simply make a copy. Maybe it is the act of actually taking something away from someone that makes stealing tangible property immoral in our minds.
[DM: I think how one feels about this depends on who the "we" is. A person may not be stealing from his friend, at least not directly, when he burn a copy of a CD his friend bought, but those who would benefit monetarily from a legal purchase have been stolen from. They offered a good or service in exchange for payment, and that payment was denied.
Additionally, even if one doesn't care, or care very much, about the financial aspect, there's a feeling of violation. Tim Krabbe stopped his fantastic, wholly free Open Chess Diary* (at http://www.xs4all.nl/~timkr/chess2/diary.htm) at least in part because people were stealing his web content, and I've had deal with some similar lowlifes myself.
* I see he updated it last month with one entry, his first and only one in 20 months.]
Just to be clear: I'm not suggesting that it is moral or any less immoral to make illegal copies of copyrighted works than it is to steal a bike.
I'm simply pointing out that, in my experience, "regular" folk (the kind that you would run into at a supermarket) don't have the same reaction when someone makes an illegal copy of a music CD as they would have if someone stole a physical item, e.g., a bike.
I agree, it is a 'no-no' to distribute or make illegal copies of copyrighted works such as a music album. It is definitely the same as stealing someone else's property.
Anthony Green
http://www.dp-db.com/secrets-of-successful-traders