Choice Screens Move the Needle, Self-Preferencing Fight Looms

On 27 April 2026, the National Bureau of Economic Research published working paper w35112, a difference-in-differences analysis of mobile browser usage across the European Union following the Digital Markets Act's rollout of mandated choice screens. The paper, authored by a team of seven economists including Michael Luca and Stefan Hunt, asked a question that has haunted antitrust enforcers for a decade: can an active-choice intervention actually dislodge the power of a preset default? The authors' answer, parsed carefully, is yes. The effect is real, it is measurable, and it is modest. That last word, 'modest,' will be the one that lawyers on four continents argue over for the next five years.

The NBER study examined browser-choice screens presented to iOS and Android users in the European Economic Area after the DMA took effect. The researchers measured what happened when a screen appeared during device setup, asking users to select a default browser from a list of options rather than silently inheriting Safari on iPhone or Chrome on Android. The finding that a statistically significant share of users switched to alternatives recasts a debate that had been dominated by theory and anecdote. As the European Commission's own two-year DMA review, released the following day, concluded the law was 'fit for purpose,' the NBER paper supplied something the review did not: causal evidence, not just compliance metrics, that the architecture of a choice intervention matters. The NBER working paper will be cited in regulatory proceedings from Brussels to Washington for precisely that reason.

The DMA's choice-screen obligation, housed in Article 6, is the most visible expression of a regulatory theory that has gained extraordinary traction since 2020: that platform self-preferencing is a distinct competition harm, and that a structural or quasi-structural remedy, rather than a conduct prohibition coupled with a fine, is required to correct it. The theory holds that a platform that simultaneously operates a marketplace and sells its own goods or services on that marketplace faces an irreconcilable conflict. It will design ranking algorithms, default settings, and user interfaces that steer consumers toward its own offerings, not because those offerings are better, but because the platform controls the infrastructure. The choice screen is the remedy designed to break that feedback loop without breaking up the firm.

In the United States, the same theory underpins the Department of Justice's search antitrust case against Google, decided in the government's favor by Judge Amit Mehta in August 2024. The liability finding was clear: Google illegally maintained its monopoly in general search services and search text advertising through exclusive default-distribution agreements with browser developers, device manufacturers, and carriers. The remedy phase, however, has been far more contested. Judge Mehta's final remedial order, issued in late 2025, stopped short of the structural separation the DOJ had initially sought. The government appealed that order in February 2026, arguing the remedy failed to address the self-preferencing dynamic at the core of the violation. The DOJ appeal places the choice-screen question directly before the D.C. Circuit.

The procedural posture matters. The D.C. Circuit will review Judge Mehta's remedial order under an abuse-of-discretion standard, a deferential bar that requires the government to show not that a better remedy exists but that the chosen remedy is legally inadequate to cure the violation. The DOJ's opening brief argues that prohibiting Google from paying for default placement while permitting the company to continue self-preferencing its own search product on Chrome and Android is an internally contradictory remedy. The competitive harm was the foreclosure of rival search engines from distribution channels. The remedy leaves Google in control of the two largest distribution channels and does not require a choice screen. The D.C. Circuit, whose docket includes a series of antitrust rulings skeptical of structural remedies for conduct violations, will hear argument in the fall.

The remedy leaves Google in possession of the very distribution channels through which the illegal monopoly was maintained, without imposing any neutral-intermediation obligation on those channels., DOJ opening brief, United States v. Google LLC, No. 23-5153 (D.C. Cir. filed Feb. 4, 2026)

Meanwhile, the European Commission has been operating under a different legal framework and a different standard of proof. The DMA does not require the Commission to prove anticompetitive effect. It requires designated gatekeepers to comply with a set of ex ante obligations, including the choice-screen requirement. The Commission's enforcement has been aggressive. On 28 April 2026, it published a formal review finding the DMA had given users 'more control over their experiences and devices,' while acknowledging that many citizens do not know the law exists. The review noted that the Commission had opened non-compliance investigations into Apple, Google, and Meta, and that fines totaling more than seven billion dollars had been levied against U.S. technology firms over the preceding two years, a figure reported by CNBC that has drawn sharp pushback from the Trump administration.

What the NBER Study Actually Found, and What It Did Not

The NBER paper measures daily active usage of mobile browsers, not installation counts or survey responses, which matters because prior studies of choice screens often relied on self-reported preference data that did not capture whether users actually changed their behavior. The authors find that the DMA-mandated choice screen increased the daily active usage of non-default browsers by several percentage points, concentrated among users who were presented with a screen during device setup. The effect persisted through the study period, suggesting it was not merely a novelty response. Critically, the magnitude of the shift varied by platform: the effect was larger on iOS, where Safari was the entrenched default, than on Android, where multiple browsers already competed for pre-installation placement.

