Google v. Oracle: The Curse of Being Popular?

A Look at Google’s “Lock-In” Argument and Whether User Investment (or Expectation) Should Be Considered In Assessing Copyrightability.

Google is currently petitioning the Supreme Court to hear Google v. Oracle, a dispute over the copyrightability of certain lines of computer code used in Oracle’s Java platform. The case is another gloss on copyright’s most ubiquitous issue – what constitutes protectable expression as opposed to a non-protectable idea or function. As expected, much of Google’s argument concerns the proper jurisdictional limits of copyright law and patent law. However, it was a separate user-centric argument from Google – termed the “lock-in effect” – that jumped out at me. In this post, I want to discuss Google’s lock-in argument and whether user-based considerations, such as user investment or expectations, have a role in determining whether a creative work is copyrightable.

For those unfamiliar with Google v. Oracle, the case concerns source code Google utilized in its Android platform for mobile devices. In the 1990s, Sun Microsystems (which would be later acquired by Oracle in 2010) developed the Java programming language and Java software platform. The Java language was a series of words and symbols along with syntax rules to allow programmers to write code. The Java platform was a means to allow programmers to write code that would run on different operating systems and devices – hence it’s slogan, “Write once, run anywhere.”

Included in the platform were ready-to-use, pre-written Java programs (or “packages”) to perform common functions, which would allow programmers to focus on developing their own programs. The packages consisted of two types of source code – declaring code and implementation code. The declaring code (also known as a “method header”) is an expression that identifies the pre-written function. The implementation code is the step-by-step instructions for carrying out the declared function.

To promote its platform, Sun (and Oracle) created a two-tier licensing scheme. The first tier was an “open source” license where programmers could use the software packages, so long as they contributed back their creations. The second tier was for those users who wished to utilize the Java platform to create their own programs, but wanted to retain control of their proprietary work. There, a business, like Google, could purchase a license to use Oracle’s declaring code but develop its own implementation code.

Although Google and Sun (and later Oracle) entered negotiations about a possible licensing deal for the Java platform, no agreement was reached. Notwithstanding the failed negotiations, Google made verbatim copies of the declaring code in thirty-seven packages created by Oracle, and duplicated the structure, sequence and organization of the Java Application Programming Interface into its Android platform. Google did, however, write its own implementation code.

Oracle filed suit, alleging Google’s Android platform infringed copyrights and patents Oracle held in the Java platform. On the copyright issue, the district court held that Oracle’s method headers were not subject to copyright protection. According to the court, “[s]o long as the specific code used to implement a method is different, anyone is free under the Copyright Act to write his or her own code to carry out exactly the same function used . . . in the Java [platform]. It does not matter that the declaration or [packages] are identical.” Oracle Am., Inc. v. Google, Inc., 872 F. Supp. 2d 974 (N.D. Cal. 2012). On appeal, the Federal Circuit reversed, finding that Oracle’s packages were original and entitled to copyright protection. Oracle Am., Inc. v. Google, Inc., No. 13-1021 (Fed. Cir. May 9, 2014).

Although the Federal Circuit remanded the case to determine the issue of fair use, Google filed a petition for writ of cert directly to the Supreme Court on October 6, 2014 (available here). As noted earlier, the Supreme Court has yet to decide whether it will hear the case, but did invite the U.S. Solicitor General to submit a brief in the case.

The issue itself is not particularly novel. Once you strip the case of its references to smartphones and operating systems, the question is simply, “Is this a copyrightable work?”

Under Section 102 of the Copyright Act, copyright protection extends to original works of authorship fixed in a tangible medium. However, protection does not “extend to any idea, procedure, process, system, method of operation, concept, principle, or discovery, regardless of the form in which it is described, explained, illustrated, or embodied in such work.” This reflects copyright’s basic distinction – an expression is protected, while an idea is not.

In its petition, Google argues that Oracle’s method headers, while original, are ineligible for copyright protection. According to Google,

There is no dispute, for example, that the implementing code that instructs a computer how to perform a method may be subject to copyright protection. . . . But whether the method headers are entitled to protection is exclusively a question for patent law because the headers constitute, or embody, a system or method of operating the pre-written programs.

Oracle Am., Inc. v. Google, Inc., No. 14-410, Petitioner’s Writ for Certiorari at p. 27.

Why do the method headers constitute a system or method of operation? Google makes two arguments. First, the code was “necessary” to access certain functionalities in the Java platform. Second,

[p]rogrammers have invested significant time and effort in learning the Java language, including the shorthand commands. But now, long after Sun lured computer programmers into the Java community and after any patent protection likely would have expired, Sun’s successor Oracle is attempting to build a wall around use of Java’s method headers.

