Power Without Responsibility: Intermediaries and the First Amendment
by Rebecca Tushnet, Professor of Law
Georgetown University Law Center, rlt26@law.georgetown.edu
76 Geo. Wash. L. Rev. 101 (2008)
Georgetown University Law Center rlt26@law.georgetown.edu
August 2008
NOTICE: THIS WORK MAY BE PROTECTED BY COPYRIGHT
YOU ARE REQUIRED TO READ THE COPYRIGHT NOTICE AT THIS LINK BEFORE YOU READ THE FOLLOWING WORK, THAT IS AVAILABLE SOLELY FOR PRIVATE STUDY, SCHOLARSHIP OR RESEARCH PURSUANT TO 17 U.S.C. SECTION 107 AND 108. IN THE EVENT THAT THE LIBRARY DETERMINES THAT UNLAWFUL COPYING OF THIS WORK HAS OCCURRED, THE LIBRARY HAS THE RIGHT TO BLOCK THE I.P. ADDRESS AT WHICH THE UNLAWFUL COPYING APPEARED TO HAVE OCCURRED. THANK YOU FOR RESPECTING THE RIGHTS OF COPYRIGHT OWNERS.
Power Without Responsibility: Intermediaries and the First Amendment
Rebecca Tushnet*
Introduction
At least since Alexander Meiklejohn wrote that “[w]hat is essential is not that everyone shall speak, but that everything worth saying shall be said,”1 First Amendment theorists have debated the implications of speaker-focused versus audience-focused theories of free speech.2 Jerome Barron’s classic article is, in this vein, deeply concerned with providing citizens greater access to conflicting viewpoints and nonmainstream subject matter, not because speakers with disruptive ideas have a right to be heard, but because we as a society have an interest in hearing them.3
Law and technology help constitute the audience for speech, shaping both what speech reaches an audience and what that audience can do in response.4 An audience-centered theory of free speech, therefore, cannot accept that the First Amendment is satisfied by government nonintervention into the market. Indeed, the concept of nonintervention is incoherent from an audience-oriented perspective because the private property arrangements that law enables will determine what the audience hears and in what manner it will be able to respond. In this essay, I will discuss the law’s shaping role mainly in the context of intermediaries’ claims to control, and simultaneous denials of responsibility for, the content provided by end users.
As Barron recognized, the First Amendment rights of speakers and audiences must be evaluated in the contexts of their relationships to larger structures.5 To the extent that there is a right to speak or a right to hear, who is on the other side of that right? The system of free expression is not atomized, but pervasively structured by conduits such as television broadcasters and Internet service providers (“ISPs”). Here I will focus on (potentially) harmful speech as it relates to claims for greater access to those conduits. Any effective proposal for access rights should deal with the recruitment of intermediaries to police and deter unlawful speech and the many and varied ways in which individual speakers will violate existing laws.
Creating incentives and obligations for intermediaries is a quintessentially legislative task, as Congress has already recognized by enacting various regulations of, and liability protections for, Internet intermediaries. The multiple competing interests involved help make the case that the legislature should be given substantial leeway by the courts in crafting solutions. It is this conclusion, perhaps, that makes access rights so difficult for individualist free speech theories: it would be much easier if there were one right answer that could be enforced by courts. The promise of systemic approaches such as Barron’s is that they reveal how speech works—or fails—in practice; the danger is that we lack the political power or will to structure that system in beneficial ways.
Part I of this essay reviews how Barron’s arguments about the vulnerability of individual viewpoints to corporate control remain salient in a vastly changed communications environment. The default of access to the means of expression has changed, in that it is easier than ever for individual speakers to find a platform that could in theory reach millions. But chokepoints remain. Rather than filtering out unpopular views entirely, Internet-based media are more likely to allow all content by default, but channel attention to favored content, and then suppress specific troublesome speech once it’s brought to the attention of corporate owners. Part I.A considers how Barron’s arguments fare online. Part I.B then recounts some decisions by a popular online journaling service, LiveJournal, that illustrate the continuing importance of intermediaries, and background law, in shaping individual speech.
Part II considers more generally how intermediary liability for users’ unlawful speech does and should affect individuals’ opportunities to reach audiences. Right now, intermediary liability is a patchwork of different rules for different substantive areas. Moreover, from the perspective of access rights, intermediary liability for users’ speech is largely uncoupled from intermediary control over such speech: intermediaries possess power over individual speakers, but they have no corresponding responsibility to individuals for the use or abuse of that power.
