The Problem Isn't the Use of Analogies but the Analogies Courts Use

Genevieve Lakier is an Assistant Professor of Law at the University of Chicago Law School.

The search engine is one of the most important technologies of the twenty-first century. Just as was true of the television, the motion picture, and the radio before it, the widespread dissemination of this new technology is reshaping the speech environment in which we operate in profound, and often unexpected, ways. Search engines raise, as a result, a whole host of novel constitutional, as well as regulatory, questions.1 Yet, as Heather Whitney’s interesting and incisive paper demonstrates, courts confronted with free speech challenges to laws that regulate search engines tend to analyze them by means of an often-unpersuasive analogy to older technologies, chief among them the newspaper. This is a problem because it makes complex issues seem unduly simple.

Take, for example, one of the more influential of the cases Whitney discusses in her paper, Zhang v., Inc.2 As Whitney notes, the plaintiffs in that case sued the Chinese internet company Baidu after it prevented their pro-democracy websites, and other websites that provided information about the democracy movement in China, from appearing in the results of the very popular search engine it operated.3 The plaintiffs claimed that Baidu censored the search results at the behest of the Chinese government, which did not want its citizens exposed to democratic ideas, and that Baidu’s actions violated their federal and state civil rights.4 The district court dismissed the complaint because it found Baidu’s decision not to include the plaintiffs’ websites in its search engine results to involve the same kind of editorial discretion as the decision by a newspaper editor not to publish an article in the newspaper.5 Because the First Amendment clearly immunizes the latter decision against the kind of civil rights claims the plaintiffs asserted, the court concluded that it immunizes Baidu’s decision as well.

The Baidu court strongly suggested that the conclusion it reached was inevitable—and obvious—given existing precedents. It noted, for example, that “[t]o allow [the] suit to proceed would plainly violate the fundamental rule of protection under the First Amendment, that a speaker has the autonomy to choose the content of his own message.”6 And the court suggested that although the constitutional analysis might be different in a case in which the search engine was not alleged to have exercised “editorial control over its search results,” the First Amendment “plainly shields Baidu from Plaintiffs’ claims.”7 Two “plainlies” in one paragraph! How much more straightforward can you get?

In fact, there is reason to doubt that the First Amendment issues raised by the case were as plain and simple as the district court suggested. For one thing, if Baidu truly was acting at the behest of the Chinese government — as the plaintiffs alleged in their complaint — was it actually exercising its own expressive autonomy when it removed their websites from its search engine results? Or was it instead acting as an agent of the censorial Chinese state?8 The answer to this question surely matters — or should matter — to the First Amendment analysis.

Even if one assumes that Baidu acted on its own volition when it removed the plaintiffs’ websites from its search results, it is far from clear that requiring it to list the plaintiffs’ websites in its search results would create the same kinds of expressive harms as would requiring a newspaper to print articles it did not choose. This is because of the important differences in the relationship between search engines and the websites they find, on the one hand, and newspapers and the articles or advertisements they print, on the other. Newspapers are legally responsible for the defamatory or otherwise unlawful content of the articles and advertisements they publish, but search engines are not.9 Newspapers also tend to be associated in readers’ minds with the ideological and aesthetic viewpoint of the articles or advertisements that fill their pages. This is not true of search engines and the websites they link to. How could it be? The list of websites produced by a single search will often express a cacophony of competing views; it would be very difficult, as a result, to glean from the actions of the search engine a discernible perspective. From both a legal and a cultural standpoint, then, the websites that populate search engine results are not understood to express the viewpoint of the search engine that finds them. And this, in turn, suggests that preventing search engines from removing websites from search results would not undermine the search engines’ ability to “control the content of [their] message.”10 It would not create, in other words, the kinds of harms the Court was worried about when it struck down the right-of-reply statute in Miami Herald Publishing Co. v. Tornillo.11

