Next Wednesday, George Mason University is holding a conference on competition, search, and social media. I will be on a panel regarding antitrust remedies, and this post is about a particularly misguided remedy.
Over the past few years, Google’s ubiquity and has success captured the attention of both competitors and regulators. Google’s competitors have raised concerns over the company’s practices in their bid to encourage an investigation and regulatory action against the search engine, with some particularly related to copyright matters. Their complaint: that Google search engages in “theft” of the content of other sites, puts that content in search results, and therefore use the stolen content to generate ad revenue. (For example, see page 29 of this report issued by Google’s competitors including Microsoft and Yelp). Their proposed remedy: to forbid Google from displaying in search results “snippets” from other sites without permission. They claim that Google is forcing them to choose between (a) allowing the company to steal their content and be included in Google search results without negotiating over how the snippets will be displayed or commit “webicide” by removing yourself from Google search results.
This post explains why the copyright theories of Google’s competitors’ are so mistaken. Google’s competitors are wrong on both the complaint and the remedy, both of which betray a deep misunderstanding of copyright law and of how the Internet has developed. Essentially, the competitors suggest that Google cannot exercise the right to fair use without violating antitrust laws—even though fair use is a limitation on copyright grounded in the First Amendment itself, announced by courts, and codified by congressional statute.
Snippets and Fair Use in Copyright Law
We can begin our analysis with this fact: displaying a “snippet” is generally a “fair use”—whether done by Google, Bing, Yahoo, or Facebook and Twitter users. What’s a “snippet” and what’s “fair use”?
A snippet is a small piece of content, usually text, from a website. It is essentially a short quote. In the pre-Internet world, perhaps a short quote for a book review: “Hemingway’s asserts that ‘War is not won with victory.’” Search engines display snippets in their results to help users determine whether a linked page is what a user is seeking. A search for “Apple” will turn this displayed snippet in Google: “iPad is a magical window where nothing comes between you …” The use of snippets is fundamental to how search engines work. Without snippets, navigating the web would be more laborious and time consuming.
Fair use is a legal concept. It is the right to reproduce/copy or display portions of a copyrighted work when certain factors are met. The Supreme Court has declared fair use to be a “traditional contour” of copyright protection, and a “built in First Amendment accommodation.”
There are two main policy reasons for fair use. First, freedom of speech. Copyright laws limit the speech of people who don’t own a copyright. The government can fine or send us to jail for saying, singing, or writing certain things (“speech”) because of the words chosen. For decades, legal scholars and judges have discussed how copyright fosters certain speech and suppresses other speech. Because of the burden on speech, copyright law accommodates free speech in several ways: an ideas/expression dichotomy (ideas cannot be copyrighted) and an originality requirement (facts cannot be copyrighted). But perhaps the most celebrated accommodation for free speech is fair use. Fair use lets people transform, criticize, research, and reference content without needing permission from anyone—government or copyright holder.
Second, economics. There are high transaction costs for clearing the use of every snippet or quote or use of content under copyright—particularly when copyright automatically attaches to almost everything we create. These transaction costs are probably astronomical for search engines—they would need copyright permission from every site on the Internet to display a snippet of their content in search results. Even if almost every site would happily oblige, the transaction costs required for all the permissions would be crippling.
Fair use turns on a context-based four-factor test. The four factors are: (1) the purpose and character of your use (transformative, critical, scholarly uses are more likely fair uses), (2) the nature of the copyrighted work (nonfiction or less creative works more easily support fair use), (3) the amount and substantiality of the portion taken (using a small, not large, portion), and (4) the effect of the use upon the potential market of the copyrighted work (a more limited harm). For “de minimis” uses, of particularly small amounts of the copyrighted work, courts do not even bother to engage in the four-factor fair use analysis. There is simply a “de minimis defense.” (See here.)