The study's limitations are as instructive as its findings. The paper measures browser choice, not search engine choice, and the distinction is not academic. A user who switches from Safari to Firefox but continues to type 'google.com' into the address bar has not altered the search market at all. The DOJ's case against Google was about search distribution, not browser distribution, and the remedy the DOJ wants includes a search-engine choice screen presented during device setup and at periodic intervals thereafter. The NBER study provides proof of concept for the choice-screen mechanism but does not answer the question of whether search-engine choice screens would produce shifts of comparable magnitude in a market where Google's brand is far more entrenched than any browser's.

A separate line of inquiry, pursued by the Commission's DMA enforcement team, concerns the design of the choice screen itself. The order in which options appear, the number of options shown, the descriptive text accompanying each option, and the visual weighting of the interface all affect outcomes. In 2024, the Commission opened a non-compliance investigation into Apple after competitors complained that the iOS browser-choice screen was displayed only once, during device setup, and was not shown to users who had already completed setup before the DMA took effect. Apple subsequently modified the screen's design and frequency. The NBER study captures data from the period after those modifications, meaning its results reflect a relatively mature intervention, not the first iteration of the policy.

The self-preferencing question also reaches beyond browsers and search engines. On 16 March 2026, European publishers and technology firms, including several browser developers and search startups, sent a joint letter to Competition Commissioner Teresa Ribera urging the Commission to conclude its nearly two-year investigation into Google's alleged favoring of its own services in search results. That investigation, separate from the DMA's ex ante obligations, proceeds under traditional Article 102 competition law and addresses whether Google's ranking algorithms systematically demote competing vertical search services, such as travel and shopping comparison sites, while elevating Google's own offerings. A finding of infringement would require a remedy, and a choice screen for vertical search is among the remedies under consideration.

Google, for its part, filed a set of proposed concessions with the Commission on 6 May 2026, offering to change how its search engine ranks news content in an effort to settle a separate DMA investigation. The proposals, reported by The Next Web, do not address the broader self-preferencing investigation but signal the company's willingness to negotiate interface-level remedies when the alternative is an infringement finding carrying a fine of up to ten percent of global turnover.

The transatlantic dimension of the self-preferencing debate has become more pointed since January 2025. The Trump administration has characterized EU digital regulation as a de facto tariff on American technology exports, and administration officials have raised the prospect of retaliatory measures under Section 301 of the Trade Act. The CNBC report on EU fines noted that the cumulative total levied against U.S. firms since the DMA's entry into force exceeds seven billion dollars. Commission officials, for their part, have emphasized that the DMA applies to all gatekeepers regardless of nationality, though to date all six designated gatekeepers are American. The political economy of the choice-screen debate is now inseparable from the trade-policy friction, a reality that complicates the appellate path of any remedy that imposes a mandatory choice architecture on a U.S. firm.

Apple has mounted its own challenge to the DMA's choice-screen mandate, arguing in a statement released 5 May 2026 that the obligation to present users with a browser ballot during device setup creates a 'privacy-threatening' vector by exposing users to unfamiliar third-party software before they have configured privacy settings. The argument, reported by Macworld, is unlikely to persuade the Commission, which views the privacy objection as a repackaging of the commercial objection. But it will resonate in Washington, where policymakers on both sides of the aisle have shown increasing interest in the intersection of competition policy and data privacy, and where the incoming administration's antitrust enforcers have signaled that they view European digital regulation as a competitive threat to American firms.

What the appellate panel in the D.C. Circuit will care about, and what the Commission's enforcement staff already cares about, is the counterfactual question that the NBER study begins to answer. If a platform is prohibited from self-preferencing, will users choose alternatives in sufficient numbers to make the market contestable? The study suggests the answer is yes, but only if the intervention is well designed, repeatedly presented, and accompanied by meaningful information about the alternatives. A poorly designed choice screen, presented once and easily dismissed, is worse than no remedy at all: it creates the appearance of consumer choice while entrenching the incumbent through the very mechanism purporting to correct the market failure. The Commission understands this. The D.C. Circuit will have to decide whether Judge Mehta understood it too.

The debate over choice screens is, at bottom, a debate about where the burden of competition should lie. In a traditional market, the burden is on the entrant to build a better product and persuade consumers to switch. In a market where the incumbent controls the infrastructure through which consumers discover and adopt products, the burden is functionally impossible to meet. The choice screen shifts a portion of that burden back to the incumbent: it must compete, however briefly, on the merits of its product alongside rivals, in a setting where the user's attention is undivided. The NBER study provides the first rigorous evidence that this shift changes behavior. The remaining question is whether the legal systems on both sides of the Atlantic will demand it.

Watch for the D.C. Circuit's scheduling order in United States v. Google. The panel's composition will determine the range of plausible outcomes far more than the briefing will. A panel with Judges Srinivasan and Pillard, both of whom wrote separately in the Microsoft consent-decree enforcement action to emphasize the importance of remedial adequacy, will ask different questions than a panel drawn from the court's more business-deferential wing. On the European side, the Commission's non-compliance decisions against Apple and Google, expected before the summer recess, will specify precisely what constitutes an acceptable choice screen under the DMA. Those decisions will become the compliance baseline for every gatekeeper-designated platform. The experiment is no longer theoretical. The data is arriving. The courts and the Commission now have to decide what to do with it.