In light of yesterday’s jury verdict in Oracle v. Google, I thought it worth revisiting the concept of fair use in copyright, including the factors that go into determining whether or not a use is fair, the issues having such a test presents, and its application to APIs in particular.
What’s Copyright Anyways
I’ve previously discussed what copyright is, what it protects, and how it’s distinguished from a copyright registration, but generally speaking copyright is the exclusive right to a creative work that you’ve made. To get copyright protection, your work has to be: i) original; ii) creative; and iii) expression. That is, it has to be yours, it has to have some creative element (i.e., not be a basic recitation of facts), and it has to be expressed in some manner, which in essence means it must be written down or recorded (or as copyright lawyers like to say: fixed in a tangible medium of expression). Once you satisfy those elements, you now hold the exclusive right to do certain things with your work, like make copies, distribute it, publicly display it, perform it, and create new (derivative) works based upon it. Prior to the 1976 Copyright Act, you had to take certain steps to make sure you got all that protection, but now you get it as soon as the work is created, by default.
Fair use is, then, an exception where a third-party can make use of a work in a way that typically only an owner (or licensee) would be able to do. Some classic examples of fair use are: criticism and comment (like a book review quoting a passage from a book so as to discuss it); news reporting (like a snippet of a home movie showing the events of a major news story); and parody (like 2 Live Crew’s “Hairy Woman,” a parody of Roy Orbison’s “Pretty Woman”). Fair use also covers a wide variety of internet activity that it’s now difficult to imagine could have ever been questioned, like creating search engines or image thumbnails.
Fair use is a statutory creation under 17 U.S.C. § 107, which sets forth four factors to be considered in analyzing whether any use is fair:
- the purpose and character of the use, including whether such use is of a commercial nature or is for nonprofit educational purposes;
- the nature of the copyrighted work;
- the amount and substantiality of the portion used in relation to the copyrighted work as a whole; and
- the effect of the use upon the potential market for or value of the copyrighted work.
The Problem With Fair Use
The problem with fair use is that the statute sets up a test that is, well, squishy. That is, looking at just the statute, there’s no way of telling whether what you want to do is legal or illegal. You have to weigh the factors and place your use somewhere along the spectrum, and then make a judgment call as to which side you’re closer to. As you can imagine, reasonable, and unreasonable, minds disagree all the time about this.
Placed in the context of litigation, fair use is a defense to a lawsuit that must be proved by a defendant, and whether or not a use is fair is a legal determination that can only be made after development of evidence through discovery, either through summary judgment or a trial. For a defendant, even one making a classic fair use, that means the burden is effectively on you to prove innocence, and for both parties, that means that no final resolution can come without potentially years of litigation and incredible costs. Where potential costs are so high and the outcome so uncertain, the result is a chilling effect on whoever the less-wealthy party is.
Fair Use of APIs
APIs (application program interfaces) are sort of like application-specific languages that allow different pieces of software from different makers to talk to each other. Every piece of software is in essence a series of different commands, and in every piece of software those commands are different. In order for one piece of software to send information to another, it has to know how that piece of software is set up to receive information and use the right command. The API, then, is the set of commands that each piece of software uses. Initially, the trial judge in Oracle v. Google decided that APIs weren’t eligible for copyright protection at all, because they didn’t meet the factors above, but the appeals court disagreed and sent the case back for trial, leaving Google with the lone defense of fair use.
In order for a software developer to make actual use of an API, they have to include the commands from the API in their own code. That is, if someone speaks english, and you want to talk to them, you’re going to have to speak some english. If the english language were protected by copyright, then you’d be committing copyright infringement simply by communicating to an english speaker. Oracle’s position was essentially that this is exactly as it should be, they created the language and if Google wanted to use snippets to communicate, then they should pay a licensing fee. Google’s position was, in essence, that everyone in the development community recognized that that made no sense for the industry, that the common practice was to make use of APIs, and that Oracle only changed its mind when their Android competitor failed.
So Google won (for now, the ruling will no doubt be appealed), which I think is absolutely the right decision. The problem is that who knows how the next case will play out. Even if it stands, this is a trial court decision, unique to its own facts as decided by ten individuals whose thinking we’ll never know. The appeals court ruling, that APIs are subject to copyright protection, will continue to be good law, and that means that anytime a large company writes a small company a cease and desist for making use of its API, the small company will have to decide whether it wants to (or even can) jump into bet-the-company litigation. Where small companies (and big companies) are spending their money on litigation rather than development, innovation suffers.