Friso Bostoen & Nicolas Petit: “Platforms’ Treacherous Turn”

The Network Law Review is pleased to present you with a Dynamic Competition Initiative (“DCI”) symposium. Co-sponsored by UC Berkeley, EUI, and Vrije Universiteit Amsterdam’s ALTI, the DCI seeks to develop and advance innovation-based dynamic competition theories, tools, and policy processes adapted to the nature and pace of innovation in the 21st century. The symposium features guest speakers and panelists from DCI’s first annual conference held in April 2023. This contribution is co-authored by Friso Bostoen, Assistant Professor at Tilburg Law School, and Nicolas Petit, Professor of Law at the European University Institute.


The ‘open early, closed late’ strategy is taking a heavy toll on consumers in the digital economy. The pure case of ‘open early, closed late’ occurs when a digital platform imposes prices after a period of free and unlimited supply. More generally, ‘open early, closed late’ encompasses broader forms of consumer exploitation, like when Netflix, Uber, or Twitter raise prices or reduce quality. Today, ‘open early, closed late’ strategies may well be one of the most visible forms of consumer harm. And yet, the antitrust laws have so far remained toothless. This is not a small paradox. Both the US and EU antitrust systems consider that the consumer welfare standard should be the ultimate goal guiding the application of the law. But when real consumer losses show up, the toolbox looks empty.

The issue is this. Antitrust law offers no straight liability route against ‘open early, closed late’ strategies. Plaintiffs and agencies must use mini doctrines of antitrust liability developed in the context of an industrial economy. FTC Chair Khan had this to say about refusal to deal: ‘This type of “open first, closed later” scheme does not quite fit a traditional “refusal to deal” framework, and the tactic can be anticompetitive even if the platform did not have a duty to deal.’ Other routes like excessive pricing are equally barred by strict judicial interpretation. As we know from Justice Scalia’s Opinion in Trinko, the charging of monopoly prices ‘is not only not unlawful; it is an important element of the free-market system.’ The EU does not fare much better. Its competition law embeds more mini doctrines of antitrust liability. And yet, ‘open early, closed late’ strategies have mostly evaded antitrust intervention.

The ‘open early, closed late’ paradox is a good metaphor for a lot of what’s wrong with antitrust law today. Show antitrust law a new case of straightforward consumer harm, and it will cling to history or ideology in service of inaction. Academics will no doubt be reminded of the student not knowing the answer to a hard exam question, who ends up regurgitating litanies of related precedent without solving the problem. There is manifest consumer harm though. As Carl Shapiro noted, ‘The most troublesome cases arise when the dominant firm appears to be following a strategy of “open early, closed late.”’

At the same time, past episodes of antitrust expansion underline the need for guardrails. Broadening the types of business conduct attracting antitrust liability is not cost-free. A legitimate concern for economic efficiency and administrability should inform any attempt to formulate a framework for intervention against ‘open early, closed late’ strategies.

How should antitrust law approach platforms’ ‘treacherous turn’? Current mini doctrines of US and EU antitrust law dance around the problem. We propose a more actionable antitrust standard that does not require leaving behind the efficiency and administrability concerns embedded in judicial precedent.

1. A closer look at the “treacherous turn” story

Before we discuss antitrust laws’ ineffective treatment of ‘open early, closed late’ strategies, let’s take a close look at the treacherous turn story. In economic terms, the treacherous turn occurs when a unilateral change in a platform’s price, terms, or conditions leads to a decrease in user surplus not offset by an increase in quality or quantity.

Several illustrations of the treacherous turn story spring to mind. Last year, Disney+ increased the price of its ad-free streaming service from $7.99 to $10.99 – a 50% increase – while introducing ads for consumers staying with the original $7.99 subscription. But the treacherous turn story need not concern consumers. For a decade, Google allowed app developers like Spotify to conclude transactions outside of its Play Store (thus avoiding its 30% commission fee) but now insists on receiving its cut from all in-app purchases. Or take Apple, which used to allow app developers to track users to support their advertising-funded business models but now severely restricts that possibility under its new App Tracking Transparency policy.