Id. at p. 32. (Internal citation omitted).

The latter is the “lock-in” argument. According to Google, programmers’ investment of time and effort caused them to be bound to the Java platform. Google included the shorthand commands to accommodate (if not capitalize on) programmers’ familiarity with the language. To permit Oracle to copyright its shorthand commands after programmers were “lured” into the Java community would essentially amount to fraud. Instead, the shorthand commands should be considered the basic vocabulary of the Java language, which the copied method headers operate.

The (perhaps unintended) thrust of Google’s argument is that downstream factors, such as how consumers interacted with the work, could impact whether copyright protection ever attached to the work in the first place. This notion would carry a steep price for copyright holders; namely, that user considerations, such as popularity or familiarity, could eventually invalidate their copyright interests and force their works into commons. (Ed. Note: An example of one such argument and its flaws can be found in Terry Hart’s post on Martin Luther King, Jr. and his I Have A Dream speech). Is there any precedence for such a concept? In intellectual property, Google’s argument does resemble the trademark concept of genericide.

In trademark law, trademarks protect a business’s use of words, symbols, or images that identify the business as a particular source of a good or service used in commerce. The theory behind doing so is that protecting a business’s mark will, in turn, promote market efficiency. Mark P. McKenna, The Normative Foundations of Trademark Law, Vol. 82:5 Notre Dame L. Rev. 1840, 1844 (2007). Consumers can distinguish between the goods and services of competing businesses, reducing consumer search costs, and consumer recognition of brand origin will encourage businesses to invest in product (and brand) quality. Id. at 1844-1845.

But not all terms can be trademarked. Specifically, generic terms. Why? Because “[g]ranting exclusive use of a term that is the only publicly recognized name for a category of goods unfair limits competition because it confers a monopoly on the trademark owner by rendering competitors unable to describe their goods effectively. . . Competitors are hampered in the sale of their goods, and consumers cannot discover whether products similar to the trademarked goods are available from other sources.” Jacqueline Stern, Genericide: Cancellation of a Registered Trademark, 51 Fordham L. Rev. 666, 675 (1983) (available at

Some trademarks can become generic, however, and lose protection. Under Section 1127 of the Lanham Act, a trademark loses its legal protection if the mark becomes the common name of the relevant product or service, as used by the consuming public or commercial competitors. See also Bayer Co., Inc. v. United Drug, Co., 272 F. 505 (S.D.N.Y. 1921) (the test is one of public perception). This is known as genericide and it has led to the terms aspirin, thermos, escalator, and trampoline losing their legal status as trademarks.

Essentially, once the public is locked in (or invested) to a certain understanding – that the mark merely represents a generic term for the good or service – it serves the market to treat the mark as a generic term. There is a sound rationale for that conclusion. If marks are intended to distinguish goods and services of competing businesses, but users are only familiar with a particular term as an indicator of a common category of goods or services, not a particular business, the mark serves no value as a source of origination. Instead, continued protection of the mark could disrupt market balance by conferring an unfair competitive advantage to a business or creating opportunities for consumer misunderstanding.

Is there any comparable rationale in copyright law for granting weight to user investment or expectation in determining whether a work is copyrightable? There doesn’t appear to be.

Unlike trademark law, the primary objective of copyright law is to incentivize creation. Feist Publications, Inc. v. Rural Tel. Serv. Co., 499 U.S. 340, 349-350 (1991). Google’s position appears to undermine that objective.

First, downstream considerations would distort the legal rights of copyright holders and the public. Under the Copyright Act, copyright protection attaches once an original work is fixed in a tangible medium. 17 U.S.C. § 102(a). User investment and expectation, however, place importance in considerations that could not exist until after the work was fixed. Would copyright protection initially attach to Oracle’s declaring code at its release, but later enter the public domain when it became popular? If so, when does a work become so popular that it loses copyright protection?

Second, if downstream considerations could strip Oracle of its work’s copyright protection, what incentive would Oracle have to provide that additional content? And, what about other creators? While the pre-written programs were certainly an attractive perk to programmers, it is undoubted that it also granted programmers leeway to focus on other original creations.

Consider the merits of Google’s argument outside of the context of computer programming. A stage performer no doubt invests time and energy in learning his or her lines for a play. Does that mean that would be an appropriate rationale to divest the screenplay’s author of their copyright interest in the work? Most scholars would certainly agree it would not.

Furthermore, Google’s position lacks statutory support. Unlike Section 1127 of the Lanham Act, there is no provision in the Copyright Act that suggests user investment or expectation can deprive a work of copyright protection. Instead, copyright law considers the work itself. To that end, Congress emphasized that the appropriate dividing line for copyright protection is whether a creation is an expression, not merely an idea.