My main concern is to show that Congress is free, within rather broad limits, to determine an appropriate intermediary liability regime. The First Amendment does not currently require a particular solution. That being said, if individuals’ speech should not be attributed to intermediaries when it is unlawful, we should at least consider ways in which intermediaries could be deterred from interfering with it when it is lawful. The current regime privileges access providers over both individual speakers and third parties harmed by those speakers’ speech. Sometimes that is a mistake, and it is not one that the First Amendment bars us from correcting. Without change, Barron’s hope for communicative diversity may not be realized, even on the Internet.
I. Talking Together in Rented Rooms
A. Access to the WordPress:6 The Role of the Intermediary
Though Barron saw the press as a group of gatekeepers, he did not speak of them as intermediaries, as is more common today. Barron’s terms were “the press” and individual “speakers.”7 The very term “intermediaries,” as opposed to “the press,” emphasizes that aggregators, compilers, and other more passive conduits are not themselves the source of speech, any more than the New York Times is the source of its ads, letters to the editor, or even stories written by employees or freelancers. As a corporate entity, the Times can adopt some of that speech as its own,8 and its status as a publisher will impose certain legal duties on it,9 but before the Times can fill its pages it ultimately needs people to provide speech. Starting from the proposition that speech comes from people, not companies, Barron argued that more and different people should have access to the apparatus of speech distribution in order to correct for predictable and harmful distortions in the deliberative process.10 Indeed, he treated even major media outlets like the Times as conduits—profit-seeking entities that have no inherent interest in the particular speech they carry.11
Barron was concerned with those whose ideas were unacceptable to the mainstream media, and who therefore found it impossible to be heard in a public discourse dominated by a few large channels of communication.12 In his account, the major media were not offering substantive debates about significant political, social, and economic matters.13 Owners of mass media outlets were unwilling to present viewpoints that challenged the status quo—or, for that matter, supported it in explicitly ideological terms.14 This was in large part because it was structurally disadvantageous for them to do so: controversy would threaten profitability.15 Pandering to the part of people that enjoys mindless entertainment was easier and safer.
Accepting Barron’s analysis, then, the Internet could solve some, but not all, of the problems he identified. Aside from his condemnation of the concentration of sources, Barron’s critique of modern media, drawing on the work of Marshall McLuhan, had two related but analytically distinct components. First, visual media like television encourage style over substance, making them less valuable than media like newspapers for hashing out the issues of the day.16 Second, modern media are so expensive to produce that they can only survive by appealing to the lowest common denominator.17 As David Foster Wallace wrote,
television is [not] vulgar and dumb because the people who compose [the] [a]udience are vulgar and dumb. Television is the way it is simply because people tend to be extremely similar in their vulgar and prurient and dumb interests and wildly different in their refined and aesthetic and noble interests.18
The Internet and the “long tail”19 of media promise to alleviate the problem of the lowest common denominator, enabling smaller producers to survive by targeting niche markets even in television and film. Decreased costs of production and distribution in the digital age enable widespread access to a greater variety of content. Amazon.com offers more books, by at least an order of magnitude, than even enormous physical bookstores.20 Because listing a book costs Amazon very little compared to the costs to a physical bookstore of stocking a book that sells only one copy a year, and because Amazon sells nationwide, it can profit from books that ordinary bookstores can’t afford to carry.21 Those books naturally provide readers with access to more topics, from more viewpoints, than the relatively few popular works available in conventional bookstores.22 Netflix and iTunes offer other examples of increased diversity through new business models.23 Thus, more speakers can survive and thrive by finding the niche markets willing to pay for their speech, correcting the lowest common denominator problem one reader or viewer at a time.
Yet Barron identified another feature of modern media as also producing systematic distortions in discourse: the dominance of audiovisual media over text.24 If, as McLuhan famously said, the medium is the message,25 and if the message of film and television (not to mention video games and Internet video) is inherently antipolitical, then problems remain. An implication of Barron’s view of audiovisual media is that the availability of political documentaries on Netflix that could never survive at the multiplex will not be sufficient to restore a healthy democratic public sphere.
Perhaps Barron’s fears are better addressed by the explosion of blogging and other more text-based methods of communication used by millions of citizens on the Internet. Henry Jenkins, a media scholar at the Massachusetts Institute of Technology, has written extensively about the ways in which people—young people in particular—are using new media, and the connections they make through the Internet, to learn how to think and to write, as well as how to communicate in other ways.26 Text is part of that process, but it need not have pride of place. Jenkins, and scholars like him, argue that new media regularly support significant political discourse even in the narrowest sense of “politics.”27 At least after a new generation learns to use a medium’s particular features, that medium can provide complex and serious content as well as distracting entertainment.28
A related question is whether Barron is actually urging us to reclaim a culture that existed when information was transmitted through print or is instead imagining a utopia. Illiteracy, poverty, and—crucially—the denial of the franchise were significant historical limits on the ability of people to participate in the republic of letters.29 The dominance of print didn’t equate to a fully democratic society. Nor was print journalism immune from appeals to prejudice, short-circuiting rationality; “yellow journalism” got its name from newspaper circulation battles that slaveringly promoted war.30 Because other media are capable of communicating valuable information—images from New Orleans after Hurricane Katrina, or the beating of Rodney King—the beneficiaries and losers from new modes of communication cannot simply be sorted into the categories of the thoughtless and the thoughtful, respectively.
