By A. Douglas Melamed1
I thought antitrust law was working pretty well when I was in the Justice Department during the Clinton Administration. Both the Department and the Federal Trade Commission won significant merger and nonmerger cases, including of course the monopolization case against Microsoft, that established important precedents and moved the law in a more aggressive direction.
In the years since then, I have come to believe that antitrust law has become too conservative and needs to be recalibrated to be more aggressive. It needs to do a better job taking account of modern economic learning and to be concerned about false negatives, as well as false positives. I have written a number of articles explaining that view,2 and I have testified to that effect before Congress.3 I have also suggested that special competition rules might be appropriate for the largest tech platforms.4
Nevertheless, I believe that Congress should not enact the American Innovation and Choice Online Act (AICOA) in its present form. I summarize below what I see as the three most problematic aspects of the bill.
I. What Does “Harm to Competition” Mean?
There has been a consensus for several decades that antitrust law is about preventing harm to competition in order to enhance economic welfare. Preventing harm to competition does not mean protecting individual firms from the rough and tumble of marketplace competition as competing firms try to best their rivals by providing better products and services. It means prohibiting anticompetitive conduct that creates or maintains market power. By preventing increased market power as a result of anticompetitive conduct, antitrust law increases output, which benefits workers and results in lower prices that benefit consumers, especially those least able to pay higher prices.
AICOA pays at least lip service to the importance of harm to competition. Sections 3(a)(1)-(3) of the bill prohibit self-preferencing and discrimination by the platforms, which can be harmful in some circumstances. The prohibitions imposed by those Sections apply only where plaintiff shows that the conduct “would materially harm competition.” The covered platforms can avoid violating the other prohibitions of the bill (Sections 3(a)(4)-(10)) if they can show that the conduct at issue “has not resulted in and would not result in material harm to competition” (Section 3(b)(2)).
That language might mean that there can be no violation of the bill unless there is the kind of harm to competition with which the antitrust laws have long been concerned, but that is far from clear. In the first place, the statutory language is patently ambiguous because “would” is conditional and the bill says nothing about the contemplated conditions. More important, while the language suggests an antitrust-type concern about competition, the bill does not include the normal antitrust language (e.g., “competition in the market as a whole,” “market power”) that gives meaning to the idea of harm to competition, nor does it say that the imprecise language it does use is to be construed as that language is construed by the antitrust laws.
Plaintiffs will no doubt argue, and courts might agree, that Congress did not intend to incorporate existing antitrust concepts and that “harm to competition” means any reduction in competition by, for example, causing a weak and insignificant rival to exit from an intensely competitive market. Plaintiffs will argue that Congress could have specified that the bill incorporates the antitrust notion of injury to competition but that Congress chose not to do that, and they will emphasize that the whole point of the bill was to supplement the antitrust laws with stronger prohibitions.
The bill could be very harmful if it is construed to require, not increased market power, but simply harm to rivals. The U.S. has in the past tried laws that insulate weak firms from competition provided by more efficient firms. The results have been increased costs, reduced output, and harm to consumers and suppliers.
II. The Bill Would Prohibit Welfare-Enhancing Conduct, Even if the “Harm to Competition” Language Were Construed to Mean Increased Market Power
The antitrust laws do not punish firms that succeed by welfare-enhancing competition on the merits, even if they gain market power by doing so. Those laws prohibit only anticompetitive conduct. AICOA, by contrast, would prohibit conduct that is not anticompetitive. The bill could, therefore, reduce economic welfare, even if the “harm to competition” language is construed to mean increased market power.
For example, Section 3(a)(6) prohibits the covered platforms from using “nonpublic data that are obtained from or generated on the covered platform by the activities of a business . . . to offer, or support the offering of, the products or services of the covered platform operator that compete or would compete with products or services offered by business users on the covered platform.” In other words, the bill would prohibit Amazon from using Company A’s data to build better products or services that compete with Company B; and it would bar Google from using data about Company C to improve the targeting and value of ads sold in competition with Company D. Notably, it would prohibit the use of such data even if Companies A and C in the examples above agreed to sell the data to the platform, without being coerced by threats or otherwise, simply because they expected to benefit from the sale. In these examples, everybody is harmed except maybe Companies B and D.
The bill does provide some affirmative defenses to otherwise prohibited conduct, but those defenses are narrow and do not encompass all welfare-enhancing benefits. For example, Section 3(a)(8) restricts uninstalling software and changing default settings. The conduct prohibited by that section can be justified only if shown to be necessary for the “functioning of the platform” or to prevent transfer of data to China or a “foreign adversary;” that conduct evidently cannot be justified even if it is necessary to develop new, complementary products. The other conduct prohibited by the bill is justifiable if it is necessary to “maintain or substantially enhance the core functionality of the platform” (Section 3(b)(1)(C)), but not if it substantially enhances other features of the platform or other products or services. And does “core functionality” mean core functionality today, so that the platform cannot evolve by adding or changing core functionality? The narrowness of the justifications ensures that the bill, if enacted, will deter and likely prohibit a wide range of efficient and welfare-enhancing conduct.
Sections (3)(a)(1)-(3) could undermine economic welfare for an additional reason. Those sections would prohibit the covered platforms from preferencing their own products, but the broad language of the bill glosses over important ambiguities. Do the prohibitions on self-preferencing mean that the platforms cannot enable their products and services to take advantage of their joint scale and scope economies? Do they mean that the platforms cannot preinstall their new products and services, even if users want them preinstalled? Do they mean that the platforms have to preinstall all alternatives in addition to their own preinstalled products and services, even if doing so makes the platform more cumbersome or the products and services less accessible and less valuable to users?
