[Discuss] can you copyright an API?

[bogstad at pobox.com (Bill Bogstad)]

On Wed, Apr 25, 2012 at 9:30 AM, Edward Ned Harvey <blu at nedharvey.com> wrote:
>> From: discuss-bounces+blu=nedharvey.com at blu.org [mailto:discuss-
>> bounces+blu=nedharvey.com at blu.org] On Behalf Of Tom Metro
>>
>> In any case, we're heading off on a tangent from the thread. I'm far
>> less interested in whether Google prevails on the overall suit than in
>> the question of whether APIs can be protected by copyright.
>
> Even if an API is copyrightable, we've already concluded it's easily
> circumventable. ?They can't illegalize translating from one language to
> another, and they can't illegalize designing a language to be easily
> translatable to a specific other target language. ?In other words, you can
> easily script a trivial transliterate function that will translate. ?I
> suppose Oracle could prosecute anyone who wrote such a translator without
> licensing the copyrighted API...

Err, translations are derivative works (at least in the literary
world) and the copyright holder for the original work controls them.
I'm not sure the translation script would be a subject of copyright
violation, but Google's implementation of their "new" API would seem
to me to be a derivative work of Oracle's API.  I don't think APIs are
(should be?) copyrightable, but I don't think your sleight of hand
would work to get around the problem if they were.   Speaking of
which, if APIs are copyrightable do K&R own the copyright on the API
for the C standard library or does Stroustrup own it for C++?   That
would mean that every implementer of those libraries might be subject
to copyright violation suits.  That's why some people consider a
ruling saying APIs are copyrightable would be a major change to the
way people think about software and copyrights.

> If an API is copyrightable, the above does not mean the copyright is
> meaningless... ?Just that in the future, it will have minimal impact.
>
> The main question of interest will be whether or not the API is patentable.

As I understand it, even software patents have to do with actually
"doing" something.   An API doesn't actually do anything, it is the
underlying implementation that does.   If I could claim patent on an
API, it would seem to me that I could claim your book describing my
API violated my patent as well.  I wonder what that would do to all
the authors of Windows programming books out there.

In any case, I think it is Google's actual Dalvik VM that might be
subject to Oracle's patents (and part of this lawsuit); but I don't
think the API is relevant to paents.  And as far as patents go in this
case, I believe that all but 2? of the patents that Oracle started
with have either been tossed by the Patent Office (upon becoming aware
of prior art) or Oracle decided they weren't going to claim during
this case (perhaps their infringement case was too weak).   Even the
two that they are moving forwarded with have been whittled down by the
Patent Office by prior art.

Frankly, this case seems a lot like the SCO/IBM case.   We started
with lots of grandiose claims and even before the trial starts large
chunks have already been tossed.

Bill Bogstad