Social media platforms are not infrastructure nor natural monopolies, and they should be regulated only if they have monopolistic power and abuse it. Free speech is a fundamental right, but its protection needs to be balanced with other rights.


In the aftermath of the Capitol Hill riots, then US President Donald Trump’s account has been suspended from the majority of internet platforms, including Twitter and Facebook, to prevent any incitement to violence. As many of Trump’s supporters migrated to Parler, a micro-blogging and social network where far-right, anti-semitic, and conspiracy-based messages were posted every day, Amazon Web Services (AWS) removed Parler from its servers; Google and Apple removed it from their app stores.

In a ProMarket article, prof. Luigi Zingales claims these decisions set a “dangerous precedent.” The TAGAF companies (Twitter, Amazon, Google, Apple, Facebook), Zingales argues, “are not random private companies, they represent (as the telephone in the past) a basic infrastructure of communication.

If these platforms were non-replicable infrastructures, a strong case could be made for regulatory intervention, for the same reason telecommunication and energy networks are regulated (as the telephone metaphor suggests). In the Trump case, most commentators and prominent politicians have called for an intervention based on political arguments such as the protection of free speech and the power that social media platforms wield. Still, the TAGAF—particularly Twitter and Facebook—are private companies that compete with each other, as well as with many other sources of information. Is the infrastructure analogy relevant in this case? The Oxford Dictionary defines infrastructure as: “The basic physical and organizational structures and facilities (e.g. buildings, roads, power supplies) needed for the operation of a society or enterprise.” From an economic point of view, an infrastructure should be regulated insofar as it is a natural monopoly. Each of the TAGAF may be a de facto monopolistic provider in its own markets (as Tommaso Valletti argues, among others), but there is nothing natural in this.

Unplug Facebook or Twitter, and the web works equally fine, which was hardly the case with AT&T. Amazon, Apple, and Google are neither “platforms” nor “infrastructures” but multiproduct firms supplying, among other things, important bits and pieces of the internet infrastructure which could be substituted by other competitors without affecting the proper functioning of the internet as an infrastructure. If regulatory action has to be taken, it should be grounded on a specific economic analysis of the market segments in which they exercise the kind of market power we aim at removing. A case can be made for tough antitrust action against each of them, discussed on a case-by-case basis if market power can be shown and a reasonable theory of harm can be devised. Regulating online platforms is something different, and, more importantly, it presupposes a one-size-fits-all kind of approach.

“Unplug Facebook or Twitter, and the web works equally fine, which was hardly the case with AT&T.”

Amazon or Google tell a partly different story: While they provide services that are part of the internet as a basic infrastructure, it takes a long way to get from this basic observation to considering them natural monopolies or “infrastructures” per se.

Going back to the Trump ban, the true focus of the debate has been, in any case, Facebook and Twitter: private companies selling different services to their users who agree to specific terms of engagement. It may be true that they are exercising market power, so antitrust authorities might (and possibly should) impose remedies. They may even go so far as to impose a breakup of Facebook (Twitter has far less users and raises less serious questions). Any further acquisitions should be halted, which would reinforce their market power. Still, the choice does exist, and users exercise it, as the massive migration towards Parler showed together with the even more massive increase of Signal and Telegram users after WhatsApp, a Facebook subsidiary, updated its terms of service. Facebook and Twitter occupy, indeed, a dominant position, but even they should be analyzed separately.

In October 2020, Facebook had two billion users while Twitter had 353 million. Their market share in terms of all social media visits was 60.52 percent and 14.43 percent, respectively. Facebook is the market leader and may have committed abuses over the years, as has been variously documented. But what about Twitter? Where is the evidence of abuse? Zingales argues that the foundation for Facebook and Twitter’s decision to ban Trump lay on a discretionary choice rather than on the rigorous application of their terms of service. He believes that their failure to intervene earlier, or to suspend many other users that were as violent as Trump (or more), provides hard evidence. One might argue that Twitter and Facebook did too little too late—still, how many assaults on Capitol Hill had they ignored before? That in someone else’s view they “should” have acted earlier (shutting down pages more frequently?) bears no relevance as to whether they have been enforcing their own terms of service and can hardly provide a case for a wide-ranging regulation.

“This is what ‘free speech’ is about: anyone has a right not to be censored by the government. That does not mean that anyone has a right to say anything, anywhere: free speech is a right, but not the only one.”

If a company abuses its dominant position, there are space and procedures for imposing sanctions or remedies; otherwise, every company should be free to set its own rules. Individuals are free to use the services by agreeing to the terms or otherwise switching to an alternative platform. This is what “free speech” is about: anyone has a right not to be censored by the government. That does not mean that anyone has a right to say anything, anywhere: free speech is a right, but not the only one. Nor is it superior to other equally-fundamental rights, among them the right of private companies to set their own rules—as long as they do not infringe on the law—or to prevent a mob from assaulting Congress to undermine the rule of law and the Constitutional order. Striking a balance between these rights is what politics is about. Again, this brings to mind the phone network analogy that Zingales utilizes: a telecommunication network is (or, instead, was) a natural monopoly. An online platform like Facebook or Twitter may be a monopoly (this might be the case of Facebook, much less of Twitter), but it is not a natural one.

Different conclusions could be drawn if Facebook, Twitter, and the like were treated as publishers selling information for which they are liable. Section 230 of the Communications Decency Act allows social media platforms to deny having publishers’ nature because it is not possible (nor is it desirable) for these companies to check ex-ante all the content published by their users. This legislation helped the internet flourish, and it should be maintained. If the state of technology now allows social media to check the content and intervene in real-time as inappropriate or illegal content is uploaded, perhaps the rules might be improved (as they are in Europe, to some extent). Incidentally, even publishers have a right to allow or remove any content from their platforms without anyone crying about free speech being infringed. If, however, a publisher grows too big or abuses its dominant position, antitrust authorities have a say.

Still, companies selling different services on the internet could be treated differently according to their business model. Their conduct should be scrutinized and sanctioned if they have relevant market share and abuse their dominant position. If they contribute to the functioning of the internet as an infrastructure, their role should be clarified, properly considered, and regulated if deemed relevant. A case-by-case approach based on sound economic motives is the most appropriate one and we should avoid bringing on political arguments as a mean to impact the regular functioning of this relevant industry.

Regulation is often invoked to change the specific situation and obtain particular results that one political party or the mainstream public opinion appreciates. In this case, antitrust enforcement—rather than ex-ante regulation—should always be used on a case-by-case basis. Only by minimizing regulation and discretionary public intervention shall the market flourish, innovation shall thrive, and consumer and citizens welfare and utility shall increase.

Disclosure: Carlo Stagnaro is the Director of the Digital Economy Observatory at Istituto Bruno Leoni, which has previously received minor funding from Google.