Even if Barron’s deliberative ideal is ahistorical and discounts the value of images, however, it has many attractive features. The goal of easy access to diverse viewpoints on important political and social issues is normatively desirable. This leads back to Barron’s more extended criticism of mass media: its concentration and focus on profits, resulting in lack of interest in controversial topics other than celebrity gossip.31
Diversity of content might at first seem to solve this problem, but concentration comes in many forms. The long tail only works efficiently if there are major content aggregators.32 iTunes, Amazon, Netflix, and others profit because they offer hits to attract numerous customers. Their customers’ second, third, and subsequent choices then increasingly diverge, creating the long tail.33 There are still blockbusters, who in Barron’s terms still have substantial control over the topics of public discourse. Bill O’Reilly’s books sell many more copies than an unknown’s political rantings; The Daily Show gets many more viewers than an average original political satire on YouTube.34 At the same time, those dominant channels can be evaded on occasion, and they may in some cases be prodded to address issues carried up through the capillaries of the Internet.35 I do not mean to suggest that nothing has changed, only that there remain substantial concentrations of power over public discourse.36
It’s not just that big hits remain profitable, or that they support the businesses that bring us the rest of the long tail. Concentration is more pervasive. The spaces in which people communicate—blogs, MySpace pages, message forums, and so on—are largely spaces they do not themselves own but are provided by ISPs whose policies may prioritize many things over users’ ability to speak.37 And in many categories, there are dominant providers with substantial market control. YouTube is a vehicle for new content to receive widespread attention, but other competing video sites lag far behind, and that means that YouTube’s choices about what videos to host—screening out pornography and combat footage and limiting the length of videos—determine what most people will see.38 Aesthetic complaints about YouTube are familiar from decadesold criticism of other popular media: by structuring itself around short and popular clips, the site creates stylistic expectations that make it harder for truly innovative works to thrive.39
If anything most clearly encapsulates the continued power of aggregation and selection, it is reality television—“amateur hour” in that the participants are not professional actors or scriptwriters but are nonetheless controlled by large media companies. The ideology is that the entertainment comes from amateurs, a special kind of user-generated content: “Television has been invaded by, and perhaps risks being overrun with, ordinary folk who have seeped through the screen much as Alice smushed through the looking glass. . . . The audience is the show, the show is the audience. . . .”40 But the producers—professionals and repeat players—select and carefully position the members of the audience who get a chance at the big stage.41 And, although they don’t like to talk about it, they edit footage together to create better narratives, script important moments, and give elaborate instructions so that the amateurs onstage will behave in the ways the producers want them to.42 One significant result is to make advertising in the form of product placement seem more natural and honest, when in fact it is the result of careful planning and mandates.43 The fantasy that amateurs are in charge is a useful one, but it remains a Hollywood illusion.44
Reality television offers in concentrated form what the new media environment does more generally—the appearance of unstructured choice and cacaphony, coupled with extensive background control by large organizations. That control often may not be overtly exercised, but its existence is important both in terms of large media corporations’ ability to focus attention and their power to suppress marginal speech. It is to the latter phenomenon—the suppression of a few voices in a speech environment that seems largely unfettered—that the next section turns.
