Android’s Bionic Problem Is Not “Bogus”: Why Judge Alsup Got It Right And Linus Torvalds Got It Wrong
Posted on Tuesday, Nov 8, 2011
In September, federal judge William Alsup denied Google’s request for a ruling that the Java application programming interfaces (“APIs”) were, categorically, not protected under copyright law. In that order, which came in Google’s litigation with Oracle over Google’s use of Java in its Android mobile operating system, Judge Alsup ruled that each of the disputed files must be analyzed individually to determine whether it is protected by copyright. He also ruled that even if the individual files are ultimately determined not to be copyrightable, the selection and arrangement of those unprotected elements may nevertheless show creativity that is entitled to copyright protection.
Android’s use of Oracle’s Java isn’t the only example of Google’s cavalier attitude toward copyrights. Judge Alsup’s order didn’t address Android’s Bionic library, which I analyzed back in March, but his reasoning can be readily applied there, too. As I explained, when Google created the Bionic library, it attempted to work around GPLv2 by using an automated script to “clean” the Linux kernel headers. After analyzing the code in several files, I concluded that Google’s categorical, automated approach to “clean” the headers didn’t work as a matter of copyright law, and I observed that Google’s methods provided an easy bypass of the GPL for commercial exploitation.
My observations piqued the curiosity of a number of journalists and bloggers, and several recognized the implications of Google’s approach. It didn’t take long for Google supporters (and, surprisingly, some free software advocates) to go on the attack. After much sturm und drang in the Android community, Linus Torvalds himself reportedly declared that the idea “seems totally bogus.” He admitted, however, that he had not bothered to review the code or my analysis of Google’s approach to “cleaning” the Linux kernel header files. Although Linus has emphasized that he speaks only for himself and not for any of the other contributors who hold copyrights in the Linux kernel, critics of my analysis nevertheless declared it “debunked,” on the basis little more than a flippant quote from Torvalds in a blog.
Another blogger who criticized my initial analysis declared that “It Is Not That Simple.” At least he got that part right, but that was my very point, and it is also the point that Judge Alsup made in his recent order: the copyrightability of the Java APIs and the Linux kernel headers cannot be determined on a categorical basis. It’s necessary to examine the files themselves, perform a complex analysis of the code, and properly apply copyright law. I did that in my first paper, and I’ve done it further in this new white paper. That analysis leads me to the conclusion that Google’s approach doesn’t work. But if it does work, if the guardians of the Linux kernel and the GPL believe that it is acceptable to use an automated process to “clean” GPL’d headers or code so that you can re-distribute them under a non-copyleft license, that fact should be made clear so that the uncertainty and doubt is dispelled for good. Blogs