For those passionate about the intersection of litigation and software, this new article on Abstraction-Filtration-Comparison will be exciting. This is a Court-approved method of determining whether or not one program infringes the copyright of another.
Comments and corrections would be appreciated.
I thought the article was interesting, but it left me feeling flat with when it didn’t draw any conclusions from the analysis. And example of the comparison text would be very helpful, even if it’s artificial. The a portion of the Ingres 9 vs. 9.2 example might be instructive.
This article is based on a source code analysis. What happens when someone decompiles an interpreted language like Java or C#, steals intellectual property and repackages as a new product? In this case, string analysis on the source would look completely different, yet there is theft.
Recently, we found an old version of one of Java products decompiled and posted to Google code as open source. In this case it was obvious it was one of our products – they even left their original unique license key in the decompilation! It didn’t help that it was originally obfuscated.
A developer doing a more sophisticated decompilation would be harder to pursue in the courts.
Bill: Thanks for the feedback. If we get a burst of energy we may try finding some similar systems, abstracting them, and doing the comparison. But as I noted in the article, if the abstractions are done well, the jury and judge should be able to do the comparisons themselves!
David: The A-F-C method only becomes relevant once a lawsuit is in progress, so both parties have access to all source code through discovery. I think the method would still work to show that a decompiled copy was infringing. The similarity would reveal itself perhaps at Abstraction Level 1 or 2.
Very interesting. Are most cases of software infringement based on pure source code similarity or functional similarity? Seems to be very good for the former, but maybe not as helpful for the latter.
An example of the latter is a case from several years ago, where video game maker id Software got into a patent scuffle with Creative Labs due to id’s top software guy, John Carmack, independently arriving at a solution for rendering shadows in a video game. (http://www.shacknews.com/article/32824/creative-labs-patent-doom-3)
Unfortunately, he didn’t realize this methodology had already been patented! (id ended up paying royalties to Creative Labs to resolve the dispute)
DeAngelo: “Functional similarity” would be covered by patent law, not copyright law. Anna Karenina and the latest Danielle Steele novel perform the same function (both are novels; both can support a cup of coffee) but that doesn’t mean one has been copied from the other.
Philip, here’s a question: Would you allow a non-technical manager of senior software developers to manage said developers using the middle and end-products of the A-F-C method?
If the answer is yes, then perhaps there is room to re-purpose A-F-C as a software development management tool.
If the answer is no, then how can we let lay folk completely disconnected from the software and the organization that created it make any ‘management decisions’ about it?
@philg
I see. A company called Zynamics which has a product that I believe does something similar on executables (BinDiff) was recently acquired by Google( http://techcrunch.com/2011/03/01/google-buys-security-analytics-startup-zynamics/)