Don’t trust antitrust to tame Big Tech

In the past years, Big Tech has come under increased scrutiny from antitrust authorities in the EU and in the US. Whether through private or public litigation, cases have been brought before courts on both sides of the Atlantic, and colossal sanctions have been pronounced on monopolization charges. In a context where inequalities continue to rise and where the power of large corporations is being questioned, a form of “antitrust populism” has crept into political discourse. But does Big Tech really fear antitrust action? Is antitrust enforcement necessary to reduce concentration and abusive practices in digital markets? Is it sufficient?

The tech industry is prone to concentration

Messaging applications, social media, and other networked services strongly benefit from network externalities: utility for a new user increases with the number of current users. Take Facebook for example: the utility I derive from a social network clearly depends on the number of other users that are already using the platform. Google’s case is more complex: it benefits from network externalities if its search accuracy increases with each additional user. Alternatively, it benefits from indirect network externalities, in that advertisers, forming one side of the platform, derive greater utility from a larger number of users on the other side, and vice versa. This latter phenomenon is a key element of Amazon’s success, with each side of the platform (consumers and merchants) benefiting from the size of the other side.

What are the consequence of these network effects? When they are strong, they raise switching costs for users, which results in strong market concentration. In the case of Whatsapp, this is clear: market entry is hindered by the fact that users already have access to their network of contacts. In this context, switching to a new service usually means they have to convince all of their friends to come along with them, or at least to multi-home. As such, users are “locked in” to the network. In the case of search engines, apparent switching costs are low (“competition is just one click away”), but the lock-in is simply more insidious. Google’s access to a tremendous wealth of historical search data and the fact that most websites optimize their content to adapt to its algorithm (because it is the leader) means that entry in the market is extremely difficult.

Against this, tech leaders argue that digital markets are in constant flux, submitted to successive disruptions which make the previous technology obsolete at a moment’s notice. Certainly, this seems to be borne out by examples of social media such as AOL or MySpace. But this perspective fails to consider that with pocketable smartphones, access to personal computing may have reached its “final form”. Ben Thompson, for instance, argues that specialization will take place (e.g. connected watches), but that mainstream general purpose computing is likely to have reached a sort of equilibrium with the smartphone. This can only make concerns about services such as Android (Google) and Whatsapp more acute.

Under-enforcement can harm consumers

“So what?”, you might ask. Concentration isn’t necessarily bad, and everyone agrees that dominance, when based on selling a superior product, should not be punished. Crucially, over-enforcement has the potential to chill innovation in the long-term, which would be detrimental to consumers. After all, who is Google harming by marketing such a powerful product for free?. This reasoning is commonplace, and the associated “wait-and-see” attitude dominated US antitrust enforcement for decades. The approach has ideological roots in the Chicago School, which holds that false positives are always more costly than false negatives in antitrust enforcement. According to this doctrine, the standard of antitrust law is consumer welfare.

What is the effect of this standard? It warrants a form of laissez-faire whenever antitrust authorities are not 100% sure consumers are being harmed more than they are benefiting from a dominant player’s conduct. In fact, this was the reasoning of the Federal Trade Commission in 2013, for example, when it decided to stop investigating Google on allegations of “search bias” in the Google Shopping case: “condemning legitimate product improvements risks harming consumers.”

But the tide is turning. There is widespread awareness that dominance of digital platforms, combined with lack of transparency, means that Big Tech could be harming consumers without even raising suspicion. A key element of this issue is privacy. Accepting that the “price” that consumers pay to access most digital platforms is giving up part of their privacy, it’s important to ask what competition large platforms face. In traditional industries, market power is defined as the ability to profitably maintain prices above marginal cost. So the key question is: what level of personal data collection would make Facebook users turn away from the platform? If the answer is “none” (within the limits of data protection legislation), then Facebook enjoys market power, and its “pricing” of privacy can be described as abusive.

Antitrust, populism and the myth of omnipotence

As a result of this growing awareness, expectations towards antitrust authorities have grown strongly in the past decades. Once an arcane and frankly dry field of law reserved to experts, antitrust has become a key element of public debate. It is frequently mobilized in electoral campaigns, with the promise of “breaking up Big Tech”. In Europe, it even has its very own “pop star” in the person of Commissioner Margrethe Vestager. Whether you side with Google or with the European Commission, the sudden re-emergence of antitrust into mainstream public debate is probably a good thing: heightened awareness of competition imbalances can make users more likely to hold large platforms accountable, and firms to pursue litigation when they fall prey to predatory business practices.

The flipside is that, as part of often populist discourses, antitrust is being asked to solve problems that it is not designed to tackle. Antitrust actors are called upon to address concerns about the political power of large corporations and income equality. But these clearly fall outside of its ambit. The illusion that antitrust will be able to solve these long-standing issues is likely to be counter-productive. Antitrust cannot and should not replace tax reform and the fight against corruption. Underlying idealized conceptions of antitrust is a populist vision which pits tech elites against “the people”. This vision is widespread today in the context of rising inequalities and increasing concerns about tech in other fields, such as hate speech and tax avoidance.

But the reality of antitrust enforcement is much less exciting than what populists believe. The landmark US Microsoft case, for example, was drawn out over nearly 10 years, and the result is mixed. While it prevented Microsoft from further engaging in the conduct that was identified as illegal, it arguably did nothing significant to restore competition. Despite a salutary effort towards interoperability, the case did not significantly lower barriers of entry or force Microsoft to open up across the board. Crucially, most requirements were self-enforcing. The EU Google Shopping case might suffer the same fate — the appeal lodged by Google is still being processed by the CJEU— given how backward-looking the remedies are. Despite an impressive sanction, money damages have proven to be largely ineffective in restoring competition. Moreover, the context has changed tremendously more than a decade after the initial complaint was filed, giving the case a bitter taste of outdatedness.

