RMS on Copyrightability of APIs

Posted May 1, 2012 2:53 UTC (Tue) by slashdot (guest, #22014)
In reply to: RMS on Copyrightability of APIs by wahern
Parent article: Fair use or "first excuse"? Oracle v. Google goes to the jury (ars technica)

Whether dynamic linking to a library creates a derived work is totally unrelated to whether APIs can be copyrighted.

In the latter case, both the application AND the library are newly written code, and only the API itself is from the original implementation.

Anyway, I don't see how a court could rule an API copyrightable.

It would be like allowing someone to have a copyright on the English language, and requiring everyone to pay royalties to speak or publish unless they switch to French or something else.

RMS on Copyrightability of APIs

Posted May 1, 2012 4:49 UTC (Tue) by iabervon (subscriber, #722) [Link]

It would actually be more like allowing someone to have a copyright on Klingon or Esperanto. APIs and constructed languages were at least created by a particular individual or organization, and have a certain amount of arbitrary choice. If there are a lot of fields and methods that aren't determined by any fixed scheme, that aspect could be copyrightable. If there are quirks of the API which are only necessary for interoperating with that particular implementation, those probably are copyrightable. In general, APIs should probably be copyrightable to the extent that they are poor APIs.

RMS on Copyrightability of APIs

Posted May 1, 2012 18:15 UTC (Tue) by wahern (subscriber, #37304) [Link] (7 responses)

Several courts have ruled that APIs are copyrightable. I don't have my casebook handy but I believe there was an Apple case a few decades ago where this was affirmed at the appellate level.

But by-and-large the copyrightability of APIs is still an open question, I believe.

I personally think APIs (and many, many other things) shouldn't be copyrightability. But my "beliefs" count for squat. And the problem in the FOSS community is that we hear a certain cadre of academics swear up and down that APIs aren't copyrightable, and we believe them because they're the only voices we hear.

Well, unfortunately, there are other voices. Rich voices. And those voices pay for conferences and retreats for judges and other policymakers. And though to our ears those voices sound weak and distant, that simply might say more about how far away from the legal mainstream our community is than it does about where the actual law is actually headed. Though, I tend to think most district courts aren't eager to join the party happening over in the patent circuit, and seem to be holding the line against strengthening of copyrights wrt IT.

Still, there's quite a long list of sharp turns the law has taken which surprised the vast majority of legal scholars.

RMS on Copyrightability of APIs

Posted May 1, 2012 18:41 UTC (Tue) by dlang (guest, #313) [Link] (6 responses)

> everal courts have ruled that APIs are copyrightable. I don't have my casebook handy but I believe there was an Apple case a few decades ago where this was affirmed at the appellate level.

Oracle was unable to produce any examples for the court, so I question this statement.

RMS on Copyrightability of APIs

Posted May 1, 2012 21:09 UTC (Tue) by wahern (subscriber, #37304) [Link] (5 responses)

Here's the one I had in mind: Apple Computer, Inc. v. Franklin Computer Corp., 714 F. 2d 1240 (3rd Cir. 1983). http://scholar.google.com/scholar_case?case=1006320412569...

Wikipedia also has a description, http://en.wikipedia.org/wiki/Apple_Computer,_Inc._v._Fran....

But note that the description on that Wikipedia article has some biased opinion. In the first paragraph the editor conspicuously tries to pin the ruling to the facts of the case. E.g. when the editor says that the court opinion "cited the presence of some of the same embedded strings." That's the editor's spin. As taught in law school that case is often used to represent the idea that software interfaces are copyrightable. Of course, it represents whatever a court says it does; not what academics, scholars, and wikipedia editors say.

RMS on Copyrightability of APIs

Posted May 1, 2012 21:12 UTC (Tue) by wahern (subscriber, #37304) [Link]

By "software Interfaces" I meant APIs (it's just that with an old OS the API is mostly just entry points at fixed addresses). There are several other well known cases concerning copyrightability of GUI elements which is of no use discussing here.

RMS on Copyrightability of APIs

Posted May 1, 2012 21:25 UTC (Tue) by slashdot (guest, #22014) [Link] (1 responses)

That article says "Franklin admitted that it had copied Apple's software but argued that it would have been impractical to independently write its own versions of the software and maintain compatibility".

And "The Court of Appeals overturned the district court's ruling in Franklin by applying its holdings in Williams and going further to hold that operating systems were also copyrightable".

So unless those statement are false, it seems it was about the ability to copyright the OS implementation, not the API.

RMS on Copyrightability of APIs

Posted May 1, 2012 22:56 UTC (Tue) by wahern (subscriber, #37304) [Link]

One way to read the case is that the defendant tried to argue that the OS (as compiled into object code) was so simple that it did not possess the requisite creativity necessary for copyrightability. Remember, we're not talking about a modern, multi-million line kernel, but an extremely simple OS doing basic I/O in way almost entirely dictated by the hardware (i.e. dictated under the circumstances). The creativity involved here is arguably commensurate with that which goes into creating an API. The amount of code we're talking about here is substantially less than many header files today, and certainly far less than the Java API. If something lacks sufficient creativity than verbatim copying is of no matter. This is why Google and RMS argue that Google can basically copy kernel headers files verbatim, strip the GPL notice and, somewhat ironically, affix their own license.

In this case, that argument about simplicity and lack of creativity failed miserably. Our concepts of operating systems has changed considerably over the past 30+ years, so today the case on its face doesn't look like it's concerned with APIs per se. In any event, the questions and holdings of the case make it poorly suited to use in a brief in a real case. But because the question of copyrightability of APIs has received little actual attention in the courts, this case still stands out in the academic literature.

RMS on Copyrightability of APIs

Posted May 2, 2012 3:00 UTC (Wed) by mjg59 (subscriber, #23239) [Link]

The early 80s were an era where it wasn't even entirely clear under US law that software was copyrightable. Franklin didn't attempt to demonstrate that they'd independently come up with equivalent code. They didn't attempt to demonstrate that the extent of their copying was to ensure compatibility and no more. They merely asserted that an independent implementation was impossible. Their defence was effectively "Yes, we did this, but we should be allowed to do this". It was worth a go at the time, but it's not really terribly surprising that it ended the way it did.

But that's not equivalent to copyright arguments over an API. Apple v. Franklin never really determined whether or not an API was copyrightable - it merely decided that object code *was* and that Franklin had directly copied Apple's object code. It's even indicated that third parties had produced compatible code without direct copying, but that Franklin had simply chosen not to.

I don't think this case covers anything relevant, other than indicating that courts are able to change the status quo. Before this and the Williams case, the assumption was that software might not be copyrightable. It's conceivable that Oracle might trigger a change in assumptions about whether APIs are copyrightable. We just have to hope not.

RMS on Copyrightability of APIs

Posted May 2, 2012 17:53 UTC (Wed) by dlang (guest, #313) [Link]

for what it's worth, there was just a decision in the EU clearly stating that APIs and programming languages (as opposed to implementations of the languages) cannot be copyrighted.

http://web.archive.org/web/20200208061421/http://www.groklaw.net:80/article.php?story=20120502083035371