The closest analog to user expectation in copyright is scènes à faire – the principle that copyright protection does not extend to expressions or natural similarities that are mandated by the genre. The common example is that, in the crime genre, the inclusion of disgruntled cops, foot chases, or conversations in a bar or precinct would, generally, not be copyrightable elements as they are stereotypical ideas that naturally flow from the genre itself. See Walker v. Time Life Films, Inc., 784 F.2d 44 (2d Cir. 1986). While a consumer may come to expect a genre to embody certain ideas and generalized expressions, scènes à faire examines the genre itself, not user expectations. Moreover, those genre-mandated elements remain distinct from any expressions not necessary for the genre. The latter of which remain copyrightable expressions.

Some courts have applied the concept of scènes à faire to computer programs, finding that there are certain elements and arrangements that naturally flow from the medium. See e.g., Computer Assoc. Int’l, Inc. v. Altai, Inc., 982 F.2d 693 (2d Cir. 1992) (finding it customary to list variable declarations at the beginning of a source code). But while general ideas of how to structure and organize code are unprotected, they remain “distinct from the underlying computer code because while code is necessary for the program to work, its precise formulation is not.” Lotus Dev. Corp. v. Borland Int’l, Inc., 49 F.3d 807, 816 (1st Cir. 1995), aff’d 516 U.S. 233 (1996). Paradoxically, Google cited Lotus as support for its position.

This is not to say that there is no role in copyright for user considerations. Specifically, those considerations appear apt, if not required, for a thorough application of the fair use defense. 17 U.S.C. § 107(1) (analyzing “the purpose and character of the [allegedly infringing] use”). To some, it may be a distinction without a difference to analyze user considerations during the fair use analysis instead of when assessing copyrightability. That distinction, however, could have a significant impact on an author’s rights and legal remedies as a determination that an element is not copyrightable puts it into the commons without any consideration of how it is subsequently used.

For me, the issue is where Oracle’s declaring code falls on the idea – expression spectrum. I don’t believe user-focused considerations assist that inquiry. Copyright law doctrines, such as scènes à faire, appear well suited to the task by delineating which expressions are copyrightable at the time of creation. The focus on ex-post considerations, such as user familiarity, undermine copyright’s basic principles and threaten to penalize a copyright holder for creating a successful work that has achieved market saturation.


One thought on “Google v. Oracle: The Curse of Being Popular?

  1. Frankly, I find Google’s argument that headers “constitute, or embody, a system or method of operating the pre-written programs” even less convincing than their argument for genericide in copyright. It is demonstrably untrue, as a matter of fact.

    Computer programs are a bit tricky from a copyright perspective, since they are both a process that does something, and a specific way that this process is described. Only the latter part is protected under copyright and rightly so. The classification of computer programs as “literary works” is quite natural – what is protected is how the program was written, not what it actually does.

    Where I think the district court erred in Oracle is in failing to distinguish what the headers do from how they are written – ruling that only the implementing code is subject to copyright protection. It must be pointed out that the “literary” part of headers is, for the most part, completely arbitrary.

    As part of the process of operating a computer program, the header must do two things: identify the program (or program fragment, commonly known as a method) and specify the parameters of the program – which also receive identifying names. The number and type of parameters could possibly be considered a part of the “process” side of programing, but the identifiers used most certainly cannot be. There is no requirement to use any particular identifier – other than the rules as to what may be considered a valid one.

    Anyone is certainly “free under the Copyright Act to write his or her own code to carry out exactly the same function used . . . in the Java [platform].” However, doing so does not require “the declaration or [packages] [to be] identical”. Which is why it does matter if they are.

    Google could easily preserve the process-necessary structure of the copied packages (meaning: the classes, methods and parameters etc. needed to replicate their function) without using the same identifiers for these. They could, esssentially, say the same thing “in their own words”. Given that programs are protected as literary works, this seems to me a key factor.

    Google specifically chose not to do this. They did so because the identifiers (words) used by Sun/Oracle were already well known to Java developers. Phrased in terms of more standard literary works, it is essentially the same thing as using settings and characters (the headers) from a popular work, in order to create your own story (the implementing code). If we were to apply it to books, Google’s work would clearly be derivative and infringing. (Admittedly, some of the copied parts might have been scènes à faire).

    It is exactly for this reason they are trying to advance the “lock-in” argument and counting on the courts to be unfamiliar with the intricacies of computer programming. Of course, accepting the proposition would defeat the whole purpose of copyright – especially provisions relating to derivative works – by ruling that any expression that becomes popular also becomes fair game. Such a view is irreconcilable with the proposition that copyright promotes progress by assuring creators rewards if they create popular (and thus valuable) works.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s