Whitney is entirely correct, therefore, that even if there are important similarities between newspapers and search engines, there are important dissimilarities that complicate the straightforward analogy the district court relied on in Baidu. These dissimilarities call into question whether precedents like Tornillo compel the conclusion that the court reached in Baidu and that other courts have reached as well: namely, that, because of its concern with the expressive autonomy of editorial speakers, “the First Amendment fully immunizes search-engine results from most, if not all, kinds of civil liability and government regulation.”12


I am less convinced by the conclusion that Whitney herself reaches from the problematic analogical reasoning employed in cases like Baidu: namely, that courts should whenever possible “address First Amendment coverage questions through the lens of normative theory and not through a cacophony of ill-suited analogies.” Analogical reasoning plays an incredibly important role in our law, including in the law of the First Amendment.13 Given the centrality of analogical argument to legal practice, it is hard to believe that judges would take up Whitney’s suggestion to chuck the analogies and restrict themselves instead to a more straightforward normative analysis of the constitutional and other interests at stake in these cases. Nor, I think, should they. The interests that the First Amendment is said to advance — Whitney identifies two (democracy and thinker autonomy), but there are others one could name (the search for truth, for example) — tend to be articulated at an extremely high level of generality. Their application in concrete situations will, as a result, almost always be up for debate. Analogical reasoning helps constrain judicial discretion on these matters and thereby brings some degree of consistency and predictability to the law.14 It also allows judges who cannot agree on what interests the First Amendment advances to reach agreement in particular cases.15 And, by concretizing the abstract, it helps illuminate complexities that a court more focused on abstractions might miss.16 Perhaps most importantly, analogies encourage the creation of generalizable rules: rules that extend beyond a particular fact pattern or kind of speech act. They therefore make it more difficult for courts to treat speakers differently based on a political or cultural or idiosyncratic judgment of the value of their speech. Certainly this is the role that analogies have historically played in free speech law.17

There are thus both First Amendment–specific and more general legal process reasons to believe analogies play a useful, and not just a familiar, role in judicial decisionmaking. This is not to say that analogies cannot go wrong. Nor is it to contest the fact that, as Whitney points out, analogical reasoning is fundamentally incomplete. It is absolutely true that, to know whether two things are similar or dissimilar, one must have a view of what the relevant criteria are, against which similarity is to be measured. But what the incompleteness of analogical reasoning means is not that analogies cannot provide a useful aid to judicial analysis; what it means, instead, is that analogies will prove useful only to the extent they are used thoughtfully, to illuminate the similarities and dissimilarities that matter for the purposes of the law.

The problem with the decision in Baidu need not be seen, therefore, as a consequence of the court’s reliance on analogical reasoning per se. It is better understood as a consequence of the court’s reliance on an overly formal analogy between newspapers and search engines, one that fails to take into account the very different functions that newspapers and search engines play in the contemporary public sphere. It may be true that both newspapers and search engines rank and order information that they then provide to the public—that they perform in this sense an editorial function. But they do so for very different purposes. Newspapers articulate views, and provide information, that contribute to the “uninhibited, robust, and wide-open . . . debate on public issues” that the First Amendment protects against government control.18 Search engines, in contrast, allow users to navigate among different sources of information and perspective.

This suggests that the most illuminating analogy for purposes of constitutional analysis is not to the newspaper but instead to the cable provider or, less directly perhaps, the telephone company—to other technologies that provide users access to views and information that those technologies neither author nor control. It is somewhat surprising that Whitney does not mention the cable-provider analogy as a possible alternative to the newspaper analogy she criticizes, since Oren Bracha and Frank Pasquale first suggested it almost ten years ago in an article that the district court discussed in Baidu.19 And it is a very good analogy in many respects: Like cable companies, search engines provide access to the speech of others; like cable companies, search engines exercise some degree of editorial discretion over whom they provide access to. 20 They both speak and act as a conduit of speech, in other words.

