I’m actually not that upset. The decision may or may not stand, so nobody on either side should either overcelebrate or rend their garments in anguish. And even if APIs are copyrightable, maybe that’s not so terrible. But I think the OSS community just picked up a new to-do item.
[Disclosure: While working at Google, I worked with the attorneys on certain aspects of this case, and was deposed by Oracle. I am not the slightest bit neutral in this dispute.]
Might not stand? · I read most of the Federal Circuit Appeals Court judgment and boy, the law around this is clear as mud. While some of the judges’ language made think “are they crazy?” other arguments sounded sort of plausible.
On the other hand, this does fly in the face of the industry’s perception of the body of precedent, and I’ve read several sources asserting that this court has historically been controversial in its findings, something of an outlier. Anyhow, there are lots more miles of legal road ahead, and for now it’s safe to ignore predictions about how it plays out in the end.
The right thing to do is to pass legislation making it clear that API specifications aren’t copyrightable; but those wheels turn slow.
Might not be that bad? · Suppose it becomes settled precedent law that API specifications are copyrightable. That’s going to suck for the Java world, because Oracle will obviously look for ways to extract rent from this new-found territory. But Java is past its best-before date anyhow, for interesting new applications.
But beyond that? Well, let’s think, and make a mental list of all the big APIs that are existentially important to industry, and might prove to have copyright issues. I’m thinking... but coming up mostly empty.
The first thing that leaps to mind is AWS; Amazon has done some quiet saber-rattling about the APIs, and Eucalyptus took out a license. I think this is worth worrying about. Also, has Google been clear about the status of their Cloud APIs?
An OSS to-do · I think this means anyone who ships open-source software has a new line-item in their licensing work: Make sure it’s legally clear the API is freely re-implementable. Me, like everyone else I just copy the Apache or GPL boilerplate into my files. Is that good enough? I have no idea.
This is particularly interesting when a new OSS platform that wants widespread adoption has a big organization behind it: I’m looking at you, Go and Dart.
For the moment · To be safe, let’s just assume APIs are copyright-able, however crazy that seems, and have a close beady-eyed look at the ones we depend on.
The good news is, I just don’t see that many which really matter and are open to rentier exploitation.
Contributions
Comment feed for ongoing:
From: David Megginson (May 10 2014, at 14:01)
"But Java is past its best-before date anyhow, for interesting new applications."
Except perhaps for Android apps.
[link]
From: Bob Haugen (May 10 2014, at 14:15)
If this does stand, Oracle and Java will have just shot themselves in the head.
[link]
From: clive boulton (May 10 2014, at 14:17)
Oracle's chief is obsessed with winning. Let him win the obsession. Move the Android Cheese:~
Google$ Chrome
Google$ Dart
Google$ Go
Keep him guessing.
[link]
From: John Cowan (May 10 2014, at 18:05)
If Google had taken the principled decision to release Dalvik under the GPL, this whole lawsuit would have been blocked. Instead, they knuckled under to the handset manufacturers, who wanted a permissive license so they could keep their software stack proprietary.
However, it may still be that Google's use of the "declaring code" was fair use: neither judge nor jury has ruled on that yet.
[link]
From: Tim (but not THE Tim) (May 10 2014, at 19:28)
But wait - aren't the standard 'C' libraries technically APIs? Will AT&T suddenly decide to collect rent on all of those, including those implemented in open source?
[link]
From: DaveK (May 10 2014, at 21:10)
Well, I can see one big problem that you may have overlooked: all of a sudden, it's going to be illegal for anyone but IBM to make a PC-compatible BIOS any more.
[link]
From: Bob Foster (May 11 2014, at 00:02)
This isn't terribly original, but who now "owns" the Unix APIs?
[link]
From: Wang-Lo (May 11 2014, at 09:24)
@Bob Foster "who now "owns" the Unix APIs" -- Darl McBride?
[link]
From: John Cowan (May 11 2014, at 18:50)
The various Unix headers are available from multiple sources under a variety of permissive licenses including the LGPL, the BSD, and the CDDL. They have been reimplemented from scratch more than once. So no worries.
[link]
From: Blaine Cook (May 20 2014, at 13:37)
To the point about OSS APIs – if they're copyrightable, are they not, then, copyleftable? Is it possible that by building on top of Linux (and thus using GPL'd APIs), Oracle might find themselves forced to open source their code? (Obviously that won't happen, but it seems like an interesting possibility...)
[link]