Even if such files were identical to Oracle files
(which they are not), their use is de minimis and
is not actionable."
Florian:
This baffles all description. If you think about it,
this "logic" would constitute a carte blanche for
everyone with a large software project to steal smaller
parts of other people's creations. Google's less-than-
one-percent-claim may very well be true, but this kind
of reasoning must be rejected altogether. Imagine a
music publisher who puts out a collection of 101 songs
and steals one or more of those songs from different
other companies,
Example is wrong:
The copying is not on "song" scope, but on a smaller scope.
If nobody was allowed to use a sentence that was ever used before, most writers would face a serious problem.
I have updated the blog post on reexamination and dismissal of copyright claims with an explanation concerning the scope of what was copied. So thanks for raising the issue, even though you quoted selectively because you left out the reference to 12 source files.
> I would consider a source code file to [be] a song.
To be fair, a song basically stands on its own. It may be part of a larger work, like an album or symphony, but these things are collections more than structures. The album does not fall apart if a song is omitted.
Songs themselves however are structures with dependencies: melody, harmony, rhythm, phrases, riffs, etc. If you remove the melody or the rhythm, the song is harmed in a much more fundamental way than an album or symphony would be by the removal of a song.
In other words, albums are composed of songs, but they are not built on songs. Songs are built on melody, rhythm, etc.
Now, back to source files. Applications are built on source files. It is tough to say what source files are built on, but they are composed of lines of code.
The analogy of a source file to a song breaks down, because the source file is only useful because the application is built on it. If we are determined to use this metaphor, applications are more analogous to songs, and source files are more analagous to structural elements like melody and rhythm.
That's a run-of-the-mill attempt to deny copying. However, if you look at the synopsis documents I produced (with a decompiled version of Oracle's code on the left side and the code found in the Android codebase on the right side), you can see that Google grossly overstates the importance of those differences.
The copying is not on "song" scope, but on a smaller scope.
If nobody was allowed to use a sentence that was ever used before, most writers would face a serious problem.