Platforms can afford to take such a turn when users have few outside options. For example, consumers may face high switching costs locking them into the platform. Or business users might have made significant relation-specific investments that are not economically transferable to other platforms. A successful treacherous turn requires a substantial degree of demand-side inelasticity. Cases will thus tend to concern situations of monopoly power and entry barriers on the supply side.

Now, why is the treacherous turn different from the well-known problem of opportunistic recontracting encountered in seller–buyer relations subject to transaction and mobility costs? Opportunism is both more likely and more harmful in digital markets. It’s more likely because the economics of multisided markets advise entry with user subsidies that allow the platform to grow, scale, and reach critical mass. But the economic conditions required upon entry do not supply enough returns to break even. Inevitably, successful profit-maximizing platforms must renege on them once they’ve crossed the tipping point.

It’s more harmful because digital services emerged against a background of imperfect legal institutions. Contracts of adhesion – now the industry practice for digital services – mean that platforms can coerce users into new conditions at any time (something antitrust intervention has occasionally tried to address). It is not unreasonable to expect ‘opportunistic self-interest with guile’ – as Nobel prize economist Oliver Williamson called it – in the absence of clearly applicable property, contractual, or regulatory institutions.

2. Antitrust’s refusal to supply detour

US and EU law is not entirely disarmed against a platform treacherous turn. In US antitrust law, consumer exploitation is not a cognizable harm as such (remember Trinko). But the US Sherman Act embodies a prohibition of restraints of competition leading to competitive injury that can be relevant to the issue of platform treacherous turns. Aspen Skiing offers an illustration. The Supreme Court found that Aspen Skiing had successfully monopolized the market for downhill skiing services by cutting off its main competitor’s access to its infrastructure. The existence of a years-long cooperation between Aspen and the plaintiff was critical in allowing the Court to declare the refusal to supply unlawful under Section 2 of the Sherman Act. Short of evidence indicative of recontracting, it is unclear that the Court would have been able to establish antitrust liability. As the Court later held, if Aspen Skiing is ‘at or near the outer boundary’ of Section 2, recontracting kept it within its reach.

The situation is only slightly different in Europe. EU competition law allows for more direct control of platforms’ treacherous turn. Article 102(a) TFEU prohibits dominant firms’ unfair prices and trading conditions. Reluctant to engage in price regulation, competition agencies in Europe have, however, been conservative in the application of Article 102(a) TFEU. Like in US law, recontracting comes indirectly within reach of Article 102 TFEU through the prohibition of dominant firms’ exclusionary refusals to supply. The finding of abuse in Microsoft, for example, heavily rested on the circumstance that Microsoft had withheld interoperability information that it had previously supplied to Sun Microsystems. The European Commission (EC) described the refusal as ‘open early, closed late’ in all but name.

Overall, refusal to supply law places a limited check on ‘open early, closed late’ strategies. Chair Khan sensed this when she declared ‘open early, closed late’ schemes do not fit the traditional refusal to deal framework. Refusal to supply law aims at protecting dominant firms’ rivals from competitive injury. However, the primary economic evil of ‘open early, closed late’ strategies is harm to trading partners. Refusal to supply doctrine’s insistence on a showing of injury to competition is likely to prevent the application of antitrust law in most cases. In addition, as scholars and practitioners of antitrust across the Atlantic know well, courts impose very high burdens of proof in refusal to supply cases.

In practice, agencies and plaintiffs have tried to avoid the refusal to supply doctrine. In Microsoft, the EC initially argued that the anticompetitive conduct constituted a disruption of supply rather than a flat-out refusal, which justifies a less demanding liability test (it abandoned the argument on appeal). And in its self-preferencing case against Google, the EC developed a sui generis theory of leveraging abuse to overcome the refusal to supply argument raised by the defendant. In confirming the Google Shopping decision, the General Court stressed how Google’s search engine ‘is, in principle, open’, which distinguishes itself from the infrastructure usually at stake in refusal to supply cases. Google benefits from results from third parties, which improves its search engine and drives growth. Absent anticompetitive intent, it is therefore irrational to close the engine by populating it with your own results.

The adoption of surrogates to the refusal to supply doctrine hardly improves antitrust law. These approaches, when successful, increase legal uncertainty. The proliferation of webs of mini doctrines raises the risk of different and, sometimes, inconsistent antitrust standards. As Google Shopping exemplifies, the outcome is increased litigation – courts being asked to make sense of the law – not remediation.

