Hacker News Reader

Comment by moomin | Hacker News Reader

I'm sure this would be a great result for Google, but it would kick the question of whether APIs were copyrightable into the long grass.

View on HN

For another example, part of Android's userspace runtime (Bionic) is a direct copy of the Linux headers made by automated source-scraping tools. If APIs were copyrightable, this would be a GPL violation (as others have asserted [1]).

Moreover, all userspace programs that use this information (by linking against glibc) would be considered derivative of the Linux kernel. All userspace programs that link against Windows' C/C++ runtime would be derivative works of the NT kernel. And so on.

Linus himself has asserted that the copying of such APIs is considered "normal use" and that such programs are not considered derivative works [2].

[1] http://www.fosspatents.com/2011/03/googles-android-faces-ser...

[2] http://www.groklaw.net/articlebasic.php?story=20110322014831...

I'm no expert, but you are not comparing similar cases. It is in the interest of an operating system / tools vendor to give their customers blanket permission to use their APIs with the platform they are selling, otherwise they have an unattractive product (buy our computer and write software for it, but we own whatever you write!)

Google's case is different: They have created a platform that uses APIs from a different vendor's plaform (Sun) and while Sun would have been perfectly happy to grant their own users permission to use the APIs, of course they have no interest in Google's customers being given that permission.

The whole basis of assuming that anyone can use an API that you write is that "anyone" is "all people who you have permission to use your software in the first place". Of course it limits adoption of your platform if you restrict who is allowed to use it, but so does charging money for it. Some people choose to charge money for their work, some people give it away.

I disagree. Whether or not it is in the interest of the vendor is irrelevant; if APIs are copyrightable, the vendor's license extends to any work that uses that API. If they are not, the vendor has no say in who can reimplement or use that API, as they have no ownership rights.

If APIs were copyrightable there would be no Wintel machines... IBM would have killed Compaq for daring to build a clean room BIOS.

Logically I can't see why an API should get copyright protections. By its definition it is a interface, and the moment you start allowing restrictions on implementing an interface your killing competition. It seems that in almost ever other field fair use has decided that whatever is required for compatibility is not copyrightable.

That said my beliefs and the law can at times be completely different things...

Isn't it the same with the software patents they have in the US?

I don't see why existing anti-competition laws would suddenly disappear if an API was copyrightable ?

And your argument about Wintel machines is actually backwards. Because if APIs were copyrightable there would be MORE competition in that circumstance. Due to less standards.

I am actually in favour of having standards organisations take a greater role in pushing real standards rather than de-facto ones.

The best standards are those which have been taken from real-world implementations. In other words, the way you get a good standard is to ratify a de facto one.

Reply on news.ycombinator.com