Because snippets for search engines are transformative in purpose, use content that is more fact than creative fiction, use only a small amount of content, and have minimal effect on the value of the content (even for Yelp), their use by search engines likely constitute a “fair use.” The few circuit courts that have addressed similar questions have agreed with this analysis. (See here, here, and here.) Indeed, some snippets may even be “de minimis” uses that do not trigger fair use analysis.
No Copyright Violation
If Google’s competitors disagree with my fair use analysis, they could sue Google for copyright infringement to test the question. Copyright penalties are stiff, often thousands of dollars per infringement, depending on the work. So, the competitors have more than enough incentive to sue if they think they can win: Google might have to shell over millions for copyright infringement for little snippets. But Google’s competitors have not sued Google for infringement, despite their complaining. One reason for this is simple: they’d lose their case. The second reason is just as simple: if they were to win their case, and then sites like Bing and Yahoo and most other sites online would be subject to huge copyright penalties for facilitating infringement through snippets. Instead, Google’s competitors do not sue but merely refer to the use of snippet as “theft” and “basically theft” (pages 30-31)—even though Google’s use is most likely legal under the fair use statute and therefore protected under the First Amendment accommodations to copyright.
An Antitrust Violation Nonetheless?
As a result, rather than claim Google’s use of snippets violates copyright law, they claim instead that it violates the antitrust law. This argument is peculiar. To my knowledge, exercising fair use has never qualified as an antitrust violation. Moreover, to my knowledge, preventing a company from exercising fair use has never been imposed (or suggested seriously) as an antitrust remedy. Indeed, far more often, it is the use of copyright, not the exercise of fair use, that is the root of anti-competitive behavior. Tim Wu wrote an article years ago about how dying distribution mediums use copyright and “content lock-out” anti-competitively. The remedy there is generally the compulsory sharing of the content. With Google’s search snippets, however, the competitors want special rules to forbid Google from exercising a right to fair use of content–though their rules would conflict with both the Copyright Act and the First Amendment while dramatically raising Google’s transaction costs to merely display search results today.
We can use Yelp as an example here. Yelp essentially claims that Google’s insistence on following fair use is a coercive agreement that violates antitrust law. Google will agree not to index any site that wishes not to be indexed. There is a even simple convention in the code of a site (robots.txt), as well as new companies offering differentiated services such as Distil.it. But Yelp doesn’t want to exclude Google from crawling and indexing its site. Rather, Yelp wants to have control over how Google displays snippets and information from Yelp in Google search results after Google has crawled the site. This would be analogous to Hemingway sending his book to the New York Times Book Review (i.e. making his book “crawl-able”) but then saying he wanted control over how the Times could quote him—whatever fair use would permit. Indeed, he insists that the Times could only quote pages 3, 10, and 32 of his book and write the review in Helvetica font, not Times New Roman. The Times might respond that Hemingway need not send his book to the Times at all—but that if he does the usual standards of fair use will apply. Could Hemingway then complain that the Times is giving him no choice at all because not being reviewed in the Times is necessary to reach readers?
The Yelp complaining is similar: whatever fair use permits, Yelp argues that Google should instead abide by whatever terms Yelp sets for display and that Google is not allowed to walk away based on the terms demanded–otherwise it would be threatening Yelp with the “webicide” of not being included in Google. The Yelp situation is even more problematic than the Hemingway hypothetical: if Google were forced to display snippets from all sites it crawled as those sites wished, rather than crawling sites without robots.txt and relying on fair use, it would be forced to hand control of its results to millions of conflicting sites without the ability to walk away from unreasonable demands. As a result, Yelp can only be asserting that Google acts anti-competitively by insisting on respecting fair use. There is no precedent (or logic or public policy) supporting this argument.
Courts would be unlikely to haphazardly discard fair use and adopt a radical (unpredictable, perhaps unconstitutional) expansion of copyright’s costs on search engines. It is their proposed remedy–not the long-established “traditional contour” of copyright law known as fair use– that should be discarded.
(Disclosure: I am an adviser to Google, but my ideas here are my own and should not be ascribed to Google.)