3. Towards an antitrust rule for the treacherous turn

Antitrust law should address platforms’ treacherous turn with a direct rule. But what should the rule look like? First, it must either fall within the formal requirements of the antitrust statute or be included in specific legislation. The rule should respect the basic premises of administrability and economic efficiency. And in a digital context, a concern for not overly undermining platform monetization opportunities should is paramount. That concern rules out a per se approach to platforms’ treacherous turn. But the concern does not require costly examination under the rule of reason either.

A quasi-per se rule could strike the right balance. Under a quasi-per se rule, the agency or plaintiff is required to show a threshold level of market power and the defendant is given an opportunity to advance a business justification. The quasi-per se rule is the preferred approach for practices that present procompetitive justifications like tying or vertical restraints.

As far as the market power threshold is concerned, the rule should address situations where the treacherous turn is not ‘self-defeating’. In a competitive market, the expected loss of users stemming from an adverse change in the platform’s terms and conditions (T&Cs) should make an ‘open early, closed late’ strategy economically unprofitable and in turn irrational. A showing of market power should allow an inference that the self-defeating constraint does not hold in the particular case. Relation-specific investments can play an important evidential role here.

The rule should allow the admission of business justifications that establish an absence of ‘opportunistic self-interest with guile’. The universe of possible defenses is not infinite. The evidential discussion should focus on whether the justification constitutes the only plausible explanation for the impugned conduct. For example, a surge in inflation might justify a unilateral change in subscription prices. But the justification fails when the price increase exceeds the inflation rate. Even more clearly, Twitter banning links to Mastodon because they are ‘potentially harmful’ is not a credible justification in light of past company practice and other platforms’ security policies. The more plausible explanation is that Twitter wanted to stop users from flocking to a competitor following a series of controversial policy changes.

On its face, a quasi-per se rule against ‘open early, closed late’ appears very harsh. It is not hard to conceive of an extreme case in which antitrust law would force a platform to provide its services for free in perpetuity. More generally, there are obvious objections to a law that would affirm an economic principle whereby platforms’ trading conditions are fixated upon entry, preventing incremental business model change and experimentation.

Careful remedy design can avoid the problem of business model ossification. Let us consider a hypothetical treacherous turn in which a platform starts charging for a service that had been until then free. A reasonable compromise can be found by prohibiting the platform to charge existing users while leaving it free to charge new users. In practice, the agency or court will either restrict the liability finding to existing users or will fashion its injunction such that the platform remains free to charge new users. And the remedy does not have to concern prices. Say a platform changes its T&Cs to engage in more invasive data collection. Under our rule, the platform can do so for new users, but it will have to maintain previous levels of data protection for existing users.

The proposed remedy follows the ‘surprise theory’ of harm established in aftermarket cases like Kodak. The Supreme Court found it legitimate to protect existing users from rapacious conduct but considered that new users adopting the service with full information on the terms of trade could not be deemed victims of anticompetitive exploitation. From an economic perspective, the Kodak case suggests that late adopters free-ride on experimentation and information costs incurred by early ones. Through that lens, our proposed remedy can be thought of as a risk reward for early adopters, who helped the platform scale. The remedy does not prevent the platform from developing alternative monetization models.

4. A final word

Some laws beyond antitrust place limits on platforms’ treacherous turns. Contract law can moderate the problem in some circumstances. And in the EU, the Platform-to-Business (P2B) Regulation tried to address the problem in 2019 by imposing a transparency obligation and a reasonable notice period on platforms. The P2B regulation rightly noted that ‘sudden changes to existing terms and conditions may significantly disrupt business users’ operations.’

Contract law or transparency regulation can at best place a soft constraint on treacherous turns. Individual or small business users will never enjoy enough bargaining power to negotiate effective contractual protections in their relations with platforms. In addition, neither contract law nor texts like the P2B Regulation are backed by public enforcement. This, we believe, makes the case for a more effective substantive and institutional response. It is time to experiment with an antitrust turn.

Friso Bostoen & Nicolas Petit

Citation: Friso Bostoen & Nicolas Petit, “Platforms’ Treacherous Turn”, Network Law Review, Summer 2023.


Related Posts