At the heart of the issue is the fact that monopolization and abuse of dominant position litigation is by definition ex post and that enforcement in big cases traditionally takes years. Add to this the fact that ex ante merger control has been weak in the past years (Facebook-Whatsapp comes to mind) and that authorities generally have limited resources to investigate and prosecute — especially compared to the money that is spent on antitrust compliance — and you arrive at a situation where large platforms no longer fear antitrust enforcement. Instead, they continue to rely on barriers to entry and buying potential competitors off to consolidate their dominance, and consumer welfare is harmed by lack of choice and overcollection of personal data.

So what can regulators do? Sure, they should strengthen antitrust enforcement, including merger control. Thinking about new ways to tackle monopolization in digital markets is also important. But antitrust has limited powers and resources, and it should not take on problems for which it is ill-equipped. We need an alternative means of increasing competition in tech. Acting on Lawrence Lessig’s contention that “code is law”, modifying the architecture of networks can set the terms on which they are used. We need a structural remedy to concentration in the tech world. A remedy that wouldn’t supplant antitrust enforcement, but one that could avert market power ex ante. This remedy is interoperability.

Making networks open and decentralized by default

Interoperability is the ability of systems to work together. It is closely related to the notion of compatibility. Different types of interoperability exist. Interoperability can be adversarial: this means that individuals or firms deliberately try to make their services interoperable with the dominant service, most often against the the dominant player’s will. It can also be cooperative, either in the form of industry-wide standard-setting (this is also known as full interoperability) or bilateral: firm A decides to make its product compatible with firm B. Finally, in digital markets interoperability can take the form of client interoperability: users could use several social media platforms from a single application, with the application acting as a middleman between the user and the platforms. Generally, interoperability in digital markets is based on providing third-party firms with access to data and features via application programming interfaces (APIs). Finally, interoperability is closely related to the notions of data portability (the ability to take your data with you when changing service providers) and delegability (the ability to delegate a task to that third party or to access a service via a third party).

Requiring large platforms to make their services interoperable can lower entry barriers by reducing network effects. If messenger applications can interoperate, I don’t have to worry about my which service my contacts use, and I can use the service of my choice, e.g. a service that handles my data in a way that I find acceptable. This mechanically reduces concentration and increases competition on the merits. Skeptics might point out that “market forces” should decide whether services should be made compatible or interoperable (read: if companies don’t decide to make their own services compatible, then enforcing compatibility would amount to government failure). This point of view neglects the fact that private incentives for compatibility in network industries are often too low by social welfare standards. Indeed, compatibility generally increases output, but larger (and in particular: dominant) players have low incentives to engage in standard-setting or building bilateral compatibility. This is because they are unwilling to relinquish the competitive advantage they derive from their larger user base. This explains why large platforms have been relatively proactive in pushing for data portability (especially Google), but rather timid when it comes to moving from portability to interoperability.

All of this isn’t to say that interoperability doesn’t carry risks for innovation, privacy and security. Crucially, standardization is not necessarily the best way to go forward, depending on the industry. Full interoperability supposes that companies agree on a set of interoperable features. This is clearly limiting in terms of innovation in certain industries. While messaging applications may be able to agree on a set of interoperable features, this may not be the best solution for social media, which are far more diverse and fast-evolving. Client interoperability may also be a good way to promote diversity and innovation, by enabling users to access services through the application of their choice (and with the features of their choice). Interoperability can also be risky in terms of security. The fact that email communication is fully standardized, for instance, means that universal email encryption is particularly difficult to implement, because all the email platforms would have to get around the table and agree on an encryption protocol, which commentators claim is highly unlikely. Finally, privacy concerns should also be taken into account. After all, Cambridge Analytica accessed Facebook’s data via its API, which is exactly what delegation to third parties involves. This is why policymakers need to make sure delegation is in the interest of the user and based on explicit user consent. Generally, policymakers might want to consider:

  • Extending data portability requirements
  • Setting minimum interoperability norms for large platforms
  • Banning measures that aim to restrict interoperability
  • Enabling delegation to third parties based on user consent

These issues are already being tackled in various legislatures. The GDPR introduced data portability requirements, for instance. More recently, the US ACCESS Act, which is still in the first stages of the legislative process, is putting the spotlight on these topics, and several actors have requested interoperability provisions to be included in the forthcoming EU Digital Services Act. The task is a formidable one. But its success or failure will shape the Internet for years to come.

Disclaimer: This essay was written as an assignment for the class “Regulation and the digital economy” taught by Benoît Loutrel and Betrand Pailhès at Sciences Po.

References

Cyphers, B. and C. Doctorow, ‘A Legislative Path to an Interoperable Internet’, Electronic Frontier Foundation, July 28, 2020, https://www.eff.org/deeplinks/2020/07/legislative-path-interoperable-internet.

Katz, M. and Shapiro, C., (1985) ‘Network Externalities, Competition, and Compatibility’, American Economic Review, vol. 75, pp. 424–440.

O’Shaughnessy, P., ‘Ben Thompson — Platforms, Ecosystems, and Aggregators’, Invest Like the Best, episode 176, June 2, 2020 , https://investorfieldguide.com/thompson/.

Riley, C., (2020) ‘Unpacking interoperability in competition”, Journal of Cyber Policy, vol. 5, no. 1, pp. 94–106.

Shapiro, C. (2018) ‘Antitrust in a time of populism’, International Journal of Industrial Organization, vol. 61, pp. 714–748.

Shapiro, C. (2009) ‘Microsoft: A Remedial Failure’, Antitrust Law Journal, vol. 75, no. 3, pp. 739–772.