B. Reading Lolita Online: The Case of LiveJournal
A recent series of events related to the popular web journaling site LiveJournal illustrates the complex interplay between law, social forces, and intermediary control of speech. Known as “Strikethrough” to many journal writers, the controversy began when LiveJournal suspended and deleted, without warning, a number of user accounts for noncompliance with its content policies.45 LiveJournal’s concern was with sexual content. MySpace was—and remains—much in the news for the presence on its site of pedophiles trolling for targets.46 While LiveJournal is a very different type of website, it still falls within the “social software” category, and its differences would not be significant to the reporters and regulators focused on pedophilia. Allegedly, an outside group (or perhaps a person posing as a group) threatened to contact advertisers about LiveJournal’s supposed support for pedophilia.47
Initially, however, LiveJournal did not suspend users it had determined to be pedophiles. It neither examined users’ writings nor cross-referenced identifying information with sex offender registries. Rather, according to the most common accounts, the targeted users had “interests”—phrases on their profile pages, designed to allow people to find like-minded journalers—dealing with illegal sexual conduct, such as rape and incest.48 LiveJournal had initially conceived of “interests” as reflecting users’ favorite things, but users had for a long time used the interests area of their profiles to identify topics of interest to them. So, for example, some abuse survivors listed incest as interests, as did a community dedicated to reading Nabokov’s Lolita in an online book club.49 They were caught up in LiveJournal’s purge. LiveJournal’s user base also includes a significant number of fan communities.50 This became important because a Harry Potter fan community that included fan stories with adult content— including depictions of rape and incest—was also purged.51
The resulting outcry was intensive and sustained, and even attracted notice from outside.52 Eventually, LiveJournal relented and reinstated many of the suspended users whose “interests” were as survivors or limited to fiction.53 A few months later, however, LiveJournal again permanently banned certain users, this time for posting sexually explicit drawings featuring characters determined by LiveJournal’s staff to be below the age of eighteen.54 Because the works at issue are drawn, rather than representing actual bodies, the difference between a sixteen-year-old and an eighteen-year-old is very much in the eye of the beholder. Artists thus argue they deserve leeway, while LiveJournal contends that an abundance of caution requires it to suppress material that appears to its staff to represent under-eighteen characters engaged in sexual activity.
This series of events, which is far from over, demonstrates some basic points about Internet intermediaries. First, LiveJournal’s initial miscalculation,55 driven by a moral panic that turned into a business imperative, was based on the erroneous assumption that users deployed the category of “interests” in the way that LiveJournal initially intended, when in fact they had adapted it to better fit their goals of self-expression and connection to others. One reason that intermediaries shouldn’t be liable for everything their users do is that users do unexpected things.
Second, users are highly vulnerable to intermediaries. Because the suspensions affected a user’s entire account, not just objectionable entries, the suspended users lost, in some cases, years of writing and art. There are journal backup services available, but not everyone uses them—very few people expected to be suddenly banned. Moreover, because intermediaries bring people together, there are often significant switching costs. Even if a more user-friendly environment is available, if moving there means losing connections to many friends, the gains may not be worth the costs. Or, from an external perspective:
The salience of Internet communication is famously sensitive to marginal changes in availability. . . . To assure the presence of countervailing sources of cultural power, major actors are crucial because they stand astride the attention of the central mass of the population.
. . . Even if [excluded viewpoints] are available to the segments willing to expend the time, effort, and expertise to search for them, the balance of popular perception may be skewed away from a proper evaluation of the matters before the public for decision.56
Third, users generally recognized that LiveJournal had every right to create and enforce its own policies, even if they went beyond what the law requires. The questions were, rather, ones of fairness and governance— exactly the issues Barron addressed in his analysis of private ownership of the means of communication. For example, a blogger argued:
Current controversies . . . have one stark issue in common: the conflict between corporate desire to profit from users and the content they generate, and the users’ own sense of ownership not only in their content and creativity, but in the hosted services they use to publish that content and to connect with others online. . . . .
. . . [T]here’s no such thing as “free speech” on Livejournal, because only a government with a constitutional mandate is required to provide its users with free speech. However, as civil liberties advocates have reminded us for years, the right to speech is only as good as the right to access to venues in which speech can be heard. And in an environment where public spaces are relatively rare, including the internet, there are strong arguments for corporate responsibility in voluntarily refraining from restrictions on user speech.57
As a result, the writer proposes measures to ensure user representation in corporate governance.58 Substantive disagreements about policy might be mediated through democratic procedures, perhaps both for establishing ground rules for acceptable content within user communities59 and also for providing due process for individual users whose speech is deemed unacceptable.60
These structural solutions share with Barron’s proposals a focus on institutional design.61 And they should also be possible by legislative mandate. Though such a law would affect the speech-related decisions of private companies, there is no inherent reason that private corporations must allow managers (or even shareholders) to make those decisions, given that their governance structures are creatures of state law.62
Legal requirements are never independent of social forces. With respect to adults seeking to engage in sexually explicit conversations or conduct with minors, legal pressures on social networking services have been applied at the state level. Attorneys general have pressured MySpace to disclose the identities, and even the private messages of, sex offenders who use its site.63 We don’t ban sex offenders from using the telephone,64 but we are apparently eager to ban them from communicating over the Internet.65 Access bans directed at potential child predators are already being implemented by sites like MySpace and LiveJournal, in part to ward off legal mandates.66
And this helps explain Strikethrough: the dynamics that prompted LiveJournal to act were based in part on the different, much less favorable treatment for intermediaries hosting or providing access to sexually explicit content as compared to copyright-infringing or defamatory content. Unsurprisingly, the legal regime tracks the social. The Communications Decency Act (“CDA”)67 protects ISPs from liability for most user-supplied content,68 but it specifically excepts criminal laws relating to obscenity and child pornography (as well as laws governing intellectual property).69 Congress determined that freedom of movement for intermediaries is too costly in this instance.