The analogy is, like all analogies, not a perfect one. Because cable providers have to create at least part of the physical infrastructure of their cable system before they can operate, barriers to entry in the cable market may be significant.21 The same is not true of search engines. For these companies, the only significant barrier to entry is the development of a search algorithm (and perhaps also access to sufficient marketing dollars to attract consumers away from competitor search engines). The consequence is that cable companies frequently possess a degree of monopoly power that search engines do not.22

The power of the analogy nevertheless arises from the fact that it takes into account not only the formal features of search engines’ speech, but also the functional role they play in the public sphere. In so doing, it makes clearer why it is we might worry about an interpretation of the First Amendment that immunizes search engines from most kinds of regulation—and why we might worry, in particular, about a constitutional rule that prevents the government from being able to prohibit the kind of content-based discrimination at issue in Baidu. Such a rule is problematic because, just like cable operators, search engines have tremendous influence over “a critical pathway of communication”—influence that they can use to “to restrict . . . the free flow of information and ideas” when it suits their interests to do so.23 Rather than a vindication of free speech values, in other words, what the analogy to the cable operator suggests is that the outcome in Baidu and similar cases poses a threat to freedom of speech. What it also suggests, as a result, is an alternative approach to the constitutional review of search engine expression: one that vests the government with significant, albeit by no means unlimited, power to regulate when doing so serves important (in this case, speech-promoting) ends.24 This is all to the good, I think, given the power that search engines possess in our contemporary media landscape.


Ultimately, therefore, I disagree with Whitney’s primary normative conclusion. I remain convinced that analogies play a vital role in free speech law because they force courts to toggle back and forth between the concrete and the abstract and thereby make possible a better understanding of both new legal questions and the normative principles we use to resolve them. I also think the most powerful response to the shortcomings of cases like Baidu is not to chuck the analogies but instead to suggest why another analogy works better.

Whitney’s paper nevertheless performs a valuable function by suggesting how difficult the process of analogical reasoning can be. It also makes clear—in a way I find heartening—that there remains considerable opportunity for courts to think more deeply about what analogies best make sense of the multiple interests at stake when the government regulates the power of search, and about the constitutional rules that consequently apply.

1 It is worth noting that the First Amendment problems raised by search engines are only the tip of the iceberg. For more discussion of the regulatory questions that search engines raise, see James Grimmelmann, The Structure of Search Engine Law, 93 Iowa L. Rev. 1 (2007).

2 10 F. Supp. 3d 433 (S.D.N.Y. 2014).

3 Id. at 434–35.

4 Id. at 439.

5 Id.

6 Id.

7 Id. at 439–40.

8 There is considerable evidence to support the allegation in the plaintiffs’ complaint that internet companies like Baidu face pressure from the Chinese government to help enforce its repressive and censorial policies. See, e.g., U.S.–China Econ. & Sec. Review Comm’n, 2010 Report to Congress 231–32 (2010), (noting that “the Chinese government outsources Internet censorship to the private sector” and that “Baidu is now very much a part of China’s comprehensive oppression on the Internet”).

9 See 47 U.S.C. § 230(c)(1) (2012) (absolving search engines and other “interactive computer services” of liability as the “publisher or speaker of any information provided by another information content provider”).

10 Baidu, 10 F. Supp. 3d at 437.

11 418 U.S. 241, 258 (1974) (striking down a Florida right-of-reply law on the ground that it undermined a newspaper’s ability to decide, among other things, how it would “treat[] . . . public issues and public officials”).

12 Baidu, 10 F. Supp. 3d at 436–37; see also La’Tiejira v. Facebook, Inc., 272 F. Supp. 3d 981, 995 (S.D. Tex. 2017); e-ventures Worldwide, LLC v. Google, Inc., 188 F. Supp. 3d 1265, 1274 (M.D. Fla. 2016); Langdon v. Google, Inc., 474 F. Supp. 2d 622, 629–30 (D. Del. 2007).