Economists have long understood that innovation is far more important for economic welfare than static efficiency.5 Yet AIOCA is likely to impair innovation by the platforms. As noted above, the bill would not permit conduct to be justified even if it is necessary to enable development of valuable new products and services, and it might inhibit or prevent the evolution and improvement of the platforms themselves. In addition, Section 3(a)(4) prohibits the covered platforms from, among other things, unreasonably delaying third party access to platform features that are available to the platform’s own products, except where necessary to prevent cybersecurity risk. Does that mean that a platform cannot introduce its own complementary products or features (e.g., voice recognition, mapping and GPS, camera, etc.) until it has enabled all other providers of similar or competing products or features to access the platform? One might think that others who are later in line for the technical reviews and assistance necessary for that access can reasonably be expected to obtain access later; but does Section 3(a)(4) mean that the platform cannot disclose to one of its own complementary product units technical specifications relevant to that access until it is ready to disclose them to everyone, and that it cannot even disclose competitively sensitive, tentative specifications to its own unit in order to draw upon the latter’s expertise in optimizing the platform’s suitability for the complementary product or service?
AIOCA will prohibit welfare-enhancing conduct by the covered platforms. It will also prohibit anticompetitive conduct by the platforms, although the benefits of that are unclear because at least some, maybe most, of the anticompetitive conduct that the bill would prohibit violates existing antitrust laws. Supporters of the bill say that it will promote innovation by making it easier for smaller firms that use the platforms to flourish. That is an interesting conjecture, and there is undoubtedly some truth to it. It is possible that efficient, welfare-enhancing conduct by the platforms regarding their businesses in complementary markets is so harmful to other actual and potential firms in those markets that it is harmful to welfare overall. But there have been no hearings on the bill, and Congress has made no other investigation or findings that would support a conclusion that the costs of restricting efficient, innovative conduct by the platforms will be less than the benefits from insulating rivals in complementary markets from the conduct that the bill would prohibit.6
III. Content Moderation
According to newspaper reports, Senator Cruz, Congressman Buck, and other Republicans are saying that the AICOA will be a useful tool against content moderation or, as they think of it, liberal bias on the digital platforms. Senator Klobuchar and other Democrats have said that the bill is about commercial practices, not content moderation. In support of their position, the Democrats point in particular to language in the bill that limits the protected Business User category to persons that use the platforms for “advertising, sale, or provision of products or services” (Section 2(a)(2)(A)).
There are three problems with the Democrats’ position on this issue. First, if Republican supporters of the bill say that it is intended to help reign in content moderation, that will be part of the legislative history and might affect judicial interpretations of the bill. Second, two of the prohibitions set forth in the bill – Section 3(b)(5), which prohibits conditioning access to the platform on use of other products or services, and Section 3(b)(8), which prohibits restrictions on uninstalling preinstalled software or changing default settings — are not limited by the definition of Business User. With AI advances, one can imagine platforms conditioning access to the platform on use of, or restricting uninstalling or changing the default settings on, truth filters. The bill would block that.
Third, and most important, content moderation might be thought appropriate for certain types of Business Users. Consider, for example, a cake shop that advertises on the platform that it does not serve LGBTQ persons and includes in the advertising false statements about them. Or a PAC that advertises the sale of MAGA hats, the profits from which will be used to fund efforts to overturn the stolen 2020 presidential election. Or a third-party app or service that disseminates false information. The platform will defend its refusal to run those ads or enable those apps or services on the ground that it is using neutral criteria based on truth and is not discriminating, but how should counsel advise the platform if she expects that the plaintiff will be able to find hundreds of false ads, or false statements disseminated by other apps or services, that slipped through the truth filters over many months in support of an argument that the criteria are not being neutrally applied? The plaintiff will have a tough time winning such cases, but given the number of potential enforcers of the new law (including states like Florida and Texas), the litigation risk might nevertheless have a real deterrent effect on content moderation.
The misinformation and bias issues on social media and other platforms are very important. They should not be confounded by a he-said, she-said debate about the reach of legislation intended, at least by the Democrats, for other purposes.
There are many hard issues raised by current antitrust controversies and the digital platforms. New legislation might be appropriate. Sound legislation requires a clear, factually-based understanding of the problems the legislation is seeking to address and of the costs and benefits of the solutions. AICOA is not yet such a bill.
Click here for a PDF version of the article
1 Professor of the Practice of Law, Stanford School.
2 See, e.g., Competition Law as Common Law: American Express and the Evolution of Antitrust, 168 U. Penn. L. Rev 2061 (2020) (with Michael L. Katz); Antitrust Law and Its Critics, 83 Antitrust L. J. 269 (2020); see also Mergers Involving Nascent Competition (March 2, 2022), http://ssrn.com/abstract=4009229; Innovation under Section 2 of the Sherman Act, 84 Antitrust L. J. 1 (2021) (with Richard J. Gilbert).
3 Prepared Statement of A. Douglas Melamed before the United States Senate Committee on the Judiciary, Subcommittee on Antitrust, Competition Policy, and Consumer Rights, Hearing on “Does America Have a Monopoly Problem? Examining Concentration and Competition in the U.S. Economy” (March 5, 2019), Melamed Testimony.pdf (senate.gov).
4 See, e.g., Antitrust Law and Its Critics, supra, note 2.
5 See, e.g., Innovation under Section 2 of the Sherman Act, supra note 2, at 2 n.6.
6 The so-called House Report (the November 2020 report of the majority staff of the Antitrust Subcommittee of the House Judiciary Committee) did not even attempt to make such an assessment and is, for than and other reasons, far from sufficient for that purpose. See, e.g., https://www.competitionpolicyinternational.com/a-missed-opportunity/.