With anxiety rising about children’s sexual vulnerability, and without any constitutional requirement that intermediaries be immunized for content provided by their users, more businesses may choose to act preemptively, as LiveJournal did. By contrast, ISPs are less likely to police for defamation, copyright infringement, or other socially detrimental expression such as that found in pro-anorexia communities, whose accounts LiveJournal repeatedly declined to suspend despite protests from numerous users.70 The next Part addresses the ways in which liability regimes shape access in the absence of formal access rights.
II. Intermediary Control and Intermediary Liability: Power Without Responsibility?
Current law often allows Internet intermediaries to have their free speech and everyone else’s too. As just noted, § 230 of the CDA allows ISPs to set their own content standards and still avoid being treated like publishers.71 In fact, under the CDA, ISPs can apparently continue to host defamatory content that the original author wishes to have removed.72 Even common carriers face more potential liability than this.73 This Part explores the alternatives, and it argues that there is room for legislatures to maneuver in setting liability regimes that encourage—or discourage— various access policies.
Because the CDA protects ISPs from most tort claims, the standard example of a law governing Internet intermediaries’ liability is the Digital Millennium Copyright Act (“DMCA”).74 Under the DMCA, ISPs can avoid monetary liability for copyright infringement by disabling access or removing links to material when they receive a properly formulated notice that it is infringing a copyright owner’s rights.75 Avoiding liability is a powerful incentive to comply with this notice-and-takedown procedure.
The DMCA also provides mechanisms for users to counternotify—i.e., inform the ISP that content it has removed was mistakenly identified as infringing another’s copyright—and have material restored.76 Nonetheless, most users who receive notice do not counternotify, even when they might have valid defenses.77 The DMCA has, as intended, mobilized the power of intermediaries to control individual infringers—as well as a certain percentage of noninfringing uses. Because DMCA notice requirements are minimal and ISPs have no incentive to investigate, the notice-and-takedown process can be used to suppress critical speech as well as most users aren’t thinking about copyright or free speech when they choose providers, ISPs do not generally compete to protect user rights. As a result, ISPs serve as a chokepoint for copyright enforcement.
The government can target intermediaries in other ways in order to control speech, using that intermediary power to its own advantage.79 Requiring libraries and schools to use filters intended to screen out indecent content in order to get federal funding is a particularly blatant example of such targeting,80 with disproportionate effects on low-income users whose Internet access is more likely to depend on those institutions.81 As the DMCA exemplifies, however, regulating the circumstances under which an intermediary is liable for a user’s speech can be used to shape the overall marketplace of speech even without the use of the spending power.
The existence of the legal scheme set forth in the DMCA demonstrates that the CDA’s policy of conferring complete immunity on ISPs is not inevitable and, most significantly, not currently understood as a First Amendment requirement. Though free speech is certainly an element of the policy debates over intermediary liability, Congress, like the courts that have applied congressional policy, has generally assumed that the First Amendment put few limits on calibrating secondary liability for Internet intermediaries. We have several different intermediary liability rules, depending on the substantive body of law: immunity for most state-law torts;82 injunction-only safe harbors when an ISP follows the DMCA for copyright infringement;83 common law secondary liability for noncopyright intellectual property torts such as trademark infringement;84 and possible criminal accessory liability for obscenity and child pornography.85 As their diversity suggests, these regimes are all optional, depending on the policies lawmakers have sought to implement.86
A. The Constitutional Role of Intermediaries
There is an alternate constitutional story one could tell, starting (as modern First Amendment law does) with New York Times Co. v. Sullivan.87 Sullivan was a case about the Times as intermediary, displaying another entity’s supposedly defamatory ad after only minimal screening.88 What the actual malice standard protected was not the speech of the Times as such, but its business model—accepting the speech of others with only limited fact-checking.89 The Court was quite clear that it endorsed the paper’s business model as a means of implementing First Amendment values. Denying First Amendment protection to messages about the Civil Rights Movement because the paper was paid to run them, the Court held,
would discourage newspapers from carrying “editorial advertisements” of this type, and so might shut off an important outlet for the promulgation of information and ideas by persons who do not themselves have access to publishing facilities—who wish to exercise their freedom of speech even though they are not members of the press. The effect would be to shackle the First Amendment in its attempt to secure “the widest possible dissemination of information from diverse and antagonistic sources.”90
And, as the Court had recognized before, intermediary liability has snowball effects in limiting the speech available to the public, making it particularly problematic.91
Consistent with this concern, Sullivan’s rules limiting defamation are especially useful for intermediaries. A printer reproducing his own words can more easily assess whether he has taken reasonable care to verify truth; the real speech-chilling effects of a negligence standard come when he must guess whether someone else who wants to use his press has also taken reasonable care. Moreover, the printer-intermediary is likely to be less committed to getting a message out than a printer-speaker; more inclined to doubt the truth of another’s claims than of his own, and thus not overconfident about his chances of success in a lawsuit; and overall more risk-averse than individual speakers, not least because of the likelihood that the printer has deeper pockets and is a more attractive defendant from a plaintiff’s perspective. Sullivan, though of course protecting individuals as well, removes barriers that disproportionately discourage intermediaries from carrying others’ speech.92 A requirement that the plaintiff prove falsity, and some basic fault on the defendant’s part, obviously encourages the speech even of individual speakers. But the extra protections—clear and convincing evidence and actual malice—constitutionalized by Sullivan are especially useful for national newspapers and other intermediaries. Thus, Sullivan analyzed what the Times knew about the truth of the statements at issue, not what the individual author of the ad knew.
But Sullivan has not generally been understood as a case about intermediary liability.93 With the rise of the Internet and the appearance of multiple new business models, many of which relied on carrying unscreened-by-default content, it was unclear how far Sullivan’s rationale—protection for certain speech-based business models—would extend past its rule—no liability for defamation without actual malice. Specifically, before the CDA, ISPs appeared vulnerable to defamation suits, at least in instances in which they were given notice of defamatory content and subsequently refused to disable access to that content. The CDA was enacted on the theory that no ISP would accept the risk of standard Sullivan-type liability, given the massive amounts of user-generated content that the Internet allows.94 The Times can scrutinize its stories, letters to the editor, and ads and make reasonable judgments about its libel exposure; Google cannot review its entire index. Absent a constitutional right to operate a search engine free of liability for the indexed content—something not much argued95—Congress believed that it needed to alter the common law, even more than it had been modified by the First Amendment, to give Internet intermediaries the chance to make their business models work.96 In essence, the CDA, and even the DMCA, subsidize new intermediary models by protecting them from otherwise applicable law, but only as a matter of legislative grace.97
The flip side of this legislative grace is that the corporation’s powers and freedoms stem from laws designed to give it special advantages,98 but those need not include the ability to claim both speaker status as against the government and also immunity from treatment as a speaker as against private claimants. I am not arguing necessarily for greater intermediary liability for users’ behavior. The basic protection against intermediary strict liability, and even against any requirement to mediate disputes about appropriate content, is an important protection against unanticipated and practically uncontrollable liability for torts committed by individual users.99 Rather, I am arguing that if we limit intermediary responsibility, whether by § 230 or by the DMCA, we should also limit intermediary power to control speech. There is no reason that any speech rights that Internet intermediaries possess should be vested in intermediaries’ management, rather than attributed to users only when those users misbehave.
B. Balancing Power and Responsibility
Individual users’ speech can do harm, and absolute immunity for ISPs—even those that refuse to remove content after the original speaker concedes liability, or even those that deliberately induce the creation of content for the ISP’s own advertising purposes—may go too far. The CDA’s protection against third-party suits need not depend on an ISP’s unfettered ability to do anything it wants to its users. We could, for example, make certain ISPs into common carriers, or something near, banning content discrimination and also ensuring that they wouldn’t be liable for what users did with that service.100 Such a rule might help fulfill Barron’s ideal of access for even controversial and unpopular speech.101 But because it is easy to predict that problems of unlawful user-supplied content will persist, a neutrality policy could not stop at requiring access ex ante. It would have to specify how intermediaries should deal with illegal speech once it was made available. Whereas Barron focused on equal access, intermediary liability draws our attention to unequal outcomes.
As a practical matter, recruiting intermediaries to police objectionable content is simply too popular to make any total immunity-plus-nondiscrimination law politically viable. Section 230, whose general liability provisions have become so vital to ISPs, is in fact titled “Protection for private blocking and screening of offensive material,”102 and general immunity for user-supplied content was granted along with immunity from liability to users for “any action voluntarily taken in good faith to restrict access to or availability of material that [a] provider . . . considers to be obscene, lewd, lascivious, filthy, excessively violent, harassing, or otherwise objectionable, whether or not such material is constitutionally protected.”103
Section 230, that is, always attempted to further two objectives: protecting ISPs from liability and thus fostering free speech, and encouraging ISPs to monitor and suppress offensive speech.104 But the simultaneous support of freedom and suppression requires us to ignore the question of whose speech is supposed to be freed and whose suppressed.
Ironically—given § 230’s title—immunity alone has not generally been sufficient to convince ISPs to monitor content. Thus, though there seems to be no immediate prospect of general legislative action, various commentators have proposed cutting back on ISP immunity to encourage them to act against unlawful speech, especially when the speaker is anonymous or difficult to identify. The DMCA’s notice-and-takedown regime has seemed an obvious model for dealing with the situation of ISPs that transmit enormous amounts and types of speech and thus cannot be expected to detect unlawful speech without specific notice.105 Looking beyond the DMCA, Mark Lemley has recently explored various possible safe harbors for ISPs who provide access to substantial amounts of content, some of which is predictably going to violate some law.106 He argues that all such situations should be treated the same, and he endorses an intermediate standard that would limit the DMCA’s incentives to overblock while being less freewheeling than the CDA.107
To the extent that such proposals cover all sorts of illegal speech, they make us confront the question of exactly what free speech rights an ISP ought to be able to assert. That is, if the government may only constitutionally punish threats when the speaker intends to communicate a threat, could an ISP be held liable for failing to remove a threatening post on a blog after proper notice? If the ISP is seen as the speaker, then it probably does not have the requisite intent. Yet if the ISP is not truly engaging in any process of selecting speech, then there may be no reason to impose an intent requirement before a court could order the speech to be removed. A newspaper might report on a threat in the context of a story about the threat; in such a case, liability for the newspaper would be inappropriate. But ISPs don’t routinely put threats in context.
More generally, ISPs may be agents of free speech, but that does not mean that they automatically take on the interests of every speaker whose speech they carry.108 By default, access providers like America Online (“AOL”) and Google do not select or approve content and are not generally understood to do so. Just as a telephone company is not engaging in speech of its own when its users speak, ISPs regularly facilitate others’ speech rather than speaking for themselves. As conduits, ISPs’ concerns are different than those of initial speakers. Free speech doctrine could be tailored to protect their interests as transmitters. A notice-and-takedown procedure or, as Mark Lemley suggests,109 a scheme that protected all “innocent” ISPs would not impose the kind of affirmative monitoring costs that ISPs feared would drive them out of business. Another possibility worth exploring might be a modified notice-and-takedown with an arbitration component, in which a complainant would have to submit evidence supporting its assertion of illegality, rather than a bare claim (as suffices under DMCA).110
The definition of ISP in the CDA is quite broad, extending well beyond Internet access providers like Comcast or AOL, and even covering bloggers insofar as their blogs allow comments.111 Some bloggers might well have speech interests in hosting comments from others. But perhaps the definition of ISPs should be refined to take this into account, and those who claim a speech interest should be asked to take the bitter with the sweet: if they want to assert free speech claims on behalf of content provided by others, then the substantive standards for holding them liable for facilitating that speech should apply, rather than absolute immunity. Failure to comply with notice-and-takedown under the DMCA, for example, merely subjects an ISP to the underlying common law of secondary liability for copyright infringement, and the same would be true for defamation and other torts if the CDA were amended to be more like the DMCA. Given the underlying law of defamation, it might be difficult to hold a blogger responsible for a commenter’s defamatory statements even without absolute immunity.112
In the past, the Supreme Court has been willing to tinker with the procedure, rather than the substance, of speech torts in order to balance the costs of harmful speech with the benefits of speech that is useful but vulnerable to chilling effects. Most notably, the Court determined that negligent defamation of private figures could constitutionally justify an award of actual damages, but presumed or punitive damages in such cases would only be available when actual malice was shown.113 By reducing the size of the possible penalty, the Court believed that it decreased the chilling effect of a negligence rule to an acceptable level. Likewise, a regime that limited available remedies against ISPs to injunctive relief—whether conditioned on compliance with notice-and-takedown, as with the DMCA, or as a blanket rule for ISPs that lacked actual knowledge of illegality— would substantially decrease the chilling effect on ISPs of altering § 230.
If the DMCA model were extended, there would be a legitimate concern over chilling effects on individual recipients of takedown notices. Copyright owners have been aggressive enough using the DMCA; there is no reason to think that offended individuals acting on behalf of their own interests, rather than those of their copyrighted works, would be any more restrained.114 Of course, when an individual speaker is identifiable, it is already possible to threaten her with a defamation suit, but more people might use notice-and-takedown than would threaten to sue. Under the DMCA, the recipient of a notice can counternotify and have the right to have the ISP return the challenged material unless the sender files suit within a short period.115 In practice, very few people counternotify in DMCA cases, and it seems likely that this pattern would continue. But whether that would mean more benefits from the removal of ill-considered defamatory speech than costs in suppression of nondefamatory speech is more a matter of intuition than confident prediction.
Being threatened with a lawsuit definitely has a chilling effect, but not one that First Amendment doctrine has targeted. Possible proposals for reforming § 230 deal with intermediary behavior alone; the substantive standards for holding an individual speaker liable would remain stringent. To the extent that notice-and-takedown led to more implicit threats of lawsuits—that is, situations in which a recipient would perceive a likelihood of suit if she contested the notice by filing a counternotification, even if the notice sender had no real intention of taking the matter further— courts would have to confront the question of how to factor this into a First Amendment analysis. Formal legal doctrine is a rough tool for dealing with perceptions of the law, and to date First Amendment doctrine has only attempted to deal with the problem of chilling effects by making it harder to win, not harder to threaten. Doctrine might not be able to do much more than that, because it is almost always possible to threaten to sue, regardless of whether success is likely, and the threat is often a frightening one. Perhaps penalties for misuse of a notice-and-takedown procedure, as exist for the DMCA, could mitigate this risk.116 Moreover, changes in ISP immunity should, as noted above, be accompanied by increasing the governance rights of users over the services they use. Without such changes, reforms in § 230 are likely to increase the number of sudden, surprising deletions of speech like those experienced by LiveJournal users. Thus, § 230 reforms must not focus only on ISP immunity from third-party claims, but must also address their immunity from user complaints about censorship.
I want to be clear: there is no constitutional requirement that Internet intermediaries be regulated consistently. LiveJournal can have its cake and eat it too, even though I think it’s a substantively worse solution than an alternative that tied immunity for users’ speech to some type of procedural due process, democratic self-governance, or nondiscrimination rule. The flexibility the legislature has with respect to intermediary speech, and the resulting effects on individuals’ speech, highlight the absence of a neutral background rule defining speakers’ rights and duties.117 Individualist theories of free speech cannot answer the pressing questions posed by intermediaries—and intermediaries are everywhere. “As a constitutional theory for the communication of ideas, laissez faire is manifestly irrelevant.”118 Regardless of what we choose to do about it, we cannot pretend that our brave new online world has rewritten the rules of access. Therefore, we should be thinking carefully about the best regimes that will balance promoting speech with reducing the harm of unlawful speech. This menu of choices should include, at a minimum, alternatives for empowering users of major ISPs substantively and procedurally, as well as alterations in § 230 to better calibrate power and responsibility.
Conclusion
The proliferation of content from new sources challenges the mass media, but new speakers remain dependent on larger organizations. While Barron wrote approvingly of the useful and valuable unheard perspectives lurking in the audience, waiting only for the opportunity to speak, current regulations concern themselves more with dangerous volunteers from the audience, creating a legal structure that protects intermediaries from thirdparty claims but provides them scant legal incentive to promote diversity of speech. Meanwhile, the same economic incentives Barron identified for mass media like newspapers and television stations push ISPs towards promoting advertiser-friendly content. LiveJournal was willing to sacrifice individual users for a better image for advertisers and investors, and this pattern is likely to continue as aggregators attempt to monetize popular spaces such as Facebook and YouTube.
Online as well as offline, government determines who gets to structure speech, and thus who gets to speak. The CDA empowers LiveJournal to monitor content but allows it to ignore complaints of defamation. The DMCA sets out a separate rule for copyright infringement allowing copyright owners to send notices of claimed infringement and have accused material taken down. Another law requires libraries and schools to filter if they want federal money, which many must have to survive. These allocations of power are not required by the First Amendment, nor are they barred by it. Indeed, intermediaries’ power to disseminate ideas and material is so highly structured by discretionary legal rules that nondiscretionary legal rules such as constitutional requirements cannot provide substantial guidance in dealing with intermediaries’ power.
To say that an ISP’s servers are its property, and thus so too the content stored on those servers, would ordinarily be to imply some responsibility when the content turned out to be unlawful. But the CDA and to a fair extent the DMCA uncouple those things. That is not a neutral policy (as if there ever was one). My version of Barron’s argument, then, is a policy-based call for action, backed by a theory that the legislature can legitimately determine that free speech will be served by particular restraints on intermediaries’ ability and incentives to interpose themselves between speakers and audiences.
I would be happier if the Constitution required my preferred allocation of speech rights. But in the absence of such a mandate, there are still important questions about the best way to fulfill our needs to talk and to listen.