13 This is true when it comes to cases dealing with First Amendment coverage questions, as it is true of cases that deal with other areas of First Amendment law. Whitney mentions the recent decision in Packingham v. North Carolina, 137 S. Ct. 1730 (2017), but this is by no means the only First Amendment opinion that relies upon analogical reasoning to determine how broadly the First Amendment applies. See, e.g., Brown v. Entm’t Merchs. Ass’n, 564 U.S. 786, 790 (2011); Winters v. New York, 333 U.S. 507, 510 (1948).

14 See Cass R. Sunstein, On Analogical Reasoning, 106 Harv. L. Rev. 741, 783 (1993) (“[B]y constraining the areas of reasonable disagreement . . . analogical reasoning introduces a degree of stability and predictability. These are important virtues for law, and they sharply reduce the costs of reaching particular decisions.”).

15 See id. at 782 (“[R]easoning by analogy may have the significant advantage of allowing people unable to reach anything like an accord on general principles to agree on particular outcomes. Sometimes it is exceedingly difficult to get people to agree on the general principles that account for their judgments. But it may be possible for them to agree on particular solutions or on low-level principles.”).

16 See id. at 779 (“[A]nalogies . . . are not simply unanalyzed fact patterns; they are used to help people think through contested cases and to generate low-level principles. In this way they have a constitutive dimension, for the patterns we see are a product not simply of preexisting reality, but of our cognitive structures and our principles as well.”).

17 Consider here the 1948 decision in Winters v. New York, 333 U.S. 507 (1948). A bookdealer was convicted of violating a New York law that prohibited the distribution of what were colloquially known as “true crime” magazines—magazines that provided, often in salacious detail, stories and pictures of criminals and the crimes they committed. Id. 508. Both lower courts upheld the conviction against the bookdealer’s First Amendment challenge because they found that the magazines in question contained so little “social value” that their unimpeded distribution did not contribute to the First Amendment’s democratic or social purposes. People v. Winters, 63 N.E.2d 98, 101 (N.Y. 1945), aff’g 48 N.Y.S.2d 230, 234 (App. Div. 1944). The Supreme Court reversed because it found that these magazines were, in relevant respects, like other kinds of literature the First Amendment protected and that their democratic or social value could not therefore be so easily dismissed. Winters, 333 U.S. at 510.

18 New York Times Co. v. Sullivan, 376 U.S. 254, 270 (1964).

19 Oren Bracha & Frank Pasquale, Federal Search Commission? Access, Fairness, and Accountability in the Law of Search, 93 Cornell L. Rev. 1149 (2008).

20 As the Supreme Court noted in Turner Broadcasting System, Inc. v. FCC, “[t]hrough ‘original programming, or by exercising editorial discretion over which stations or programs to include in [their] repertoire,’ cable programmers and operators ‘see[k] to communicate messages on a wide variety of topics and in a wide variety of formats.’” 512 U.S. 622, 636 (1994) (third alteration in original) (quoting Los Angeles v. Preferred Commc’ns, Inc., 476 U.S. 488, 494 (1986)).

21 A recent report found that capital investment by providers of cable television and internet equaled $87.2 billion in 2015. Free Press, It’s Working: How the Internet Access and Online Video Markets Are Thriving in the Title II Era 5 (2017),

22 See Turner, 512 U.S. at 656 (“[S]imply by virtue of its ownership of the essential pathway for cable speech, a cable operator can prevent its subscribers from obtaining access to programming it chooses to exclude. A cable operator, unlike speakers in other media, can thus silence the voice of competing speakers with a mere flick of the switch.”).

23 Id. at 657. For an influential, relatively early discussion of the power that search engines possess to “shape the web,” see Lucas D. Introna & Helen Nissenbaum, Shaping the Web: Why the Politics of Search Engines Matters, 16 Info. Soc’y 169 (2000).

24 Turner, 512 U.S. at 653 (rejecting the proposition that a law requiring cable operators to transmit local broadcast stations “trigger[s] strict scrutiny because [it] compel[s] cable operators to transmit speech not of their choosing” and instead upholding the law under an intermediate standard of review).