Friday, December 5, 2014

Cisco suing Arista over set of 500 commands: original interface material IS copyrightable

Cisco agrees with Google on a lot of patent-related issues (so do I), but strongly disagrees with it on interface-related copyright questions (so do I, too). That's one of my conclusions from having looked at the copyright infringement complaint Cisco filed today against Arista. Cisco simultaneously brought a patent suit and a copyright suit, alleging "promotion of theft."

Google and the amici curiae (1, 2) supporting its Supreme Court petition in Oracle's Android-Java copyright case incorrectly claim that the industry at large had for a long time relied on the non-copyrightability of anything related to application programming interfaces (APIs). No matter how many IP professors and programmers the Electronic Frontier Foundation can persuade (or doesn't even have to persuade because they're Google employees or otherwise Google-aligned) to put their good names under bad claims, it's simply not true. Today's copyright infringement action by Cisco is just the latest example of how the IT industry has actually relied on copyright protection for all creative, expressive code (whether or not interface-related) ever since.

I said Cisco is just the latest example. Oracle received overwhelming support from the software industry last year in the form of amicus briefs submitted to the Federal Circuit. Cisco wasn't active at that time, but its Arista complaint is, silently and indirectly, a wholesale endorsement of Oracle's core positions on copyright.

If Cisco is right on this one (and I am already convinced that it is), the expression, organization and hierarchies of at least 500 multi-word commands from Cisco's command-line interface (CLI) is copyrightable. I don't have a position on whether Arista infringed, but Cisco alleges outright copying and is rather specific in its pleadings.

I have looked at Exhibit 1 to Cisco's copyright complaint (you can find it further below), which lists those 500 multi-word commands. On that basis I have a few observations to share regarding copyrightability:

  1. This stuff is expressive, original material to me. The authors of this set of multi-word commands could undoubtedly have written and structured it in a practically infinite number of different ways. That's why I believe this material does enjoy copyright protection under long-standing U.S. law. And it isn't deprived of copyright protection only because of someone's subsequent choice to provide programmers with a "familiar" interface.

  2. Just like any body of copyrightable material, it obviously contains some elements that wouldn't be copyrightable if taken out of context and viewed absolutely separately (just like this blog post is composed of words and I don't own copyright in any one of those words, but in the text as a whole). For example, "ip host" all by itself isn't copyrightable. Same with "show inventory." Arista could have copied one or two of those and Cisco couldn't complain if that were the case. But when one looks at the whole list of 500 multi-word commands, many of which truly involve creative choices (for example, "show ip igmp snooping querier" or "spanning-tree potfast bpdufilter default"), the threshold for copyrightability is easily met.

  3. While the threshold for copyright protection is a non-issue in my view for the 500 CLI multi-word commands Cisco is asserting against Arista, the degree of creativity that went into the 7,000 lines of Java API declaring code is undoubtedly far greater. Most Java function and variable names also consist of multiple words each, and then there are additional parameters. I believe there are strong arguments for saying that the average line from the Java API declaring code is more expressive than the average CLI command, but even if one disagreed with me and only considered them to be at a similar level, the Java API material would be far more massive and far more creatively-structured.

    As for "massive," it's not about quantity being a substitute for quality; I'm saying this because the size of such a work has exponential implications for the number of alternatives that the authors could have chosen.

    In a recent blog post I noted that Google's supporters increasingly recognize the flaws of the overturned ruling from the Northern District of California (where Oracle filed in 2010 and Cisco filed today), and one of them (who demonstrates enormous strength and confidence by conceding he had overlooked one or more key issues when he originally sided with Google) wrote that "[o]ne thing Oracle really had going for it [...] is the sheer size and complexity of the Java API library", going on to say that "surely somewhere in there is a spark of creativity."

    As for "far more creatively-structured," those 500 multi-word commands are simply an alphabetical list, while the asserted Java APIs were taken from 37 packages, and the decision on which package should contain a particular element is also a creative act that simply doesn't occur when someone creates an alphabetical list. (Cisco would perhaps argue that even this alphabetical list has some sort of structure because, for example, one could view all multi-word commands starting with the same word, like "show", as a kind of package. But those Java API packages involve a more creative kind of structure.)

    So, to put it that way, if Cisco prevails on its CLI claim, there's absolutely no way that Google is right with its non-copyrightability position in the Oracle case. (But theoretically--though I don't think that's where it is--the legal threshold could fit comfortably in between Cisco's 500 multi-word commands and the 7,000 lines of Java API declaring code, in which case Cisco would lose and Oracle would prevail.)

  4. Cisco's complaint disagrees with the "Google camp" in an additional, highly interesting way. While Google's supporters blame Oracle's appellate victory on the Federal Circuit and continue to claim that Ninth Circuit law supports their position and that the Federal Circuit allegedly got Ninth Circuit law (which it had to apply here) wrong, Cisco is demonstrably convinced that it will prevail if and when the case is appealed to the Ninth Circuit: if it had brought a combined patent and copyright complaint (as Oracle did, though patents are no longer at issue in that dispute), the copyright part would also have gone to the Federal Circuit in the event of an appeal, but Cisco filed a copyright complaint that is separate from the simultaneous patent infringement complaint, undoubtedly making the Ninth Circuit the future appellate forum.

  5. One last observation. It has unfortunately become pretty common that patent holders allege "copying" even in cases that actually involve independent invention and legitimate kinds of imitation and, for the most part or entirely, lawful competition. I oppose that kind of stretch, though I do recognize that willful and reckless infringement does sometimes occur (and in those cases, "copying" isn't necessarily the word I would consider my first choice but may be somewhat appropriate). In Oracle v. Google, the patent case might have fared better under a different judge (even in the same district, thinking of Judge Koh, for example), but it's probably no coincidence that in cases involving literal copying, copyright is an even more powerful intellectual property right to enforce than patents. Cisco has just filed its two complaints and it remains to be seen where those cases go, but at this stage it's already clear that copyright is not just an afterthought but instead an absolutely central part of Cisco's litigation strategy.

Here's Exhibit 1 to Cisco's copyright complaint with the list of those multi-word commands:

14-12-05 Exhibit 1 to Cisco v. Arista Copyright Infringement Complaint by Florian Mueller

If you'd like to be updated on the smartphone patent disputes and other intellectual property matters I cover, please subscribe to my RSS feed (in the right-hand column) and/or follow me on Twitter @FOSSpatents and Google+.

Share with other professionals via LinkedIn: