ICLE White Paper

Steering in the Fog: The DMA and the Turn from Market Oversight to Market Ordering

Executive Summary

The European Union’s Digital Markets Act (DMA) subjects conduct long governed by Article 102 TFEU—product integration, most-favored-nation clauses, self-preferencing, data combination, and duty-to-deal obligations—to categorical rules stripped of effects analysis, efficiency defenses, and the consumer-welfare standard. This white paper argues that the DMA marks a decisive move from market oversight to market ordering. Where competition law polices market processes through evidence, effects, defenses, and judicial review, the DMA empowers the European Commission to prescribe market outcomes through open-ended standards of “fairness” and “contestability.”

The Commission’s early enforcement record against Meta, Apple, and Google shows how the DMA’s structural-unfairness axiom gives the Commission continuing authority to redesign digital markets from the center. Business-user satisfaction becomes the regime’s proxy for the public interest, while consumer choice is instrumentalized: respected when it moves users away from gatekeeper ecosystems, treated as evidence of insufficient intervention when it does not.

The paper situates the DMA within a broader political-economy shift. Its logic reflects ordoliberal confidence in a nobilitas naturalis capable of defining the market order, Karl Polanyi’s account of market re-embedding, and Herman Schwartz’s cycles of state-market reassertion. But under the DMA, re-embedding occurs through the administrative state, not society. The result is economic statecraft: competition policy used to redistribute rents, shape market structure, and advance strategic objectives. Parallel moves in Article 102 guidance, merger control, and industrial policy suggest that digital markets may be the leading edge of a broader turn toward market ordering.

The paper ends by arguing that the DMA’s costs are not drafting defects but entailments of the ordering model. By removing effects analysis, efficiency defenses, meaningful merits review, and consumer-welfare constraints, the DMA heightens error costs, aggravates the knowledge problem, weakens resistance to capture, and creates an autopoietic evaluative architecture that reproduces its own assumptions rather than tests them. Whether competition law’s disciplining safeguards can be reconstructed inside a regime designed to dispense with them may determine the legitimacy of the next decade of European market governance.

I.                    Introduction

And now hearing you speak I trust you less, no more than Boromir. Nay, stay your wrath! I do not trust myself in this, and I refused this thing, even as a freely given gift. You are strong and can still in some matters govern yourself Denethor; yet if you had received this thing, it would have overthrown you. Were it buried beneath the roots of Mindolluin, still it would burn your mind away, as the darkness grows, and the yet worse things follow that shall come upon us.

— Gandalf, in J.R.R. Tolkien, “The Return of the King” (1955).

The Digital Markets Act (DMA)[1] is the European Union’s effort to empower the European Commission to set the rules of the game in digital markets. It rests on two assumptions that command broad support in Brussels. The first is that digital platforms benefit from significant power asymmetries, allowing them to capture a disproportionate share of the gains from trade by unilaterally dictating the terms of access and interaction.[2] The second is that, given their central role in the modern economy, digital platforms warrant greater public oversight.[3]

This paper argues that the DMA marks a structural shift in how competition and markets are regulated: from market oversight to market ordering. Under Article 102 of the Treaty on the Functioning of the European Union (TFEU),[4] the Commission policed the boundaries of permissible competitive conduct. Under the DMA, it increasingly prescribes competitive outcomes and product-design choices. It sets—or will inevitably end up setting—the prices gatekeepers may charge for alternative app distribution. It dictates the data-sharing architectures through which platforms interact with business users. It evaluates whether consent mechanisms are sufficiently “free” and whether user interfaces are sufficiently “neutral.”

The Commission exercises these powers through standards that are, by legislative design, indeterminate. Rather than anchoring intervention in measurable consumer harm, the DMA rests on a structural-unfairness axiom that treats the competitive advantages of designated firms as presumptively suspect. At the same time, that axiom offers no cognizable way to determine when digital markets are sufficiently fair or contestable, and thus no obvious endpoint. The combination of per se rules and indeterminate standards creates a regime with no internal mechanism for contesting the Act’s foundational presumptions and only limited avenues for disputing the Commission’s interpretation of whether, and when, its objectives have been achieved.

By elevating the Commission’s ability to reshape digital markets above meaningful contestation, the DMA shifts the focus from overseeing markets to ordering them; from correcting specific, demonstrable market failures to redesigning products a priori through centralized administrative control.

To be sure, European Union competition law—which serves here as the DMA’s counterpoint—has never been merely a neutral referee of market competition. The distinction between oversight and ordering is best understood as a spectrum rather than a categorical divide. The Commission has always exercised some degree of market-shaping discretion, and this paper acknowledges that competition law occasionally shades into ordering territory.

The DMA’s novelty is nevertheless twofold. Procedurally, it strips away many of the constraints that historically cabined the Commission’s discretion, including burdens of proof, effects analysis, efficiency defenses, meaningful judicial review, and, importantly, an external yardstick of success. Substantively, it extends regulatory control into the product-design, pricing, and interface choices through which digital markets are constituted.

That discretion is not unlimited. The constraints that remain, however, discipline form rather than intensity. The Act’s foundational presumptions cannot be contested, while “fairness” and “contestability” remain sufficiently elastic that additional intervention can almost always be justified in their name. Together, these procedural and substantive changes move the regime decisively toward the directive end of the oversight-ordering spectrum.

This paper makes three contributions. First, it offers an account of the DMA as a qualitative transformation in the regulation of competition and markets. Second, it situates that transformation within a longer tradition of political economy. Third, it offers a critical account of the market-ordering model of governance that the DMA embodies.

The paper proceeds as follows. Section II introduces the oversight-ordering spectrum as the paper’s organizing framework, explains the two poles, and justifies its use as a lens for understanding the DMA’s significance. Section III shows how European Union competition law generally exemplifies market oversight, while acknowledging that the Commission has always occupied some position on the spectrum rather than a pure endpoint. Section IV demonstrates how the DMA departs from that model, drawing on its per se prohibitions, its novel objectives, and the early enforcement record against Meta, Apple, and Google. Section v situates that departure within the broader post-liberal turn in economic governance. Section VI critically assesses the ordering model. Section VII concludes.

II.   From Oversight to Ordering

The DMA is best understood as part of a broader shift in economic governance: from market oversight to market ordering. Oversight describes a model in which public authority polices the boundaries of competition, intervening only when specific conduct produces legally cognizable harm. Ordering describes a more directive model, in which the state uses law to shape market structure, redistribute opportunities among firms, and steer outcomes toward regulatory objectives.

This section develops that framework. Section II.A defines the oversight-ordering spectrum and situates it within ordoliberal thought, Herman Schwartz’s political economy, and Karl Polanyi’s “double movement.” Section II.B explains why the spectrum is the right lens for understanding the DMA. Section II.C then shows why ordoliberalism itself complicates any simple contrast between neutral market oversight and discretionary market ordering.

A.             The Oversight-Ordering Spectrum

Three intellectual traditions supply this paper’s theoretical vocabulary for the oversight-ordering spectrum. The ordoliberal tradition, which laid the intellectual groundwork for postwar European competition governance, understood oversight and ordering as distinct levers of economic governance. Oversight operated through indirect, legally governed competition; ordering operated through direct command and state direction.[5]

Herman Schwartz’s political economy offers a complementary macro-level framework. The relationship between states and markets is cyclical rather than linear, with periods of market ascendancy alternating with periods of state reassertion, in which governments reclaim authority to define market outcomes.[6] Karl Polanyi’s “double movement” thesis identifies one mechanism driving that cycle: the extension of market logic into new domains provokes a countermovement of social protection, as societies seek to re-embed markets in normative and institutional frameworks.[7]

Each framework illuminates a different, though interrelated, dimension of the oversight-ordering spectrum. Each also casts a different light on what the Digital Markets Act (DMA) represents within the political economy of economic governance, as Sections II.C and v explain.

At one end of the spectrum sits market oversight in its purest form. Under this model, the state polices competitive processes from the outside. It identifies and corrects specific, demonstrable failures but otherwise allows competitive markets to determine outcomes. The paradigmatic instrument is general competition law: a body of principles that applies equally to all market participants, is triggered by identifiable anticompetitive conduct, and proceeds through a judicial or quasi-judicial process.[8] The enforcer must demonstrate harm, consider countervailing efficiency arguments, and withstand scrutiny on appeal. The ordoliberals called this the “indirect steering mechanism.”

At the other end sits market ordering. Here, the state does not merely police existing competitive processes. It actively shapes the structure and outcomes of markets according to its own conception of what those outcomes should be. The tools are correspondingly different: ex ante prohibitions and obligations; price regulation and access mandates that redistribute rents among market participants; and ongoing administrative monitoring that allows regulators to judge not only whether conduct violated a rule, but whether the resulting market structure matches a preferred template.

Classic examples include public-utility regulation, where regulators set access prices for electricity networks or railway infrastructure, though even there the nominal goal remains economic efficiency and output maximization. Postwar dirigiste industrial policy offers a more explicit example: state agencies allocated investment and steered the economy toward politically preferred ends.[9] Some forms of social regulation also fall on the ordering side of the spectrum, particularly when they steer users toward conduct or outcomes regulators deem socially desirable.[10]

The distinction between these two poles is largely analytical. Most regulatory regimes occupy positions between them. The point is not to force every instrument into one category or the other, but to use the spectrum as a tool for assessing where a regime sits and whether its position shifts over time.

B.             Why the Spectrum Matters

The oversight-ordering spectrum is the appropriate lens through which to understand the DMA for three related reasons.

First, the DMA targets conduct already governed by Article 102 TFEU through a categorically different institutional framework. It employs competition law’s enforcement machinery while removing many of the constraints that give that machinery its oversight character. Determining where the DMA sits on the spectrum is therefore also a way of measuring how far it departs from the model it nominally supplements. Because the DMA regulates substantially the same conduct as competition law and is enforced by the same institution, two variables remain constant: the enforcer (the European Commission) and the subject matter (the conduct at issue). That leaves the legal framework itself as the independent variable, and the regime’s character—its degree of discretion, its position on the spectrum, and the outcomes it produces—as the dependent variable.

Second, the spectrum allows the analysis to hold two propositions simultaneously. Competition law has its own ordering moments, catalogued in Section III.C. At the same time, the DMA effects a deliberate and substantial shift toward the directive end of the spectrum. The framework therefore captures movement before either pole is reached and allows that movement to be evaluated by reference to the poles toward and away from which it tends.

Third, the spectrum connects doctrinal analysis to broader developments in political economy. Once the DMA is understood as a move toward ordering, its affinities with other transformations in economic governance become visible: the rehabilitation of industrial policy, the post-liberal turn in competition enforcement, and the resurgence of state authority in areas previously governed through private ordering. On this view, the DMA emerges neither as a truncated version of competition law nor as a wholly separate regulatory instrument. Rather, it represents a new paradigm for structuring the relationship among states, markets, firms, and consumers—less a novel tool designed to complement competition law than a new way of regulating competition itself.[11]

C.             Ordoliberalism and the Oversight-Ordering Spectrum

A brief note on ordoliberalism is necessary before proceeding. Scholars generally understand the ordoliberal tradition as a protective, rather than proactive, philosophy of economic governance.[12]  The nobilitas naturalis—the “aristocrats of the public spirit” charged with safeguarding the competitive order—existed to preserve undistorted competition and prevent concentrations of private economic power from capturing the state.[13] Ordoliberalism is therefore most commonly associated with the idea that competition should be embedded within an “economic constitution”: a rules-based framework insulated from day-to-day political pressures. The enforcer’s role was to guard that order impartially, not to direct outcomes within it.

Less appreciated is ordoliberalism’s reliance on public administration both to enforce that framework and, where necessary, to depart from it. Franz Böhm himself defended a position that reveals a latent compatibility between the ordoliberal model of expert authority and the more discretionary, ends-oriented governance the orthodox account is often taken to reject. The “true core of an economic constitution,” he argued, was “a steering norm, which guides economic happenings in a politically desirable direction.”[14] The competitive order should operate within the “boundaries set by the National Socialist state’s constitutional edicts,” and the state should suspend “indirect economic steering by legally-governed competition” whenever “direct market steering by methodical means of command” was deemed more appropriate.[15]

The public administration—the “aristocracy of the public spirit”—is thus entrusted to interpret political needs and navigate political realities, using the two levers of economic governance accordingly: direct control (ordering) and indirect steering (oversight). What separates oversight from ordering, even within a tradition explicitly committed to the former, is therefore not a categorical difference in institutional design, but a position along a spectrum. That position can, and arguably should, shift under particular political conditions.

The categories themselves are also imperfectly bounded. Even oversight contains an element of design. Determining what counts as “undistorted” competition, what degree of concentration is acceptable, and how those judgments should be measured are themselves exercises in ordering. For the ordoliberals, the crucial distinction was that these political judgments occurred upstream, at the constitutional level, rather than downstream in the enforcer’s case-by-case decisions.

Against this backdrop, European Union competition law settled well toward the oversight end of the spectrum, though never at its extreme. The constraints on the European Commission’s ability to engage in direct steering of competition and economic outcomes remained substantial, particularly when compared with more dirigiste alternatives that largely abandoned competition as an organizing principle.[16]

The DMA’s significance lies in its systematic removal of many of those constraints. In their place, it substitutes per se rules and discretionary standards that leave the Commission to determine, case by case, what “fair” digital markets should look like for each gatekeeper and each core platform service.

III.   Competition Law as Market Oversight

EU competition law provides the baseline against which the DMA’s shift toward ordering can be measured. Article 102 TFEU does not prohibit size, dominance, or vertical integration as such. It intervenes only when the European Commission can show, through a contextual and effects-based inquiry, that specific conduct harms competition to the ultimate detriment of consumers.

That does not mean competition law is purely supervisory. It contains its own ordering elements: presumptions, structural remedies, access obligations, and forms of administrative discretion that allow the Commission to shape market behavior. The point is comparative. Competition law remains anchored in evidence, effects, defenses, and judicial review in a way the DMA deliberately loosens. This section develops that contrast by first describing Article 102 as an oversight regime, then showing how that model operates in practice, and finally acknowledging the ordering elements that already exist within competition law.

A.             Article 102 TFEU as Market Oversight

Article 102 TFEU exemplifies market oversight. No firm becomes subject to special obligations merely because of its size. Dominance is not unlawful in itself and is often understood as the product of competition on the merits.[17] When dominance is established, the dominant firm assumes a “special responsibility” not to distort competition.[18] Even then, Article 102 does not treat any conduct as automatically prohibited or required. While presumptions of harm may apply in some circumstances, they remain rebuttable.[19]

To establish an abuse of dominance, the European Commission must show that, in the specific circumstances of the case, the challenged conduct harms competition to the ultimate detriment of consumers.[20] Article 102’s analytical framework can be sketched as follows.

First, market definition. Because dominance must be assessed within a relevant market, the enforcer must identify the competitive constraints the firm actually faces.

Second, effects analysis. The Commission must demonstrate actual or likely anticompetitive effects, rather than rely solely on structural characteristics that might eventually produce them. The assessment of exclusionary conduct must account for all the circumstances of the case.[21]

Third, the as-efficient-competitor principle. Article 102 targets conduct that excludes rivals capable of competing on the merits. It does not seek to protect rivals that fail because they are less efficient.[22]

Fourth, objective justifications and efficiency defenses. Conduct that might otherwise appear exclusionary may nevertheless be lawful if it produces efficiencies or serves legitimate business purposes.[23]

Fifth, meaningful judicial review. The Commission’s theory of harm must withstand scrutiny against an evidentiary record, and courts may annul decisions that rest on flawed legal or economic analysis.

The overarching logic is that intervention is reserved for cases in which a firm’s market power is sufficiently substantial and durable to escape ordinary competitive discipline. Competition rewards efficient firms, penalizes inefficient ones, and channels resources toward their highest-valued uses. Article 102 intervenes only when market power becomes strong enough to undermine that process by foreclosing as-efficient rivals and allowing a firm to capture surplus that competition would otherwise return to consumers.[24]

B.             Effects-Based Analysis in Practice

Two categories of conduct—most-favored-nation (MFN), or “price-parity,” clauses and self-preferencing—illustrate the oversight model in practice and provide useful counterexamples to the DMA’s per se approach. A third category, tying, is closely related to self-preferencing and warrants brief attention as well.

Under EU competition law, MFN clauses are evaluated through careful, contextual analysis. No EU court has classified MFNs as restrictions “by object” under Article 101 TFEU, a category reserved for practices that experience shows are inherently likely to restrict competition and harm consumers.[25] Because MFNs are vertical agreements whose effects are economically ambiguous, a “by object” approach has consistently been regarded as inappropriate.

In Booking.com, the Court of Justice held that parity clauses could not easily be justified as ancillary restraints, but declined to treat them as automatically unlawful.[26] The German Federal Court of Justice likewise conducted a detailed effects analysis of Booking.com’s narrow MFNs, emphasizing that “the extent to which the narrow price parity clause affects competition depends on the individual case.”[27] The Vertical Block Exemption Regulation presumes narrow MFNs lawful and requires individualized assessment even for broad MFNs that fall outside the safe harbor.[28] The European Commission’s handling of Amazon’s parity obligations under Article 102 TFEU followed the same contextual approach, explicitly recognizing legitimate commercial purposes, including the prevention of free-riding.[29]

Herbert Hovenkamp, the scholar most frequently cited by the U.S. Supreme Court in antitrust cases, has argued that MFNs become problematic only when combined with product-specific market power.[30] That reflects a broader point. MFNs are not categorically prohibited because it is not obvious whether they restrict competition or promote it. They may encourage platform investment, reduce free-riding, and generate inter-platform pricing benefits through the bargaining leverage that parity clauses provide. Those gains may be modest or absent where platform competition is already weak, and they may be outweighed by higher retail prices.[31] The point is that the welfare calculus cuts both ways and depends on the facts of the case and the competitive conditions in the market.

The treatment of self-preferencing is no less contextual. Google Shopping is the foundational EU case. The Court of Justice held that Google’s promotion of its own comparison-shopping service constituted an abuse of dominance, but it was equally clear that not every instance of a firm favoring its own products departs from competition on the merits.[32] Liability depended on a contextual, effects-based inquiry that examined, among other things, causation, foreclosure effects, and the relevant counterfactual.[33] The Google Android litigation reflects the same pattern, requiring extensive effects analysis before liability could be established.[34]

That framework makes sense if the objective is to oversee markets rather than to shape them. Under conditions of endemic uncertainty, conduct that appears anticompetitive may in fact be competition on the merits.[35] Self-preferencing—a common business practice across industries[36] —may reflect vertical efficiencies, including lower transaction costs, stronger quality control, better matching between users and products, or simply the return on investments that created the platform in the first place.[37]

To be sure, those benefits may be modest or outweighed by foreclosure effects where downstream rivals are equally efficient and preferential treatment proves decisive. They may also be substantial. The difficulty is compounded by the fact that “self-preferencing” is an imprecise legal category. The term encompasses a variety of practices, including tying and refusals to deal, which may or may not be harmful depending on the circumstances and the legal test applied.[38]

That observation leads naturally to tying, to which the same principles apply mutatis mutandis. Self-preferencing is often analytically related to tying because both involve a dominant firm leveraging a position in one market to extend its reach into another.[39] Here again, EU courts have never treated tying as per se unlawful. Industrial-organization economics has long recognized that tying can generate efficiencies, including lower costs, quality improvements, reduced search costs, the elimination of double marginalization, and, in digital markets, technological integration that creates functionalities the products could not provide separately.[40]

More broadly, vertical integration—of which tying and self-preferencing are downstream manifestations—can reduce transaction costs, improve coordination, and increase consumer value. For that reason, EU competition law does not treat it as presumptively unlawful.[41]

C.             The Ordering Elements of Competition Law

This account would be misleading if it portrayed European Union competition law as purely supervisory. To a committed free-marketeer, even competition law in its most modest, efficiency-oriented form is no nightwatchman state. It empowers an administrative authority to break up firms, rewrite contracts, limit the exercise of private property rights, compel or prohibit dealings with particular parties, and second-guess pricing decisions. Many classical liberals and libertarians would regard such powers as fundamentally inconsistent with a genuinely “free” market.[42]

First, even under the consumer-welfare standard, enforcers and judges necessarily retain some discretion. How much of a price increase does a quality improvement justify under the Article 101(3) efficiency defense? What time horizon is appropriate for measuring a merger’s competitive effects? These are not purely technical questions. The standard itself also contains categories that bypass full effects analysis. Although narrowly construed, the “by object” category under Article 101(1) permits condemnation without a comprehensive inquiry into competitive effects.

Second, EU competition law authorizes enforcers to intervene deeply in firms’ business models and corporate structures. The European Commission may regulate prices, restructure distribution arrangements, prohibit mergers, order divestitures, and impose access or duty-to-deal obligations on dominant firms, among other remedies.[43] As Lee McGowan observed, EU competition policy has long been distinctive as “one of the few areas where the Commission not only is responsible for direct policy implementation but also possesses wide discretionary powers both as regulator and enforcer of policy.”[44] The Commission therefore occupies a central role in shaping the operation of the regime.

Recent developments suggest a further expansion of the Commission’s capacity to shape markets and make complex value judgments. The Commission’s 2024 Draft Guidelines on Article 102 TFEU, for example, point toward a partial revival of form-based approaches. By de-emphasizing effects analysis and reducing the Commission’s obligation to engage with all available evidence, form-based approaches lower the evidentiary threshold for intervention against dominant firms.[45]

The Draft Merger Guidelines point in the same direction. They expressly acknowledge the Commission’s discretion to weigh “imponderable” considerations such as sustainable development, opportunities for small and medium-sized enterprises, and supply-chain resilience—considerations that historically fell outside the core remit of European Union merger control.[46]

Recognizing these ordering elements within competition law does not undermine the oversight-ordering thesis. The claim is not that competition law ever occupied the pure oversight pole. It is that competition law sat considerably closer to that pole than the DMA does, and that the DMA’s significance lies in how far, and how deliberately, it relocates the regime toward ordering. The next section develops that argument more directly.

IV.   The DMA as Market Ordering

The DMA does not merely supplement EU competition law. It transforms the mode of governance. Where Article 102 TFEU asks whether specific conduct harms competition in a defined market, the DMA begins from a legislative judgment that certain firms, practices, and market structures are presumptively unfair. It then empowers the European Commission to translate that judgment into continuing obligations, product-design mandates, pricing constraints, and data-sharing architectures.

This section shows how that shift operates. Section IV.A explains how the DMA replaces effects analysis with categorical rules. Section IV.B shows how “fairness” and “contestability” function as open-ended ordering principles rather than measurable competition-law standards. Section IV.C examines early enforcement against Meta, Apple, and Google as examples of market design. Section IV.D argues that the DMA treats business-user satisfaction as a proxy for the public interest. Section IV.E then shows how consumer choice becomes a regulatory instrument, respected only insofar as it moves users away from gatekeeper ecosystems and toward rivals and third parties.

A.             From Effects Analysis to Categorical Rules

The DMA’s subject matter overlaps substantially with conduct already addressable under Article 102 TFEU: tying, self-preferencing, data combination, refusals to deal, and most-favored-nation clauses.[47] The key difference lies in the legal apparatus. Once a firm is designated as a “gatekeeper,” it becomes subject to a comprehensive set of obligations regardless of its specific business model, the competitive dynamics of the markets in which it operates, or any evidence that its conduct causes harm.[48]

Once designation is established, the obligations in Articles 5–7 apply automatically. The European Commission need not show, case by case, that a particular gatekeeper’s conduct causes harm.[49] Effects analysis is discarded. No efficiency defense is available beyond narrow security and privacy carve-outs. Judicial review is confined largely to manifest error.[50]

The absence of market definition further reflects the DMA’s structuralist and static character. Competition law requires a fresh market definition in each case because markets and competitive dynamics change. The DMA instead fixes its scope to a predetermined list of core platform services, which can be expanded or narrowed only by amending the regulation through the ordinary legislative procedure.[51]

The inquiry into effects and competitive dynamics that characterizes competition enforcement under uncertainty thus gives way to taxonomy.[52] The question becomes whether conduct is “allowed” or “not allowed.” The focus shifts from whether a remedy is justified to whether regulated firms have complied with remedies already imposed.

The contrast with the oversight model described in Section III is stark. The guardrails that the Court of Justice erected in Google Shopping—including its insistence that not every instance of favoring one’s own products infringes Article 102 TFEU[53]—are rendered largely moot by the DMA’s per se approach.[54] The DMA retains the ban on self-preferencing and extends it beyond the limits that Google Shopping, on any reasonable reading, would allow.

Under the DMA, self-preferencing is not unlawful at a particular juncture for a particular undertaking because the evidence shows that it forecloses rivals to the ultimate detriment of consumers. It is branded categorically “unfair” for any current or future gatekeeper, across the Act’s regulated products and services—indefinitely.

The next question is why the DMA regulates conduct that EU competition law already covers in detail.[55]

B.             Fairness and Contestability as Ordering Principles

A deceptively simple explanation accounts for the DMA’s divergence from competition law: the DMA embeds a different understanding of competition’s role in choosing winners and losers and distributing rents in digital markets. That understanding is, in important respects, incompatible with the logic of EU competition law.

As the European Commission has repeatedly recognized, including in the text of the DMA itself, EU competition law protects competition for the ultimate benefit of consumers. Under that standard, consumer harm is necessary but not sufficient for liability. Conduct is condemned only where it strengthens a firm’s power to extract surplus by weakening competitive constraints and is not reasonably necessary to secure offsetting efficiencies. Article 102 TFEU therefore reaches conduct that harms competition “to the detriment of consumers,” not conduct that merely disadvantages a less efficient rival.

The consumer-welfare standard supplies a measurable, effects-based yardstick against which intervention can be tested. Defining “consumer welfare” is not always easy, but the standard at least directs the inquiry toward output, price, quality, innovation, and other welfare-relevant effects. Thomas Lambert has recently framed antitrust injury in similar terms: consumer harm is necessary but not sufficient; challenged conduct must also enhance a firm’s ability to extract surplus by weakening competitive constraints, and it must not be reasonably necessary to secure offsetting efficiencies.[56]

The final DMA text deliberately omits “consumer welfare” after the European Parliament’s proposed amendments to include it were rejected.[57] That omission matters. If consumer welfare is the linchpin of modern European Union competition law, a regulation designed to institutionalize a different model of competition governance would naturally dispense with it. One feature of the consumer-welfare standard that the DMA expressly cuts out is efficiency analysis and, more broadly, economic analysis. Recitals 10 and 11 make clear that efficiency arguments are generally immaterial to DMA enforcement. The operative provisions confirm the point: the Act leaves no meaningful room for economic effects analysis.

Indeed, the DMA states that “fairness” and “contestability” are its objectives “independently from the actual, potential or presumed effects of the conduct of a given gatekeeper covered by this Regulation on competition on a given market.”[58] Recital 33 defines “fairness” only negatively: unfairness involves “an imbalance between the rights and obligations of business users where the gatekeeper obtains a disproportionate advantage.” That definition supplies no benchmark, counterfactual, or limiting principle. The implicit comparison is not between observed conduct and a competitive equilibrium. It is between the current distribution of rents and a preferred redistribution known in advance only to the Commission.

The one clear premise is that gatekeepers should be made worse off because they currently “obtain a disproportionate advantage” from core platform services. The DMA’s operative provisions reflect that premise. Article 6(5), which prohibits self-preferencing, permits gatekeepers to favor third-party offerings over their own but not the reverse. Data-sharing and interoperability obligations require gatekeepers to make competitive advantages available to rivals, often at no charge. Article 6(2) bars gatekeepers from using business-user data to compete with those users, even where that competition might benefit consumers.

In each case, what the DMA calls “fairness” simultaneously levels down gatekeepers and levels up business users. It redistributes rents and, in practical terms, competitive positions. That redistribution rests on the axiomatic premise that existing arrangements cannot be fair because they emerged from structurally skewed markets.[59]

“Contestability” operates in the same way. Unlike its classical economic analogue, which is descriptive and agnostic about market structure,[60] the DMA’s contestability is normative and outcome-oriented. It asks not whether incumbents could, in principle, be disciplined by entry threats, but whether rivals are in fact able to enter and take market share from gatekeepers. Where they fail to do so, the market is treated as insufficiently contestable, and the absence of the desired structural outcome becomes presumptive evidence of a regulatory deficit.

The structural-unfairness axiom—the idea that designated core platform services produce unfair outcomes because they are structurally unfair—therefore carries a further implication. “Fairness” and “contestability” become circular and open-ended. The reasoning forms an ouroboros: whenever outcomes are not fair enough, structural unfairness persists; whenever any asymmetry remains, the distribution of rights, obligations, and rents is insufficiently fair.

There is no external yardstick to confirm or refute that diagnosis. There is no “but for” competitive scenario, no measurable equilibrium, and no limiting principle that can show when a market has become fair or contestable enough. Jacques Crémer, Yves-Alexandre de Montjoye, and Heike Schweitzer, among others, urged the DMA’s drafters to define fairness and contestability more precisely because enforcement, compliance, and judicial review all require a shared understanding of what the Act’s organizing goals demand.[61]

That advice is analytically sound, but it may be structurally threatening to the DMA’s design. If “fairness” were defined precisely enough to generate determinate answers, it would become a justiciable standard capable of constraining Commission discretion. It would invite the sort of effects-based analysis and lengthy evidentiary inquiry the DMA was designed to circumvent.[62]  In short, it would function too much like competition law.

Conversely, the Commission’s ability to iterate on the meaning of “fairness” and “contestability” case by case—to redistribute rents and impose its preferred market architecture—depends on those standards remaining directionally clear but ultimately incalculable. This purposive indeterminacy is a condition of governance whose aim is not to correct identifiable market failures, but to reshape markets. What Crémer and others identify as a potential shortcoming of the DMA may therefore be one of its core features: a deliberate facilitator of the Commission’s new role as helmsman of digital markets.

C.             Enforcement as Market Design

Three enforcement episodes illustrate the scope of the Commission’s expanded authority and the extent to which it has assumed functions that competition law traditionally reserved for markets, courts, and consumers. The DMA does not give the Commission unlimited discretion.[63] But it does make it easier for the Commission to impose its preferred vision of digital markets on designated companies, requiring them to operate on terms selected by the regulator rather than revealed by markets.

In its April 2025 decision, the Commission found that Meta’s “consent or pay” model breached Article 5(2) DMA.[64] That model offered users a choice between free, ad-supported Facebook and Instagram services and a paid, ad-free subscription priced at €10 to €13 per month. The Commission’s reasoning rested on its interpretation of the “equivalence” requirement: because a subscription “entails a monetary burden,” it constitutes a different “condition of access” from a free service.[65] If the personalized-ads version is free, the less-personalized alternative must also be free.

The Commission added that it “need not consider” the economic consequences of this interpretation for Meta’s business model unless Meta invokes Article 9(1) DMA.[66] When fewer than 1% of users chose the paid option, the Commission treated that result not as revealed consumer preference for ad-supported services, but as confirmation that the paid alternative had been priced to deny users a genuine choice.[67] In effect, the Commission substituted its own view of how the market should operate—more privacy, fewer personalized ads, and a free alternative—for the market’s revealed outcome. Under EU competition law, that result would have required a considerably more demanding showing.[68]

The enforcement record against Apple is more extensive and more revealing of the Commission’s willingness—indeed, its duty under the DMA—to dictate product design. In its April 2025 Article 5(4) decision, the Commission found that Apple’s user-facing warnings about leaving the App Store ecosystem, the so-called scare screens, were “not neutral and objective.” It therefore imposed a €500 million fine and a 60-day compliance deadline.[69] The Commission simultaneously issued preliminary findings that Apple’s terms for alternative app distribution breached the DMA: its Core Technology Fee allegedly disincentivized alternative distribution, its eligibility requirements were “overly strict,” and its multi-step user journey for installing alternative app stores was “overly burdensome and confusing.”[70]

The Commission also articulated a three-part test for permissible fees. Any commission must be related in time and scope to the initial acquisition of the user, commensurate with the value of the matchmaking function actually performed, and adjusted for any remuneration the gatekeeper already receives from facilitating that acquisition.[71] Establishing the parameters of a “permissible” fee charged by a private actor on its own platform is, in substance, ex ante price regulation—though without the institutional apparatus that rate-setting ordinarily requires. As Oscar Borgogno and Giuseppe Colangelo have noted, there is no cost model, no specialist tribunal, and no formal appeals procedure.[72]

Under EU competition law, a similar result would have required the Commission to show either that Apple’s fees were excessive in the United Brands sense—that is, bearing no reasonable relation to the economic value of the service, typically on the basis of a cost analysis—or that they foreclosed as-efficient rivals. In either case, the Commission would have had to confront Apple’s security and privacy arguments as objective justifications. None of that was required under the DMA.

The scarcity of excessive-pricing cases under Article 102 TFEU reflects the difficulty competition authorities face in determining the “right” price. The problem is harder still for a dominant platform, where there may be no competitive comparator against which to apply the second limb of the United Brands test.[73] The Commission is likely to face the same conceptual problem when determining what counts as a “fair” fee for Apple’s ecosystem. The difference is that much less now stands in its way, and much more incentivizes it to proceed.

Once the evidentiary safeguards of an Article 102 excessive-pricing case fall away—no need to prove harm, no foreclosure test, no specialized rate-setting tribunal—the Commission must still follow its own three-part test, but little else constrains its ability to set fees or revise the test as it goes. The problem that makes excessive-pricing cases so rare—the need for an enforcer to second-guess market mechanisms and price signals—therefore reappears inside a legal architecture that has discarded the burdens that once forced the enforcer to confront the challenges of price-setting. The DMA replaces those burdens with a presumption of structural unfairness that invites continuing oversight and intervention.[74]

The Google proceedings illustrate a third dimension: the Commission as quasi-legislative drafter.[75] Through specification proceedings under Article 8(2) DMA, the Commission prescribed, in granular technical detail, the architecture of a data-sharing regime for Google Search: which data must flow, through what mechanism, at what frequency, to whom, subject to which anonymization standards, and on what commercial terms.[76] That requires selecting, from among alternative architectures that might each plausibly comply with the statute, the configuration the Commission prefers.

The obligation to share data on fair, reasonable, and nondiscriminatory terms does not itself dictate whether data must flow in real time through a streaming application programming interface (API) or in periodic batches, how finely it must be disaggregated, or where the anonymization threshold should be set. Each design choice permits multiple reasonable, statute-compliant answers. Google’s own compliance proposal—quarterly releases covering more than a billion queries—diverged sharply from what the Commission required.[77]

The specification proceedings also expanded the recipient class. By extending eligibility to AI chatbots with search functionality, the Commission allocated the data-access advantage across a broader and more heterogeneous set of competitors than the statutory text alone requires.[78] This is a policy choice about which services compete “in substance” with Google Search, made through specification proceedings rather than through the ordinary tools of competition analysis used to assess substitutability, market definition, and competitive pressure.

The point is not that extending the duty to AI chatbots is indefensible on economic grounds.[79] It is that the classification occurred without the usual evidentiary and statutory safeguards, and in a more unilateral manner than Article 102 proceedings would permit. Under Article 102, the relevant product market would be an element the Commission had to define and prove on the evidentiary record. Google could contest that definition and obtain judicial review of the resulting economic finding. Under the DMA, there is no market-definition requirement, and the Commission’s interpretation of who qualifies as a beneficiary is reviewable only for manifest error. Google could make submissions in the specification proceedings, but it could not put the Commission to proof of a relevant market.

D.            Business-User Satisfaction as Public Interest

It is unclear what principles will guide the European Commission’s understanding of fairness and contestability in these and future decisions. When is a fee sufficiently fair? Who should benefit from data-sharing obligations, and on what terms? As the previous section showed, by including AI chatbots among the beneficiaries of Article 6(11), the DMA—and the Commission as its enforcer—picks winners among complementors by forcing Google to share data with a specific and well-resourced class of emerging rivals.[80]

The DMA’s architecture makes it likely, and perhaps inevitable, that the Commission will use business-user satisfaction as a proxy for the public interest. That insight crystallizes the Commission’s ordering power in digital markets. Its function becomes, in effect, Solomonic: deciding who should receive how much of what, on terms the Commission is empowered to set. Disputes that markets once resolved through prices, contracts, and entry are now arbitrated by a single institutional actor wielding standards—“fairness” and “contestability”—whose content it also defines.

Competition law again offers a useful contrast. It rests on a relatively clear distinction between social harm to the competitive process and private harm to individual competitors.[81] The point is captured by the familiar maxim that competition law protects competition, not competitors. Competitors may lodge antitrust complaints and benefit from remedies, but those benefits are contingent on, and secondary to, a finding that competition itself has been harmed and that consumers stand to gain.

Private plaintiffs may therefore benefit from antitrust enforcement, but that benefit is incidental rather than the law’s purpose. Indeed, the conflation of public and private harm is so consequential in antitrust law that some have argued competitor suits should be dismissed at the threshold because they are presumptively self-interested and hostile to the public interest.[82] The U.S. Supreme Court’s antitrust-injury doctrine reflects that concern, requiring that a plaintiff’s claimed injury be “of the type the antitrust laws were intended to prevent.”[83] The European Court of Justice has likewise held that “competition on the merits may, by definition, lead to the departure from the market or the marginalisation of competitors that are less efficient.”[84]

The DMA departs from that logic by blurring the line between private and public costs—and between private and public benefits. It imposes private costs on gatekeepers to transfer benefits to rivals and complementors, prioritizing redistribution over efficiency and over growing the proverbial pie. That sits uneasily with competition law because it conflates two things the consumer-welfare standard tries to separate: harm to competitors and harm to competition, understood in terms of output, innovation, lower prices, or other public-regarding welfare effects.[85]

As President John F. Kennedy observed, “consumers, by definition, include us all.”[86] Kennedy was not referring to antitrust law’s consumer-welfare standard, but the observation captures that standard’s universalizing aspiration. It sets the public interest against the private interests of groups seeking to use law to shield themselves from competitive pressure: laggard competitors, monopolists hoping to live the “quiet life,”[87] and anyone else looking to escape market discipline.

As I have written elsewhere:

Competition cases routinely hinge on the fundamental distinction between conduct that anti-competitively serves to exclude competitors, on the one hand, and competition on the merits that may lead firms to exit the market, on the other… The reason is simple: anticompetitive foreclosure and competition on the merits both ultimately result in the same observable outcome: namely, that rivals exit the market. In order to draw the line, policymakers must infer both the root causes and the effects of firms’ market exit.[88]

In competition law, tools such as the as-efficient-competitor principle draw the line between practices that disadvantage competitors because they are anticompetitive and practices that merely disadvantage competitors while remaining neutral or procompetitive.[89]

“Fairness” follows a different logic. It does not aim to distinguish efficient from inefficient market exit, encouraging the former while suppressing the latter. The DMA’s redistributive objectives render that distinction largely beside the point. The Organization for Economic Cooperation and Development (OECD) has acknowledged that fairness is “strongly tied to redistribution,”[90] and the expert panel commissioned by the Commission described the DMA’s fairness objective in terms of achieving “a higher standard of fairness in the distribution of the social value generated by large platforms.”[91]

That logic sits uneasily with the foundations of competition law. Robert Bork long ago warned against using antitrust to redistribute rents from efficient producers to “pressure-group favourites.”[92] Judge Frank Easterbrook’s observation that competitor-filed suits should be summarily dismissed reflected the same concern. What competition law has tried to avoid—the conversion of a public-policy tool into an instrument of private gain—now sits within the remit of a new form of competition regulation. And under the DMA, that private gain is not treated as contrary to the public interest. It becomes the public interest: competitors should be made better off through regulatory intervention because core platform services are presumed to be structurally unfair.

In hindsight, the DMA’s genesis is indicative of the regime’s goals and ethos. The Coalition for App Fairness, founded in September 2020 by Epic Games, Spotify, Deezer, Basecamp, Proton, and others, launched contemporaneously with the Commission’s legislative proposal and organized explicitly to lobby for the DMA’s passage and aggressive enforcement.[93] Internal Commission correspondence shows that Spotify policy staff reached out directly to the Ursula von der Leyen cabinet in late 2020 to promote the coalition’s agenda.[94] By 2026, the coalition had grown to more than 80 members and had become a standing interlocutor in the Commission’s compliance workshops and specification proceedings. It is unsurprising, then, that business-user satisfaction and the benefits flowing to business users have repeatedly served as proxies for the DMA’s success.[95] The system arguably set itself up for this outcome from the start.

The result is a regime that obscures the boundary between public and private interest. In the DMA’s own terms, the public interest lies in elevating business users—that is, in promoting private interests through public means. Any “asymmetrical” distribution of rights and obligations is treated as a market failure to be corrected. Because distributional asymmetry is baked into the DMA’s structural-unfairness axiom, this effectively removes any limiting principle on the Commission’s redistributive ambition. The line between correcting a market failure and redistributing surplus to well-organized complainants cannot be reliably drawn. The uncomfortable possibility is that it was never meant to be.

The DMA also invites the Commission to make complex redistributive value judgments because business users are not a monolithic class. This creates a second kind of redistribution: not merely between gatekeepers and business users, but among business users themselves. The Commission must therefore choose winners and losers within the very class the DMA purports to protect.

The Apple Core Technology Fee proceedings illustrate the point. Any fee architecture Apple designs will produce different winners and losers among developers. The Core Technology Fee is largely neutral for paid-app developers with small install bases, but potentially prohibitive for high-volume free apps that monetize through advertising.[96] The Google specification proceedings raise a similar problem. By extending data-sharing eligibility to AI chatbots, the Commission spread the data-access advantage to a broader and more heterogeneous set of recipients. Traditional search engines remain entitled to the data under Article 6(11), but the value of that entitlement is diluted when the same data is shared with AI-driven services that are themselves emerging rivals. The intended beneficiaries of the obligation may therefore find that the Commission has simultaneously armed their competitors.

In each case, the Commission is not merely transferring rents from gatekeepers to business users. It is allocating rents among business users with competing interests, divergent business models, and asymmetric stakes. Had the Commission accepted Apple’s Core Technology Fee, it would likely have benefited some developers while burdening others. Had the Commission excluded AI chatbots from Google’s Article 6(11) obligations, it would have strengthened the position of traditional search engines in Europe, and perhaps beyond, against AI-driven alternatives. Whether that would aid, block, or merely delay future technological pathways is unknowable. Each fork in the road favors one business model over another—an implicit bet about where markets are headed, and how quickly. If industrial policy means government intervention targeting specific sectors or firms to shape market structure and allocate competitive opportunities, this looks remarkably close.[97]

The Commission’s enforcement metrics build this dynamic directly into the DMA’s operative framework by treating business-user satisfaction as evidence of effective compliance.[98] The statute thus formalizes dependence on feedback from parties whose structural interest is that compliance never be declared complete. Under that setup, the Commission will rarely hear from the millions of consumers who benefit from an integrated gatekeeper product. It will consistently hear from the organized developers, alternative app stores, and complementors that stand to gain from disintegration.[99]

E.              Consumer Choice as Regulatory Instrument

The DMA gives the European Commission another redistributive function: allocating benefits between business users and consumers. The DMA openly disavows the consumer-welfare standard, yet its provisions plainly seek to change how consumers interact with core platform services. That creates a paradox. The DMA borrows consumer law’s characteristic methods—targeting “unfairness,” correcting perceived imbalances of bargaining power, and mandating changes to product design and choice architecture—while redirecting those tools toward a different set of beneficiaries: business users, including competitors. Consumers are not absent from the regime, but they are instrumental.[100]

The DMA will inevitably redistribute value between business users and consumers. Its overarching objectives of fairness and contestability require benefits, opportunities, and market shares to be distributed more evenly across the digital economy. That project necessarily reshapes the consumer experience in service of a restructuring exercise whose intended beneficiaries are often business users.

There is already evidence of that shift. Mozilla Firefox has documented increased traffic attributable to the DMA: 6 million Firefox selections through browser-choice screens and a 113% increase in daily active users on iOS in the European Union. But that gain came with a tradeoff: users faced more friction during navigation.[101] The Chamber of Progress has catalogued other concrete changes. Sideloading requirements have weakened App Store content moderation; parental controls do not extend to apps installed outside the App Store; and Apple has delayed or withheld the EU rollout of features including iPhone Mirroring, AirPods Pro Live Translation, and its Siri-powered Apple Intelligence capabilities, in each case citing DMA compliance risk.[102] A consumer survey reports a “regulation paradox”: 55% of respondents favor strong rules for digital platforms, yet 39% report needing more steps to complete previously simple tasks, and roughly one-third say their digital experiences are less seamless.[103]

One response is that consumer inconvenience reflects malicious compliance and is therefore not an inevitable consequence of the DMA.[104] Another is that the DMA deliberately accepts a tradeoff: reduced convenience, quality, seamlessness, and possibly price in exchange for a “fairer” distribution of rights, obligations, and rents; less concentrated digital markets; and more competition in the long run.[105]

On that view, weak uptake of rival services—or persistently high uptake of gatekeepers’ own services—can always be read as evidence that enforcement has not gone far enough and that consumers must be pushed further in the desired direction. Yet the DMA lacks a framework for interpreting the alternative possibility: that consumers, fully informed, may simply prefer the gatekeepers’ products. The regime is structurally inclined to meet that outcome with further steering rather than to treat it as evidence that the steering itself may be misdirected.

Consumer choice is therefore not ignored so much as admitted on one condition: that it supports more intervention. If enough users switch away from the gatekeeper, the regime is working and should continue. If they do not, intervention has not yet gone far enough. The one reading the framework cannot easily accommodate is that consumers are simply content with the gatekeeper’s products. That is what it means to say the DMA instrumentalizes consumer choice. Choice is not a verdict to be respected, but a lever to be pulled in a predetermined direction—away from gatekeeper ecosystems and toward rivals and third parties.[106]

 V.                    The Political Economy of Market Ordering

The Commission’s expanded discretion, the business-user-centered conception of compliance, and the reshaping of consumer products are interrelated features of the new regime. They follow the same logical thread. If digital markets are structurally unfair, as the DMA presumes, the Commission must more actively manage how rents are distributed both between groups and within them. The theoretical frameworks introduced in Section II.A each illuminate a different dimension of that shift. Having explained how the DMA works, this section brings those frameworks back to the fore to explain the DMA’s broader political-economy role.

Karl Polanyi’s double-movement thesis supplies the broadest lens. The DMA can be understood as a re-embedding intervention: an attempt to subordinate digital market outcomes to a normative order after a period of perceived platform autonomy. But there is an important twist. “Society,” in this context, means something idiosyncratic. The countermovement is not into Polanyi’s “society” in any diffuse sense. It is into the administrative state. The Commission—not communities, workers, or consumers—reclaims authority over digital markets. The normative order into which those markets are re-embedded is the Commission’s own account of what is “fair” and “contestable,” concepts whose content the Commission defines and whose practical meaning it develops through enforcement.

Herman Schwartz’s framework makes this legible at a historical level. What Polanyi describes as societal re-embedding becomes, in the DMA’s case, state reassertion: a phase in the cyclical relationship between states and markets in which governments reclaim authority to define market outcomes after a period of perceived market dominance.[107] Ordoliberalism completes the picture by explaining the mode of governance. The Commission, cast as nobilitas naturalis, decides when the economy should be governed by indirect steering through competition oversight and when it requires direct ordering through prescriptive rules and enforcement.

The Commission’s choice of which firms to control more directly situates the DMA in its broader political context and confirms the role of political goals in moving the regime from oversight to ordering. That shift replaces the relatively agnostic correction of market failures with a more openly political, redistributive, and moralized allocation of costs and benefits from the center. It also enables the pursuit of broader strategic goals. Competition law was too constrained an instrument for that project. The DMA provides tools competition law could not.

French President Emmanuel Macron framed the adaptation of competition law, which he regarded as too focused on consumer welfare, as a precondition for “technological sovereignty.”[108] Pursuing that goal required either a new, competition-law-like instrument that was not quite competition law, at least not in the traditional consumer-welfare sense, or the adaptation of existing competition law itself. Europe appears to have gotten both.

Indeed, the DMA has coincided with a parallel intensification of competition enforcement—a regulatory ratchet effect. Rather than substituting for competition law or merely complementing it, the DMA has coincided with its expansion, suggesting a broader shift in governance. The Commission’s 2024 Draft Guidelines on Article 102 TFEU, for example, seek to revive a more form-based approach to abuse of dominance, reversing aspects of the More Economic Approach.[109] The Commission also invoked Article 22 of the EU Merger Regulation in Illumina/Grail to claim jurisdiction over a transaction that met neither EU nor national merger-control thresholds, a maneuver the European Court of Justice ultimately rejected.[110] The April 2026 Draft Merger Guidelines fold treaty-level policy objectives into competitive assessment and acknowledge that their weight is left to the Commission’s discretion.[111] In Android Auto, the Court extended dominant firms’ duty to deal in a manner more consistent with Article 6(7) DMA than with established essential-facilities doctrine.[112] Each instrument expands the Commission’s discretion along a different axis. Their cumulative effect exceeds what any single reform could have achieved alone.

Scholars have begun to identify the same broader trend. Ioannis Lianos describes the conversion of competition law from a “politically neutral technocratic exercise into an instrument of economic statecraft.”[113] That account aligns with this paper’s reading of ordoliberalism and Schwartz: the ordoliberal nobilitas naturalis always claimed authority to determine when ordering was required, and Schwartz’s cycles describe precisely this kind of state reassertion. Ariel Ezrachi and Maciej Bernatt likewise document the rising politicization of competition enforcement across jurisdictions, describing an instrumentalization in which enforcement decisions increasingly reflect short-term political objectives rather than independent assessments of competitive harm.[114]  That account aligns with this paper’s Polanyian claim: re-embedding occurs through the state, not society. Elias Deutscher goes further, treating the DMA as “a political bargain that rebalances the relations between the haves and have-nots of the digital economy.”[115] That formulation dovetails with all three theoretical pillars of this paper.

There is growing evidence that this reorientation extends beyond digital markets, although the picture is more dialectical than linear. Japan’s “New Form of Capitalism” program placed digital platform regulation alongside wage redistribution and green transformation under the broader project of “upgrading capitalism.”[116] Industrial policy more generally has been rehabilitated after decades outside polite economic conversation.[117] At the same time, the neo-Brandeisian agenda in the United States has largely stalled;[118] the United Kingdom dismissed the chair of the Competition and Markets Authority for insufficient support of its growth agenda;[119] the Draghi Report warned of an “existential” competitiveness crisis;[120] and several jurisdictions that initially embraced DMA-like frameworks have pulled back.[121]

Those countercurrents do not refute the shift described in this paper. They do suggest that the shift coexists with significant resistance. Whether that tension becomes a permanent feature of political economy, or resolves in one direction or the other, remains to be seen. What is clear is where the DMA fits in the story. It marks a relocation along the oversight-ordering spectrum decisive enough that competition in affected markets is no longer a process the Commission polices from the outside, but an outcome it sets out to design. That the change is, for now, concentrated in digital markets makes it no less paradigmatic. Digital markets may be where the template is being built, but templates travel.

 VI.                    The Costs of Market Ordering

The preceding sections described what the DMA does and where it sits. This section asks what it costs. The central claim is that those costs are not incidental defects of drafting or enforcement that better administration can cure. They are entailments of the ordering model itself. Each flows from the features that distinguish ordering from oversight: the removal of effects analysis, efficiency defenses, and merits review, and their replacement with indeterminate standards the Commission largely interprets for itself. The constraints whose absence makes the regime ordering rather than oversight are the same constraints whose absence generates its pathologies.

Those costs are visible but unseen. They are visible because they are predictable from the DMA’s design: a regime built this way is prone to regulatory error, limited by the knowledge problem, poorly equipped to learn from failure, vulnerable to capture, and inclined to treat political discretion as legal administration. They are unseen because the regime is structurally disposed not to perceive them. It lacks the procedural moments, evidentiary obligations, external benchmarks, and public-interest constraints through which such costs would ordinarily come into view. If templates travel, the DMA’s ordering model may migrate to other initiatives that substitute direct market design for oversight. The critique that follows therefore takes the DMA as its anchor, but applies in its basic form to any regime built on the same ordering logic.

A.             Error Costs Without Error Correction

The most immediate cost of the ordering model, in competition regulation and beyond, is the risk of regulatory error in markets the regulator does not fully understand. Competition law mitigates that risk through a judicialized system of procedural safeguards: effects analysis, efficiency defenses, and judicial review on the merits. Each safeguard serves the function Judge Frank Easterbrook identified as the essential purpose of procedural design in antitrust: minimizing the sum of the costs of anticompetitive practices that escape condemnation, competitive practices that are condemned or deterred, and the system’s own administrative overhead.[122] The DMA deliberately sets those safeguards aside.

As discussed in Section III, much of the conduct the DMA prohibits has a plausible procompetitive rationale. Vertical integration, self-preferencing, bundling, and data combination are strategies whose welfare effects depend on context. Prohibiting them without effects analysis creates Type I errors that the consumer-welfare standard was designed to filter out.[123] Courts moved away from per se rules for novel conduct precisely because innovative markets heighten the risk of false positives.[124] The DMA reinstates a per se approach for the very category of conduct that the error-cost framework counsels against treating categorically.

The result is an enforcement architecture without a meaningful self-correcting mechanism. A flawed regulatory premise can persist indefinitely because the institutional channels through which it might be challenged have been deliberately narrowed. False negatives can sometimes be corrected: if a monopolist is wrongly acquitted, entry may remain possible, demand may shift, and technology may evolve. False positives are harder to unwind: the condemned behavior is abandoned, the product design is changed, and the forgone innovation may never be observed. The DMA’s architecture increases the risk of the latter while narrowing the mechanisms that might reveal or correct the error.

The nirvana fallacy is implicit throughout. The DMA compares real, imperfect digital markets with an idealized post-regulatory counterfactual in which the Commission’s interventions produce the outcomes it anticipates.[125] That counterfactual is not merely uncertain; it is unobservable. There is no control group, no parallel market in which the same gatekeeper operates free of intervention, and each core platform service is idiosyncratic enough that no other market supplies a clean benchmark.

Nor is there any guarantee that the Commission’s vision of a “contestable” app-distribution market or a “fair” advertising ecosystem will prove more efficient, more innovative, or more consumer-friendly than the arrangements it displaces. Critically, it is unclear whether the DMA contains a mechanism for discovering that it has not. That problem leads to the next concern.

B.             The Knowledge Problem of Market Ordering

The error-cost problem is compounded by an epistemic one. Market oversight, in its traditional competition-law form, is epistemically modest. It asks, after the fact, whether specific conduct caused demonstrable harm. Market ordering requires much more. It requires the regulator to know in advance what the optimal market structure should look like, how rents should be distributed between gatekeepers and business users, what fee levels are “fair,” which product-design choices should be permitted, and what degree of vertical integration is appropriate.

Friedrich Hayek’s knowledge problem helps clarify the difficulty. Hayek argued that the information needed to coordinate economic activity is dispersed among millions of individuals, embedded in local knowledge that no central authority can fully aggregate, and revealed only through the competitive process. Competition is therefore a discovery procedure, not merely an allocation mechanism—or taxonomy, as Section IV.A described the DMA’s approach.[126]

If Hayek was right, the very features of digital markets that seem to invite intervention—complexity, rapid change, and asymmetric information—are also the features that make central direction least tractable. The DMA does not engage this objection. It assumes the Commission can know, in advance and in the abstract, what a “fair” fee looks like, which data-sharing architecture is optimal, and which product-design choices should be permitted.

The enforcement record already illustrates the epistemic difficulty. In the Meta proceeding, the Commission concluded that consent or pay was insufficiently “fair,” but did not specify what alternative model would satisfy the DMA’s requirements, leaving Meta to iterate through compliance proposals against an indeterminate standard. In the Apple scare-screen investigation, the Commission judged Apple’s user-facing warnings not neutral and objective, but the question of how much friction a warning may generate before it becomes impermissible steering has no determinate answer, and the DMA supplies no metric for drawing the line.

One possibility is that the Commission will learn by doing and eventually supply comprehensive answers to these questions. Another more plausible possibility, given the dynamism of the markets at stake, is that the epistemic gap is not a transitional feature that will narrow as the Commission gains experience, but a permanent feature of the ordering function itself. If so, the implication is not merely that the Commission will make mistakes. It is that the regime cannot be designed out of making them.

The familiar rejoinder—that the regulator will calibrate over time—assumes the missing information exists to be acquired. On the Hayekian account, it does not. The relevant information is generated through the competitive process itself, which the intervention displaces by substituting ordering for oversight.

C.             Autopoietic Regulation

Sections VI.A and VI.B lay the foundation for a critique of autopoietic regulation. At a structural level, the DMA’s institutional architecture is poorly equipped to generate the conclusion that an intervention was wrong. Its preamble presents the obligations as simultaneously beneficial to end users, business users, and innovation.[127] There are no tradeoffs except gatekeepers’ losses, which are treated as the cost of doing business.

No recital seriously contemplates that mandated interoperability might degrade service quality, that prescribed fee structures might distort investment, or that redistributing rents from platforms to business users might harm consumers or other groups of business users. The absence of an efficiency defense means there is no procedural moment at which a gatekeeper can force those costs onto the table. Judicial review confined to manifest error means there is no institutional actor positioned to ask whether an intervention’s costs exceed its benefits. In this way, the DMA does not answer the error-cost question. It largely avoids it.

The Commission’s first review of the DMA under Article 53 confirms the pattern. The review declares that the DMA has been “working well overall” and catalogs enforcement outputs as evidence of success.[128] When stakeholders report concrete negative effects—for example, trade associations documenting 40% to 50% losses in organic traffic for retailers following Alphabet’s compliance measures—the review records those concerns as consultation inputs rather than findings demanding analysis. The possibility that such outcomes might reflect a structural defect in the regulation, rather than transitional friction, lies outside the evaluative frame.[129] The Commission’s assessment methodology was not designed to return a negative verdict on the DMA’s effectiveness or on its underlying axiomatic premises.[130]

The result is a system whose assumptions are reproduced rather than tested. Unfairness is structural. Gatekeepers are always beneficiaries of a lopsided distribution. Every change the DMA requires is presumed achievable. Noncompliance is most readily explained by resistance or bad faith. A regime constituted on such premises is not self-evidently “working” any more than a bus headed toward a cliff is “working” because it is moving. Movement is not progress.

Such regimes cannot readily reform themselves because they are impervious to feedback that falls outside the categories their own theory admits. And those categories are, in turn, structured to be self-confirming. That is what makes the DMA’s evaluative architecture autopoietic.

This dynamic has an endpoint that the paper’s broader argument anticipates. Franz Böhm’s concession that direct ordering is appropriate when politically warranted, and Böhm’s own political context, should give pause to those who assume that a norm-following, technically sophisticated regulator will always know where oversight ends and ordering begins. The answer does not depend on the regulator’s caliber or on the moral fiber of those who comprise it. It depends on the institutional constraints within which the regulator operates, and on whether those constraints are subject to meaningful external review.

The DMA has deliberately weakened those constraints, but at a cost to which it is structurally myopic. The question its evaluative architecture is constitutionally ill-positioned to ask—whether the costs of ordering exceed its benefits—falls by default to others: courts applying manifest-error review, legislators governing through instruments they may not fully understand, and, ultimately, consumers who use products the Commission has redesigned on their behalf and whose agency is instrumentalized toward preferred redistributive goals and market structures. Under those conditions, correction seems unlikely.

D.            Why Consumer Welfare Still Constrains

A common rejoinder is that the DMA does not uniquely empower the Commission to make political choices because the consumer-welfare standard is itself political. In that sense, the argument goes, the DMA is not fundamentally different from competition law: both rest on contestable normative premises.[131]

The observation is correct as far as it goes. The consumer-welfare standard does embed a qualitative political choice. It holds that consumers should not pay higher prices or accept lower quality so that less efficient competitors can be shielded from competition. But it is political in the way broad, settled democratic commitments are political. It commands wide assent across jurisdictions, political cycles, and doctrinal traditions,[132] because the alternative—asking consumers to bear the costs of protecting specific businesses—commands little support when stated plainly.

One cannot simultaneously maximize efficiency and protect all competitors, because some competitors are inefficient. Protecting them means condemning efficient conduct precisely because it harms them by being more competitive: lower prices, better products, greater convenience, or more successful innovation.[133] That tradeoff is widely recognized in competition law and has divided commentators for decades: to what extent should competition law protect less efficient competitors for the sake of plurality, dispersion, or some other objective?[134]

The DMA appears to have made its choice. Some economic efficiency and consumer welfare may be sacrificed for “fairer” and less concentrated digital markets. If that is the policy choice, so be it. But it should be made openly, rather than obscured in the universal language of “fairness,” whose universality is precisely what is in dispute.

The claim that all standards are equally legitimate because all are political also fails for a second reason: administrability. Consumer welfare converts a general normative commitment into a testable question: Does the conduct enable higher prices, reduced output, lower quality, or diminished innovation? Courts, litigants, and experts can organize their arguments around that basic inquiry.[135]

A polycentric standard that asks the enforcer to weigh fairness, contestability, innovation, and distributional equity without a meta-criterion for ranking those values is political in a different and more unbounded way.[136] The values in play are often incommensurable. Their reconciliation cannot be achieved through legal reasoning alone. It must be achieved through discretion, which more readily lends itself to pressure from self-interested rent-seekers.

A society may legitimately prefer to benefit organized business users over consumers. But that choice should be made explicit and subjected to democratic deliberation for what it is. The DMA never clearly acknowledged those tradeoffs, perhaps because acknowledging them would have threatened the regime’s political appeal or because its autopoietic premises made them difficult to see. As a result, the question was not taken seriously at the legislative stage. The Commission now makes these distributional choices not as a legislator accountable to voters, but as an enforcer subject only to manifest-error review.

There is, finally, a structural feature of the consumer-welfare standard that its critics tend to undervalue: it constrains capture. Because it requires the enforcer to demonstrate harm to consumers—a diffuse and numerous class whose interests cannot easily be appropriated by any single organized group—it disciplines enforcement by requiring justification by reference to beneficiaries broader than the complainants.

The DMA removes that constraint. Its standards are defined without reference to consumer welfare, and enforcement is driven substantially by business users: a concentrated, well-organized class with specific commercial grievances. A standard that constrains enforcement to benefit a broad class is not equivalent to a standard that permits enforcement to benefit a narrow one, even if both are, in some ultimate sense, political.

E.              Capture in the Ordering State

Finally, as signaled above, the shift from oversight to ordering transforms the political economy of enforcement. When the state polices competitive processes, the returns to lobbying are relatively contained. Firms may seek to influence case outcomes, but the analytical framework constrains what the enforcer can deliver. When the state defines market outcomes—setting fees, mandating access, redistributing rents among business users—the stakes of each decision rise accordingly, and the institutional architecture becomes vulnerable to the capture dynamics public-choice theory would predict.[137]

As Section IV.D explained, the structural-unfairness axiom treats asymmetry between gatekeepers and business users as prima facie evidence of a defect requiring correction, leaving the Commission without a limiting principle on its redistributive ambition. The resulting capture risk is not the crude one of an enforcer serving a single interest group. It is, instead, subtler and harder to police. Because the axiom equates business-user grievance with public-interest harm, private rent-seeking claims and public-interest claims become structurally indistinguishable at the analytical level. A complainant that can frame its commercial grievance as structural unfairness has, by definition, stated a public-interest case the Commission is institutionally predisposed to credit. The problem, then, is not merely the familiar collective-action story in which well-organized private interests overtake a diffuse public interest. It is that the two are effectively collapsed into one.[138]

The European Union’s own institutional history confirms the risk. Before 2004, EU merger control suffered from a structurally similar concentration of power. Damien Neven and Lars-Hendrik Röller observed that individual officials within the Merger Task Force could “pursue their own objectives at the expense of those assigned by the regulation,” with those objectives “in turn manipulated by third parties, including competitors and member governments.”[139] Commissioner Mario Monti acknowledged the risk directly.[140] The 2004 reforms responded by introducing peer-review panels, a chief economist as an independent analytical check, and an effects-based approach requiring the Commission to demonstrate likely competitive harm.

The DMA reverses that institutional logic. It dispenses with peer review, has no comparable chief economist function, confines judicial review to manifest error, and removes the burden of proving competitive harm.

That is the throughline of the costs canvassed here. They are not separate complaints but one institutional critique. The ordering model deliberately removes the mechanisms through which competition law learned to discipline itself: the effects test that filters error, the evidentiary record that exposes the knowledge problem, the external benchmark that allows an intervention to be judged a failure, and the demonstrable-harm requirement that keeps enforcement tethered to a broad public rather than a narrow set of complainants.

What remains is discretion that cannot easily be shown to have erred, cannot reliably learn that it has, and cannot readily be disciplined when captured. The European Union spent the decade after 2004 building safeguards that made its competition enforcement credible. The DMA, in the name of fairness, sets them aside—and does so where the stakes, and the firms, are largest.

The rationale for intervention therefore cuts both ways. If digital markets are too important to be left to the whims of the market, they are also too important to be left to a myopic helmsman steering in the fog—reordering markets under a standard that cannot reliably distinguish successful intervention from failure, in conditions too consequential, fast-moving, and hard to reverse to absorb the error.

VII.                    Conclusion

The DMA marks a decisive relocation along the oversight-ordering spectrum. It is not a categorical break from EU competition law’s tradition, but it does systematically dismantle the constraints that disciplined the Commission’s prior position on that spectrum. Effects analysis, efficiency defenses, the burden of proving harm, and meaningful judicial review have each been removed. What remains is an enforcer with expanded authority to prescribe competitive outcomes, set prices, dictate product design, and redistribute rents among business users, governed by standards—“fairness” and “contestability”—that supply no external benchmark for knowing when the exercise is complete.

The enforcement record against Meta, Apple, and Google has already shown that this authority is being exercised across the dimensions the framework predicts: setting the terms of a two-sided platform’s business model, prescribing the permissible language of a notification screen, and specifying the technical architecture of a data-sharing regime. Whether these interventions will, in the long run, produce more innovative or more consumer-friendly digital markets remains uncertain. What is not uncertain is that the DMA’s institutional architecture—per se prohibitions, indeterminate standards, thin judicial review, and reliance on business-user feedback—is structurally disposed to reproduce the regulator’s assumptions rather than test them, and to interpret any residual asymmetry as confirming that further intervention is warranted. This self-induced myopia may be an inevitable feature of the state-led mode of governance the law embodies, with implications extending well beyond digital markets.

A standard defense of the DMA is that digital markets are different and therefore require special rules. The intuition is widely shared, but the inference does not follow. If the benchmark is consumer welfare and productive efficiency, it is hard to identify a category of markets that has served consumers as well as digital markets have: prices at or near zero, rapid quality improvement, persistent entry at the application layer, and a rate of innovation few regulated sectors approximate. If the benchmark is something else—fairness in the distribution of platform rents, preservation of a particular market structure, or protection of specific classes of business users—then competition law is indeed inadequate to the task. But it is inadequate in the same way for every market. Shift the goal from consumer welfare to structural redistribution, and the toolkit of effects analysis, efficiency defenses, and judicially policed standards of harm will look insufficient wherever it is applied. Put differently, the diagnosis of inadequacy is not a finding about digital markets but an artifact of the new political economy of digital competition.

On that view, digital markets were less the natural occasion for the DMA than its convenient political target. This was a sector in which concentration was salient, the firms were foreign and politically exposed, and the case for special treatment was rhetorically easy to make. The mode of governance the DMA inaugurates is, however, portable. Similar logic could readily extend to bank cards, consoles, consultancies, supermarkets, and other sectors that exhibit some version of the conduct the DMA targets: access fees, self-preferencing, tying, data-driven advantages, and hardware- or software-driven user lock-in.

Signs of migration are already visible in adjacent EU instruments. The Draft Article 102 Guidelines reorient abuse-of-dominance enforcement around structural and distributive concerns. The Draft EUMR Guidelines incorporate non-competition objectives into merger control. The broader revival of industrial policy across the European Union supplies a hospitable political environment for ordering instruments more generally. In this respect, the DMA is of a piece with a wider international pattern that other commentators have identified: the instrumentalization of “fairness” as rhetorical cover for more discretionary, state-directed market design.[141]

The DMA, then, is not a flash in the pan. Whether the constraints that competition law painstakingly developed to discipline its own ordering moments can be reconstructed inside a regime designed to dispense with them is the question this article leaves open—and the question on which the legitimacy of the next decade of European market governance may turn.

[1] Regulation (EU) 2022/1925 of the European Parliament and of the Council of 14 Sept. 2022 on Contestable and Fair Markets in the Digital Sector and Amending Directives (EU) 2019/1937 and (EU) 2020/1828 (Digital Markets Act), 2022 O.J. (L 265) 1 [hereinafter DMA].

[2] DMA, supra note 1, recitals 2–5. Recital 3 states that the combination of gatekeeper characteristics “is likely to lead, in many cases, to serious imbalances in bargaining power and, consequently, to unfair practices and conditions for business users.”

[3] Eur. Parl. Rsch. Serv., Digital Sovereignty for Europe (EPRS Briefing PE 651.992, July 2020). European Commission President Ursula von der Leyen defined digital sovereignty as “the capability that Europe must have to make its own choices, based on its own values, respecting its own rules.” Ursula von der Leyen, State of the Union Address (Sept. 16, 2020).

[4] Consolidated Version of the Treaty on the Functioning of the European Union art. 102, 2012 O.J. (C 326) 47 [hereinafter TFEU].

[5] On the dual role of public authority in Franz Böhm’s thought—both diagnosing the dangers of concentrated power and assigning the state a discretionary role in managing it—see Franz Böhm, quoted in Thomas Biebricher & Frieder Vogelmann eds., The Birth of Austerity: German Ordoliberalism and the Eurozone Crisis 117, 119, 133 (Rowman & Littlefield Int’l 2017). On ordoliberalism as a foundational influence on European Union competition law, see David J. Gerber, Protecting Prometheus: Law and Competition Policy in Europe (Oxford Univ. Press 2000).

[6] Herman Schwartz, States Versus Markets: The Emergence of a Global Economy ch. 1 (3d ed. 2010).

[7] Karl Polanyi, The Great Transformation: The Political and Economic Origins of Our Time 136–50 (1944). Polanyi develops the “double movement” thesis principally in chapters 11–12.

[8] For the argument that general competition law occupies the oversight end of the spectrum—relying on case-by-case adjudication, rule-of-reason analysis, and an error-cost framework that prioritizes false negatives over false positives, rather than direct economic steering or industrial policy—see Geoffrey A. Manne & Lazar Radic, Competition and Competition Law in the Classical Liberal Tradition, in The Routledge Handbook of Classical Liberalism ch. 17, 258, 264–66 (Richard A. Epstein, Mario J. Rizzo & Liya Palagashvilli, eds., 2026) (describing common-law nonenforcement of restraints of trade as the classical-liberal antitrust framework), 273–74 (describing per se rules as the product of repeated rule-of-reason determinations).

[9] A familiar example is postwar France’s Commissariat général du Plan, established in 1946 to coordinate investment across strategic sectors through successive multiyear plans. See Philippe Bauchet, La planification française (1962). For a comparative treatment of similar institutions in countries such as Italy and Japan, see Andrew Shonfield, Modern Capitalism: The Changing Balance of Public and Private Power (1965).

[10] Classic examples include consumer-protection measures such as cooling-off periods and mandatory disclosures, which encourage more deliberative purchasing decisions; default enrollment in pension-savings plans, which increases savings rates; plain-packaging requirements and graphic health warnings for tobacco products, which discourage harmful consumption; and food-labeling systems such as the European Union’s Nutri-Score, which encourage healthier diets. For the foundational theory, see Richard H. Thaler & Cass R. Sunstein, Nudge: Improving Decisions About Health, Wealth, and Happiness (2008). For the distinction between economic and social regulation, see Christopher Decker, Modern Economic Regulation: An Introduction to Theory and Practice chs. 1–2 (2015).

[11] For the argument that the Digital Markets Act transforms competition law into competition regulation, see Geoffrey A. Manne, Lazar Radic & Dirk Auer, Regulate for What? A Closer Look at the Rationale and Goals of Digital Competition Regulations, 22 Berkeley Bus. L.J. (2025).

[12] On the conventional understanding of ordoliberalism as a framework of economic governance centered on the “economic constitution,” rather than discretionary intervention, see David J. Gerber, Protecting Prometheus: Law and Competition Policy in Europe ch. 7 (2000) (describing ordoliberalism as the intellectual foundation of postwar European competition law and its commitment to preserving competitive order rather than directing market outcomes); Giorgio Monti, EC Competition Law 21–28 (2007); Oles Andriychuk, The Normative Foundations of European Competition Law chs. 2–3 (2017) (arguing that the ordoliberal antitrust enforcer serves as a guardian of the competitive process); see also Werner Bonefeld, Freedom and the Strong State: On German Ordoliberalism, 17 New Pol. Econ. 633 (2012).

[13] Franz Böhm, Die Ordnung der Wirtschaft als geschichtliche Aufgabe und rechtsschöpferische Leistung (1937); David J. Gerber, Constitutionalizing the Economy: German Neo-liberalism, Competition Law and the “New” Europe, 42 Am. J. Comp. L. 25, 50 (1994).

[14] Böhm, quoted in Biebricher & Vogelmann, supra note 5, at 117. The “steering norm” formulation originates in Böhm’s writings of the 1930s, published during the Third Reich. Böhm warned that public authority over competition could be captured by authoritarian movements while also assigning that authority a discretionary role that could, in principle, serve either authoritarian or liberal political orders.

[15] Id. at 119 (“direct market steering by methodical means of command”).

[16] The contrast with more dirigiste and authoritarian currents in interwar Germany is instructive. Walter Eucken’s contemporaries within and adjacent to the ordoliberal tradition—including Alfred Müller-Armack and, in a different register, Carl Schmitt—were considerably more willing to subordinate competition to political objectives. Compare Carl Schmitt, Starker Staat und gesunde Wirtschaft (1932) (advocating an authoritarian executive empowered to override market processes in the national interest), with Alexander Rüstow, Free Economy, Strong State (address delivered at the Verein für Sozialpolitik, Dresden, Sept. 28, 1932) (employing similar rhetoric in support of a markedly different program). On the spectrum of interwar German economic thought—from laissez-faire through ordoliberalism to corporatist and National Socialist conceptions—see Volker R. Berghahn, The Americanisation of West German Industry 1945–1973 ch. 2 (1986); Werner Bonefeld, Freedom and the Strong State: On German Ordoliberalism, 17 New Pol. Econ. 633, 638–42 (2012). Ordoliberalism’s distinctive contribution was to pair a strong state with a rules-based, rather than discretion-based, economic constitution.

[17] Case C-377/20, Servizio Elettrico Nazionale SpA v. Autorità Garante della Concorrenza e del Mercato, EU:C:2022:379, ¶ 65 (“it is in no way the purpose of Article 102 TFEU to prevent an undertaking from acquiring, on its own merits, a dominant position on a market”); Case C-680/20, Unilever Italia SpA v. Autorità Garante della Concorrenza e del Mercato, EU:C:2023:33, ¶ 38.

[18] The “special responsibility” doctrine originates in Case 322/81, NV Nederlandsche Banden-Industrie Michelin v. Commission, 1983 E.C.R. 3461, ¶ 57.

[19] See Communication from the Commission, Draft Guidelines on the Application of Article 102 of the Treaty on the Functioning of the European Union to Abusive Exclusionary Conduct by Dominant Undertakings (Aug. 2024) (establishing structured presumptions for certain categories of conduct, including exclusivity arrangements, predatory pricing, and some forms of margin squeeze). For a critique that the Draft Guidelines move effects-based analysis toward quasi-per se treatment for selected practices, see Lazar Radic & Dirk Auer, The Commission’s Art. 102 TFEU Guidelines: Consolidation or Creation?, Eur. Competition J. (2025). Radic and Auer argue that, although the Guidelines purport to codify the Court of Justice’s case law, they selectively import presumptions of illegality while downplaying effects analysis.

[20] Joined Cases C-501/06 P, C-513/06 P, C-515/06 P & C-519/06 P, GlaxoSmithKline Services Unlimited v. Commission, 2009 E.C.R. I-9291, ¶ 63 (“the objective assigned to Article 101 TFEU … is to prevent agreements which restrict competition and, thereby, harm consumers”); Case C-413/14 P, Intel Corp. v. European Commission, EU:C:2017:632, ¶ 133. In Case C-240/22 P, Intel Corp. v. Commission (Intel II), EU:C:2024:915, ¶ 328, the Court reaffirmed that “it is for the Commission to prove the infringements of the competition rules which it has found and to adduce evidence capable of demonstrating to the requisite legal standard the existence of the constituent elements of an infringement.”

[21] Case C-52/09, TeliaSonera Sverige AB v. Konkurrensverket, EU:C:2011:83, ¶ 22; Case C-209/10, Post Danmark A/S v. Konkurrencerådet, EU:C:2012:172, ¶ 22; Case C-413/14 P, Intel Corp. v. European Commission, EU:C:2017:632, ¶ 138 (holding that the assessment of exclusionary conduct must be “based on all the circumstances of the case”).

[22] Case C-48/22 P, Google LLC & Alphabet Inc. v. Commission (Google Shopping), EU:C:2024:726, ¶¶ 263–67. The Court held that the Commission must assess whether conduct is capable of excluding an as-efficient competitor when the dominant undertaking submits supporting evidence that its conduct could not restrict competition. Id. ¶ 265. See also Case C-413/14 P, Intel Corp. v. European Commission, EU:C:2017:632, ¶¶ 138–41; Case C-680/20, Unilever Italia SpA v. Autorità Garante della Concorrenza e del Mercato, EU:C:2023:33, ¶¶ 56–60.

[23] Id.

[24] The logic of Article 102 TFEU, and of European Union competition law more generally, is articulated with unusual clarity in Advocate General Nils Wahl’s Opinion in Case C-413/14 P, Intel Corp. v. Commission, EU:C:2016:788, ¶¶ 41–42: “[P]rotection under EU competition rules is afforded to the competitive process as such, and not, for example, to competitors … competitors that are forced to exit the market due to fierce competition, rather than anticompetitive behaviour, are not protected … competition law aims, in the final analysis, to enhance efficiency … dominance as such is not considered to be at variance with Article 102 TFEU. Rather, only behaviour which constitutes an expression of market power to the detriment of competition and, thus, to consumers is prohibited.”

[25] Case C-67/13 P, Groupement des cartes bancaires (CB) v. European Commission, EU:C:2014:2204, ¶ 49.

[26] Case C-828/18, Booking.com BV v. Autorità Garante della Concorrenza e del Mercato, EU:C:2020:1028, ¶ 47.

[27] Bundesgerichtshof [BGH] [Federal Court of Justice] May 18, 2021, KVR 54/20 (Booking.com), ¶ 67 (Ger.).

[28] Commission Regulation (EU) 2022/720 of May 10, 2022, on the Application of Article 101(3) of the Treaty on the Functioning of the European Union to Categories of Vertical Agreements and Concerted Practices, 2022 O.J. (L 134) 4, art. 6. To qualify for the Vertical Block Exemption Regulation, the supplier’s and buyer’s market shares must each not exceed 30%.

[29] Eur. Comm’n Decision of May 4, 2017, Case AT.40153—E-book MFNs and Related Matters (Amazon MFNs), ¶¶ 44, 70, 73, 77 (examining market position, barriers to entry, multihoming, and competitive dynamics, while recognizing legitimate business justifications for parity clauses, including preventing free-riding on platform investments). The Vertical Guidelines likewise note that “the likely level of free riding must be sufficient to significantly impact the incentives to invest in the online intermediation services,” confirming that firms may still invoke procompetitive justifications. Commission Notice, Guidelines on Vertical Restraints, 2022 O.J. (C 248) 1, ¶ 373.

[30] Herbert Hovenkamp, Federal Antitrust Policy: The Law of Competition and Its Practice 213 (5th ed. 2016); see also Herbert Hovenkamp, The Rule of Reason, 70 Fla. L. Rev. 81, 159–60 (2018) (arguing that most-favored-nation clauses raise competitive concerns primarily when coupled with product-specific market power).

[31] For quasi-experimental evidence on the effects of banning price-parity clauses, see Jack (Peiyao) Ma, Andrea Mantovani, Carlo Reggiani, Annette Broocks & Néstor Duch-Brown, The Price Effects of Prohibiting Price Parity Clauses: Evidence from Global Hotel Chains, Econ. J. (2026), https://doi.org/10.1093/ej/ueag059. Analyzing France’s 2015 Macron Law, the authors find limited and statistically insignificant price effects in consumer-visible channels—hotel websites and online travel agencies—but significant price reductions in direct offline bookings, accompanied by a shift in bookings from online travel agencies to hotels’ direct offline channels. The findings suggest that the welfare effects of price-parity-clause prohibitions are channel-specific and not uniformly beneficial to consumers in the channels where most bookings occur. See also Matthias Hunold, Reinhold Kesler, Ulrich Laitenberger & Frank Schlütter, Evaluation of Best Price Clauses in Online Hotel Bookings, 61 Int’l J. Indus. Org. 542 (2018) (finding, across nearly 30,000 hotels operating under different regulatory regimes, that online travel agency commission rates did not decline following the prohibition of best-price clauses, contrary to a principal rationale offered for such bans).

[32] Case C-48/22 P, Google LLC & Alphabet Inc. v. Commission (Google Shopping), EU:C:2024:726, ¶ 170.

[33] Id. ¶¶ 168–75. The opinion’s vocabulary—“capability of foreclosing,” “as-efficient competitor,” “counterfactual scenario,” “causal link between abuse and effects,” and “burden of proof”—reflects an oversight-oriented approach to competition law. See also id. ¶¶ 218–24 (emphasizing the need to establish a causal link between the alleged abuse and its effects, a requirement absent from the Digital Markets Act’s per se prohibitions).

[34] Case T-612/17, Google Android, EU:T:2022:541, ¶¶ 427, 443, 470 (analyzing default-search-engine agreements in light of market shares, barriers to entry, switching costs, alternative options available to original equipment manufacturers, and the likely counterfactual absent the challenged restrictions).

[35] For the argument that distinguishing procompetitive from anticompetitive foreclosure under Article 102 TFEU is difficult because both can generate similar observable outcomes, see Dirk Auer & Lazar Radic, The Growing Legacy of Intel, 14(1) J. Eur. Competition L. & Prac. 15 (2023).

[36] Businesses across a wide range of industries—from restaurants and consulting firms to supermarkets—engage in some forms of self-preferencing.

[37] See Giuseppe Colangelo, Antitrust Unchained: The EU’s Case Against Self-Preferencing, 72 GRUR Int’l 538 (2023) (arguing that treating self-preferencing as per se anticompetitive allows enforcers to bypass the legal standards and evidentiary burdens ordinarily required to establish anticompetitive conduct).

[38] On the conceptual ambiguity of “self-preferencing” and the wide range of practices it encompasses, see Pablo Ibáñez Colomo, Self-Preferencing: Yet Another Epithet in Need of Limiting Principles, 43 World Competition 417, 425 (2020) (“the various manifestations of the phenomenon are far from identical; some raise issues similar to traditional tying cases, and others raise issues similar to a refusal to deal”).

[39] Id. at 417, 425.

[40] On the recognized efficiencies of tying and bundling in digital markets, and the case against treating such practices as per se unlawful, see Aditya Sushant Jain, Against Per Se Illegality of Tying and Bundling in Digital Markets: Perspectives from the EU and India, 71(1) Antitrust Bull. 96 (2026), especially § IV (surveying efficiencies associated with tying and bundling, including cost reductions, quality improvements, lower search costs, pricing efficiencies, mitigation of double marginalization, and innovation through technological integration).

[41] Patrick Rey & Jean Tirole, A Primer on Foreclosure, in 3 Handbook of Industrial Organization 2145, 2152 (Mark Armstrong & Robert Porter eds., 2007).

[42] On libertarian and classical-liberal critiques of competition law, see Ayn Rand, Antitrust: The Rule of Unreason, in Capitalism: The Unknown Ideal 63–71 (1966) (arguing that antitrust law penalizes economic success); Bruno Leoni, Freedom and the Law 25–28, 76–85 (expanded 3d ed. 1991) (criticizing the discretionary character of competition adjudication); Murray N. Rothbard, Man, Economy, and State 661–75 (2004) (rejecting the welfare-theoretic foundations of antitrust law); Dominick T. Armentano, Antitrust and Monopoly: Anatomy of a Policy Failure (2d ed. 1990). See also Dominick T. Armentano, Antitrust: The Case for Repeal (2d ed. 1999), arguing that antitrust law “cannot be reformed; it must be abolished”—not merely constrained, but repealed in its entirety, including the institutions responsible for enforcement.

[43] See, e.g., Case T-201/04, Microsoft Corp. v. European Commission, 2007 E.C.R. II-3601 (imposing an access obligation and protocol-licensing requirement, and requiring the unbundling of Windows Media Player); Case AT.40437, Apple—App Store Practices (Music Streaming), Eur. Comm’n Decision of Mar. 4, 2024 (addressing anti-steering provisions and commission structures under Article 102 TFEU). On the Commission’s broader quasi-price-setting role in competition enforcement, see Wouter P.J. Wils, Discretion and Prioritisation in Public Antitrust Enforcement, 34 World Competition 353, 356 (2011).

[44] Lee McGowan, The Antitrust Revolution in Europe: Exploring the European Commission’s Cartel Policy 3 (2010). See also Rein Wesseling, The Modernisation of EC Antitrust Law (2000) (noting the Commission’s “lack of political accountability which cannot be resolved directly at the [EU] level”).

[45] Auer & Radic, supra note 19 (arguing that the Draft Guidelines adopt a form-based approach that revives presumptions of harm displaced by the consumer-welfare standard).

[46] Eur. Comm’n, Draft Merger Guidelines ¶ 20 (Apr. 30, 2026) [hereinafter Draft Merger Guidelines].

[47] DMA arts. 5(2) (cross-service data combination), 5(3) (most-favored-nation clauses), 5(4) (anti-steering), 5(7) (mandatory use of gatekeeper payment systems), 5(8) (cross-core-platform-service bundling), 6(2) (use of business-user data to compete against those users), 6(3) (uninstallation and default-setting requirements), 6(4) (sideloading and third-party app stores), 6(5) (self-preferencing), 6(7) (interoperability with hardware and software features), 6(9)–(10) (data portability and real-time data access for business users), 6(11) (FRAND access to ranking, query, click, and view data for rival search engines), 6(12) (FRAND access conditions for business users), 7 (messaging interoperability). For a mapping of these provisions onto conduct categories traditionally analyzed under Article 102 TFEU, see Manne, Radic & Auer, supra note 11, at 14–20.

[48] DMA art. 3(2). Designation as a gatekeeper turns on quantitative thresholds—annual European Economic Area turnover of at least €7.5 billion or a market capitalization of at least €75 billion in each of the preceding three financial years, together with at least 45 million monthly active end users and 10,000 yearly active business users. The designation process does not require a showing of market power, dominance, or anticompetitive effects. Cf. TFEU art. 102; Case 27/76, United Brands Co. v. Commission, 1978 E.C.R. 207, ¶ 65 (requiring a finding of dominance in a defined relevant market).

[49] DMA arts. 5–7. The European Commission has described the Article 5 obligations as “self-executing.” Eur. Comm’n Staff Working Document, DMA Compliance Reports Summary 3 (2024).

[50] DMA recital 35 states that the obligations “are necessary to address identified public policy concerns, there being no alternative and less restrictive measures that would effectively achieve the same result.” The manifest-error standard that governs judicial review under Article 263 TFEU applies to the European Commission’s enforcement decisions, but not to the DMA’s legislative obligations themselves. Because those obligations operate per se, they are not subject to challenge based on their competitive effects. See Case C-12/03 P, Eur. Comm’n v. Tetra Laval BV, EU:C:2005:87, ¶¶ 38–39. Under the DMA, courts review the Commission’s compliance assessments, not the antecedent question whether a challenged obligation causes competitive harm. By contrast, that inquiry remains central under EU competition law.

[51] DMA art. 19(3) (authorizing the European Commission to propose amendments to the regulation through the ordinary legislative procedure).

[52] On competition enforcement as an exercise conducted under conditions of pervasive uncertainty, see Manne & Radic, supra note 8, at 279.

[53] Case C-48/22 P, Google LLC & Alphabet Inc. v. Eur. Comm’n (Google Shopping), EU:C:2024:726, ¶ 165 (holding that conduct may be categorized as abusive only when it departs from competition on the merits), ¶¶ 184–85 (requiring a case-specific assessment of a practice’s capability to restrict competition). See also id. ¶¶ 263–67 (discussing the as-efficient-competitor test).

[54] DMA art. 3; recital 11.

[55] See, e.g., Eur. Comm’n commitments decisions in Booking.com (2015) and Amazon Marketplace (2017) (most-favored-nation clauses and anti-steering provisions); Case AT.39740, Google Search (Shopping) (2017), upheld in Case C-48/22 P, Google LLC & Alphabet Inc. v. Eur. Comm’n, EU:C:2024:726 (self-preferencing); Case T-201/04, Microsoft Corp. v. Eur. Comm’n, EU:T:2007:289, and Case T-167/08, Microsoft Corp. v. Eur. Comm’n (Microsoft II), EU:T:2012:323 (tying); Bundeskartellamt, Case B6-22/16, Facebook (2019), endorsed in Case C-252/21, Meta Platforms Inc. v. Bundeskartellamt, EU:C:2023:537 (data combination). DMA arts. 5(3)–(4), 5(7)–(8), 6(2), and 6(5) remove these categories of conduct from effects-based analysis and subject them to per se prohibitions.

[56] Thomas A. Lambert, The Essence of an Antitrust Violation 2, 8 (Univ. of Mo. Sch. of Law Legal Studies Rsch. Paper No. 2024-34, 2024). For the European Union analogue, see Case C-209/10, Post Danmark A/S v. Konkurrencerådet, EU:C:2012:172, ¶¶ 20–22 (holding that Article 102 TFEU targets conduct that harms competition to the detriment of consumers, while emphasizing that “not every exclusionary effect is necessarily detrimental to competition”).

[57] Eur. Parl., Comm. on the Internal Mkt. & Consumer Prot. (IMCO), Amendments to the Draft Digital Markets Act (2021). Proposed amendments that would have identified “consumer welfare” as an explicit objective were rejected during the legislative process. Instead, the Council and Parliament adopted “contestability” and “fairness” as the DMA’s operative objectives without consumer welfare serving as a limiting principle. See DMA recital 11.

[58] DMA recitals 10–11.

[59] Id. recital 33; DMA arts. 6(2), 6(5), 6(10), 6(11). See also Pablo Ibáñez Colomo, The Draft Digital Markets Act: A Legal and Institutional Analysis, 12 J. Eur. Competition L. & Prac. 561, 562, 565 (2021) (arguing that the DMA “is crafted to grant substantial leeway to restructure digital markets and re-allocate rents”); Fiona Scott Morton & Cristina Caffarra, The European Commission Digital Markets Act: A Translation (VoxEU, Jan. 5, 2021) (“[T]he regime is not designed to regulate infrastructure monopolies, but rather to create competition as well as to redistribute some rents.”).

[60] William J. Baumol, John C. Panzar & Robert D. Willig, Contestable Markets and the Theory of Industry Structure 1–20 (1982).

[61] Jacques Crémer, Yves-Alexandre de Montjoye & Heike Schweitzer, Competition Policy for the Digital Era 4 (Rep. for the Eur. Comm’n 2019); Richard Feasey & Giorgio Monti, Implementing the Digital Markets Act, 13 J. Eur. Competition L. & Prac. 577, 578 (2022). See also Pinar Akman, Regulating Competition in Digital Platform Markets: A Critical Assessment of the Framework and Approach of the EU Digital Markets Act, 47 Eur. L. Rev. 85 (2022) (observing that “fairness” and “contestability” may be in tension because contestability implies a market has not yet tipped, while the existence of gatekeepers presupposes that it has); Giuseppe Colangelo, In Fairness We (Should Not) Trust: The Duplicity of the EU Competition Policy Mantra in Digital Markets, 68 Antitrust Bull. 618, 622 (2023) (“the endemic uncertainty surrounding the notion of fairness has traditionally made it unsuitable to operate as a stand-alone legal standard”).

[62] DMA recitals 5–11. The European Commission’s Impact Assessment expressly presented the new regime as a response to perceived limitations of Article 102 TFEU enforcement, including the duration of investigations, the evidentiary demands of effects-based analysis, and the difficulty of proving consumer harm in rapidly evolving digital markets. See Eur. Comm’n Staff Working Document, Impact Assessment Report Accompanying the Proposal for a Digital Markets Act, SWD(2020) 363 final, ¶¶ 27–34, 56–62. See also Margrethe Vestager, Competition Policy for the Digital Age (speech delivered in Brussels, Feb. 17, 2020).

[63] The European Commission’s powers under the DMA are formally constrained in several respects. First, gatekeeper designation is criteria-driven under Article 3, although the qualitative criteria in Article 3(8) confer substantial discretion at the margins. Second, Articles 5–7 establish the substantive obligations, but Article 8(2) specification proceedings allow the Commission to determine what compliance requires in practice. Third, rights of defense apply under Articles 29 and 34, although substantive determinations are reviewed under the deferential manifest-error standard. Fourth, acts adopted under Articles 8, 18, 29, and 30 are subject to judicial review under Article 263 TFEU, subject to the deference afforded to complex economic assessments. See Eur. Comm’n v. Tetra Laval BV, supra note 50, ¶¶ 38–39. Fifth, the Charter of Fundamental Rights of the European Union constrains the Commission’s exercise of its powers, including through Article 16 (freedom to conduct a business), Article 17 (right to property), and Article 47 (effective judicial protection). On the practical limits of these constraints, see Wolfgang Kerber, Taming Tech Giants with a Per-Se Rules Approach? The Digital Markets Act from a Rules-Versus-Standards Perspective, 7 Concurrences 28, 33–35 (2022).

[64] Eur. Comm’n Decision of Apr. 23, 2025, Case AT.40920—Meta/Article 5(2) DMA [hereinafter Meta Decision], ¶ 45.

[65] Id. ¶ 67. On the economics of two-sided platforms and cross-subsidization, see Jean-Charles Rochet & Jean Tirole, Platform Competition in Two-Sided Markets, 1 J. Eur. Econ. Ass’n 990 (2003); David S. Evans & Richard Schmalensee, Matchmakers: The New Economics of Multisided Platforms (2016).

[66] Meta Decision, supra note 64, ¶ 134; DMA art. 9(1). The Commission stated expressly that it “need not consider DMA interpretations’ effect on gatekeepers’ profits” unless the gatekeeper invokes Article 9(1).

[67] Meta Decision, supra note 64, ¶ 112. A substantial body of research on the “privacy paradox” suggests that users consistently report valuing privacy while behaviorally preferring free, data-funded services to privacy-protective alternatives. See Susan Athey, Christian Catalini & Catherine Tucker, The Digital Privacy Paradox: Small Money, Small Costs, Small Talk (Nat’l Bureau of Econ. Rsch., Working Paper No. 23488, 2017). If the privacy paradox holds, the sub-1% subscription rate may reflect revealed consumer preferences rather than deficiencies in the paid alternative.

[68] Under Article 102 TFEU, the Commission would have needed to show that Meta’s model constituted an exploitative abuse under the exceptional-circumstances framework developed in United Brands and Post Danmark. Meta could have invoked objective justification on the ground that subscription-or-ads models are a standard commercial practice and argued that the sub-1% subscription rate reflects revealed consumer preferences. Neither argument was available under the DMA.

[69] Eur. Comm’n Decision of Apr. 23, 2025, Case AT.40921—Apple/Article 5(4) DMA [hereinafter Apple Anti-Steering Decision], ¶¶ 89, 142.

[70] Eur. Comm’n, Preliminary Findings, Case AT.40922—Apple/Article 6(4) DMA [hereinafter Apple Distribution Preliminary Findings], ¶¶ 44, 67, 89 (Mar. 2025). Articles 6(4) and 6(7) DMA include security and privacy carve-outs that permit gatekeepers to adopt measures that are “strictly necessary and proportionate” to protect the integrity of operating systems or software. Because those provisions are subject to specification proceedings, the carve-outs may be invoked in connection with Article 6 obligations. They do not apply to the Article 5 obligations, including the anti-steering prohibition at issue in the Apple Anti-Steering Decision, supra note 69.

[71] Apple Distribution Preliminary Findings, supra note 70, ¶ 78.

[72] Oscar Borgogno & Giuseppe Colangelo, App Stores as Public Utilities, 45 World Competition 237, 241 (2022). See also Damien Geradin, What to Do About Apple’s App Store Fees?, The Platform Law Blog (“the Commission faces the unenviable task of determining what these fees should be”). In Epic Games, Inc. v. Apple Inc., 67 F.4th 946, 985–86 (9th Cir. 2023), the 9th U.S. Circuit Court of Appeals accepted Apple’s security and privacy justifications as nonpretextual, recognized Apple’s entitlement to compensation for its intellectual-property investments, and tailored the remedy to the specific anti-steering harm at issue. That remedy was also subject to appellate review. Under the DMA, the equivalent determination rests with a single enforcer and is reviewable only for manifest error.

[73] On the rarity of excessive-pricing enforcement and the difficulty of identifying a workable benchmark, see Pinar Akman & Luke Garrod, When Are Excessive Prices Unfair?, 7 J. Competition L. & Econ. 403 (2011).

[74] See supra Section III (discussing the “structural unfairness axiom”).

[75] The European Commission also served as the principal drafter of the DMA.

[76] Eur. Comm’n, Specification Proceedings DMA.100209, Preliminary Findings Communicated to Alphabet ¶¶ 22–78 (Apr. 16, 2026).

[77] Id.

[78] Article 6(11) DMA frames the obligation in terms of “third-party undertakings providing online search engines.” The provision does not define that term to include AI chatbots or generative-AI assistants. Their inclusion derives from the European Commission’s interpretation in the specification proceedings, rather than from the text of the DMA itself. See id. ¶¶ 22–78.

[79] For the view that generative-AI assistants increasingly compete with general search for informational queries, see Mohammad Amer et al., Comparing Conversational Chatbots and the Internet for Consumer Information Search, 44 Behav. & Info. Tech. 314 (2025) (describing generative AI as being “in direct competition with the internet … the incumbent search platform”). Market-tracking data point in the same direction: Google’s share of combined search-and-AI “information discovery” reportedly fell from roughly 89% to 58% between December 2022 and December 2025, while AI assistants generated sessions equivalent to a substantial share of search-engine volume. The contrary view also finds support. Most generative-AI use does not substitute for search. Aaron Chatterji et al., How People Use ChatGPT (Nat’l Bureau of Econ. Rsch., Working Paper No. 34255, 2025), finds that nearly 80% of ChatGPT activity consists of non-search tasks, predominantly writing, distinguishing chatbots functionally from search engines. Scale also remains relevant: aggregate chatbot traffic in 2024–25 amounted to only a fraction of search-engine traffic, suggesting chatbots are not yet a macro-level substitute. Experimental evidence likewise indicates that consumers continue to prefer search engines for product-information searches. See Sang-Hoon Kim & Robert Priluck, Consumer Responses to Generative AI Chatbots Versus Search Engines for Product Evaluation, 20 J. Theoretical & Applied Elec. Com. Rsch. (2025).

[80] The point holds even if the European Commission had not interpreted Article 6(11) DMA to include AI chatbots among the beneficiaries of Google’s data-sharing obligations. Some group of third parties would still benefit. The relevant point is that the DMA’s objectives entail improving the position of third parties, rather than focusing exclusively on competitive harm.

[81] On the distinction between private harm to a competitor and social harm to the competitive process, see, e.g., Brown Shoe Co. v. United States, 370 U.S. 294, 320 (1962); Brunswick Corp. v. Pueblo Bowl-O-Mat, Inc., 429 U.S. 477, 488–89 (1977). For the European analogue, see Case C-209/10, Post Danmark A/S v. Konkurrencerådet, EU:C:2012:172, ¶¶ 21–22 (explaining that competition on the merits may, by definition, lead to the marginalization of less efficient competitors). See also Richard A. Posner, Antitrust Law 18–25 (2d ed. 2001).

[82] Frank H. Easterbrook, The Limits of Antitrust, 63 Tex. L. Rev. 1, 32–36 (1984). See also Brunswick Corp. v. Pueblo Bowl-O-Mat, Inc., 429 U.S. 477, 488–89 (1977) (holding that antitrust injury must be of the type the antitrust laws were intended to prevent).

[83] Brunswick Corp v Pueblo Bowl-O-Mat Inc, 429 US 477, 489 (1977) (‘the antitrust laws were enacted for the protection of competition, not competitors,’ quoting Brown Shoe Co v United States, 370 US 294, 320 (1962)).

[84] Case C-209/10 Post Danmark A/S v Konkurrencerådet (Post Danmark I) EU:C:2012:172, para 22.

[85] On the conceptual and doctrinal difficulties that arise when the distinction between protecting competition and protecting competitors collapses, see Auer & Radic, supra note 35.

[86] John F. Kennedy, Special Message to the Congress on Protecting the Consumer Interest (Mar. 15, 1962).

[87] The phrase derives from John R. Hicks, Annual Survey of Economic Theory: The Theory of Monopoly, 3 Econometrica 1, 8 (1935) (“the best of all monopoly profits is a quiet life”).

[88] Auer & Radic, supra note 35.

[89] Auer & Radic, supra note 35, at 15–16.

[90] Org. for Econ. Cooperation & Dev. (OECD), Fairness and Competition, DAF/COMP(2018)11, at 8 (2018).

[91] Crémer, de Montjoye & Schweitzer, supra note 61, at 7.

[92] Robert H. Bork, The Antitrust Paradox: A Policy at War with Itself 347–64 (1978). For the Stiglerian inversion, see George J. Stigler, The Theory of Economic Regulation, 2 Bell J. Econ. & Mgmt. Sci. 3 (1971) (arguing that regulation is “acquired by the industry and is designed and operated primarily for its benefit”). The DMA inverts that pattern: rather than being acquired by regulated firms, it was shaped by firms that stood to benefit from the obligations imposed on gatekeepers.

[93] Coalition for App Fairness, Statement of Principles (Sept. 2020), https://appfairness.org. The founding members included Basecamp, Blix, Blockchain.com, Deezer, Epic Games, the European Publishers’ Council, Match Group, News Media Europe, Prepear, Proton, SkyDemon, Spotify, and Tile. The Coalition launched three months before the European Commission proposed the DMA on Dec. 15, 2020.

[94] Robert Gorwa, Grzegorz Lechowski & Daniel Schneiß, Platform Lobbying: Policy Influence Strategies and the EU’s Digital Services Act, 13(2) Internet Pol’y Rev. 15 (2024). The authors document that, in November 2020, Spotify policy staff emailed European Commission President Ursula von der Leyen’s cabinet—specifically digital adviser Anthony Whelan—to promote the newly formed Coalition for App Fairness and its campaign for ex ante gatekeeper regulation focused on Apple. Although the article examines the Digital Services Act package, the lobbying effort concerned the ex ante gatekeeper rules that ultimately became the DMA.

[95] DMA recital 33 frames the objectives of contestability and fairness largely in terms of benefits to business users, including “enabling business users to effectively overcome barriers to entry and expansion” and addressing “the imbalance between the rights and obligations of gatekeepers and their business users.” DMA art. 8(8) directs the European Commission, when specifying obligations under Articles 6(11) and 6(12), to assess whether “an imbalance of rights and obligations on business users” persists. Article 12 likewise authorizes the Commission to update gatekeeper obligations where such imbalances remain. The Commission’s first review of the DMA reflects the same orientation. See Eur. Comm’n, Report on the Review of Regulation (EU) 2022/1925, COM(2026) 178 final, at 4–6 [hereinafter DMA Review Report].

[96] The asymmetry arises from the structure of the fee. Apple’s Core Technology Fee imposes a charge of €0.50 for each first annual installation above a threshold of one million installations during a 12-month period and applies to developers that adopt Apple’s new European Union business terms. For paid applications, or applications generating revenue through in-app purchases, the fee is generally small relative to the revenue generated by each installation, and most developers remain below the threshold. For free, advertising-supported applications distributed at very high volume, however, the same per-installation charge may exceed the revenue generated by individual users. The burden therefore falls unevenly across business models despite a uniform nominal rate. See Eur. Comm’n, Preliminary Findings, Apple—App Store (DMA.100103, June 25, 2024) (addressing the Core Technology Fee and alternative-distribution terms).

[97] On industrial policy as government intervention targeting specific sectors, industries, or firms to shape the economy’s structure, direct resources, and promote particular competitive outcomes, see Dani Rodrik, Industrial Policy for the Twenty-First Century 1 (Harvard Kennedy Sch. Working Paper, 2004); OECD, New Industrial Policy and the Challenge of Self-Discovery, DAF/COMP(2022)4. Oles Andriychuk expressly frames the DMA as industrial policy and describes the European Commission’s evolving role as the “strategic shaping” of digital markets rather than ex post policing of competitive harm. Oles Andriychuk, EU’s New Industrial Policy for Digital Markets: Do We Know How?, EU Law Live (May 26, 2025). Andriychuk argues that competition enforcers are moving “from their reductionist role of market controllers” toward a more proactive role as “market co-designers,” with the DMA and the United Kingdom’s Digital Markets, Competition and Consumers Act marking that transformation.

[98] DMA art. 8(8) directs the European Commission, when specifying obligations, to assess whether there is “no remaining imbalance of rights and obligations on business users.” Article 12 empowers the Commission to update gatekeeper obligations when that imbalance persists. These provisions build business-user satisfaction directly into the statute’s operative enforcement metrics.

[99]Cf. Easterbrook, supra note 82, at 12.

[100] On the DMA’s influence on consumer law, see Francisco de Elizalde, Digital Platforms and Fragmentation of Consumer Law, 30 Maastricht J. Eur. & Compar. L. 88, 94 (2023). See also Francisco de Elizalde, Fragmenting EU Consumer Law: Digital Laws, Their Premises, and Their Normative Implications, 48 J. Consumer Pol’y 3 (2025) (arguing that the digital-law framework replaces the traditional consumer-trader distinction with a broader user-based model, extends protections to businesses, introduces tiered obligations based on firm size, and elevates power imbalance as the organizing principle).

[101] Gemma Petrie & Tasos Stampelos, Six Million Selections Later: How the DMA Is Giving People Browser Choice, Mozilla Pol’y Blog (May 11, 2026).

[102] Chamber of Progress, DMA Consumer Impact Report 12–19 (2025) (cataloguing Apple’s delayed rollout of iPhone Mirroring and AirPods Pro Live Translation in the European Union, weakened App Store content moderation following mandatory sideloading, and parental controls that do not extend to alternative app stores). On Siri AI, see Press Release, Apple Inc., Due to DMA, Siri AI Delayed in EU for iOS 27 and iPadOS 27 (June 2026), https://www.apple.com/newsroom/2026/06/due-to-dma-siri-ai-delayed-in-eu-for-ios-27-and-ipados-27/.

[103] Eur. Ctr. for Int’l Pol. Econ. (ECIPE) & European Platform Pol’y Ctr. (EPPC), Consumer Experiences with the Digital Markets Act 23 (Ipsos Survey 2025) (reporting that 55% of respondents favor strong rules for digital platforms, while 39% report needing more steps to complete previously simple tasks, roughly one-third report less seamless digital experiences, and 80% remain unfamiliar with the DMA).

[104] The label “malicious compliance” gained prominence in debates over the DMA after Epic Games CEO Tim Sweeney described Apple’s initial compliance proposal as “a horror show” of “malicious compliance” and “junk fees.” Tim Sweeney (@TimSweeneyEpic), X (Jan. 25, 2024, 4:57 PM), https://x.com/TimSweeneyEpic/status/1750589570880516402. See also Sarah Perez, Epic Games CEO Calls Out Apple’s DMA Rules as Malicious Compliance and Full of Junk Fees, TechCrunch (Jan. 26, 2024). Spotify CEO Daniel Ek similarly described Apple’s proposal as “a new low” and “extortion.” Daniel Ek, quoted in Sarah Perez, Spotify Calls Apple’s DMA Compliance Plan Extortion and a Complete and Total Farce, TechCrunch (Jan. 26, 2024). Competition lawyer Damien Geradin likewise argued that Apple’s proposal “takes the European Commission for fools.” Damien Geradin, research note (Jan. 2024), reported in Matthew Broersma, Apple Infringes DMA With New EU Developer Terms, Silicon UK (Jan. 29, 2024). Sweeney repeated the characterization in later commentary. See Tim Higgins, Apple Reveals Complex System of App Store Fees to Avoid EU Fine, CNBC (June 26, 2025).

[105] For an acknowledgment of the tradeoff by DMA proponents, see Zach Meyers, Will the Digital Markets Act Hurt European Consumers? (Ctr. for Eur. Reform Pol’y Brief, Sept. 2023) (“[T]he DMA could make some online services worse for consumers in the short term,” and “a degree of short-term consumer inconvenience may be needed to promote greater competition in the long run”).

[106] On the DMA’s autopoiesis, see supra Section VI.C.

[107] Polanyi, supra note 7, at 150; Schwartz, supra note 6, chs. 3–5.

[108] Emmanuel Macron, Address at the Internet Governance Forum, Paris (Nov. 12, 2018) (“If we want technological sovereignty, we’ll have to adapt our competition law, which has perhaps been too focused on the consumer and not enough on the challenge that the major platforms represent.”). For the broader policy context, see Eur. Parl. Rsch. Serv., supra note 3, at 7–8 (identifying ex ante regulation of digital gatekeepers as a central component of the European Union’s digital-sovereignty strategy).

[109] Draft Art. 102 Guidelines, supra note 19, ¶¶ 8–11; Auer & Radic, supra note 19.

[110] Case C-261/21, Autorità Garante della Concorrenza e del Mercato v. Illumina, Inc., EU:C:2023:341.

[111] Draft Merger Guidelines, supra note 46, ¶ 20.

[112] Case T-334/22, Google LLC v. Eur. Comm’n (Android Auto), EU:T:2024:000; Philipp Hornung, Android Auto and the DMA’s Influence on Article 102 TFEU Case Law, 45 Eur. L. Rev. 789, 802 (2024); Giuseppe Colangelo, The EU Essential Facilities Doctrine after Android Auto: A Wild Card without Limiting Principles?, 62 Common Mkt. L. Rev. 1299 (2025). On the broader cross-fertilization among consumer law, competition law, and the DMA, see Case 283/81, CILFIT v. Ministero della Sanità, EU:C:1982:335, ¶ 20 (requiring European Union law to be interpreted coherently as a whole). See also the argument advanced in this paper that the interaction of these regimes can produce a compounding effect, with the limitations of one discipline offset through mutually reinforcing interpretations across EU law.

[113] Ioannis Lianos, Polycentric Competition Law: The Geo-Economic Turn 6 (CLES Rsch. Paper No. 2/2026, Apr. 2026).

[114] Ariel Ezrachi & Maciej Bernatt, The Rising Instrumentalisation of Competition and Antitrust Enforcement (draft 2026) (presented at the Oxford Centre for Competition Law and Policy 2026 Event on Competition and Democracy).

[115] Elias Deutscher, Reshaping Digital Competition: The New Platform Regulations and the Future of Modern Antitrust, 67 Antitrust Bull. 302, 302–03, 340 (2022) (identifying four dimensions along which the DMA reconfigures modern antitrust: the consumer-welfare objective, the error-cost framework, effects-based methodology, and the probabilistic standard of proof).

[116] See Gov’t of Japan, The Grand Design and Action Plan for a New Form of Capitalism 2023 Revised Version 33–38, 60 (Cabinet Decision, June 16, 2023) (placing digital-platform regulation alongside wage redistribution and green transformation under the broader project of “upgrading capitalism”).

[117] Jean Tirole, Competition and Industrial Policy in the 21st Century, 3 Oxford Open Econ. i983 (2024).

[118] American Innovation and Choice Online Act, S. 2992, 117th Cong. (2021); American Innovation and Choice Online Act, S. 2033, 118th Cong. (2023). Neither bill received a floor vote. On the Federal Trade Commission’s shift under the current administration, see Mark R. Meador, Antitrust Policy for the Conservative 31–32 (May 1, 2025), https://www.ftc.gov/system/files/ftc_gov/pdf/antitrust-policy-for-the-conservative-meador.pdf (calling for Congress to “statutorily cabin[] the use of economic evidence” and codify consumer welfare as “consumer or trading partner surplus”).

[119] UK Gov’t, Press Release, Doug Gurr Appointed as Interim Chair of the Competition and Markets Authority (Jan. 21, 2025). Marcus Bokkerink resigned after the prime minister, chancellor, and business secretary urged regulators, including the Competition and Markets Authority, to “tear down the barriers hindering business.” His interim replacement, Doug Gurr, previously served as Amazon’s United Kingdom country manager.

[120] Mario Draghi, The Future of European Competitiveness pt. A, at 1 (Eur. Comm’n, Sept. 9, 2024) (describing Europe’s competitiveness challenge as “existential”). The European Commission’s Competitiveness Compass adopted many of Draghi’s recommendations, including regulatory simplification, as strategic priorities—in unresolved tension with the DMA’s interventionist architecture. Eur. Comm’n, A Competitiveness Compass for the EU (Jan. 2025).

[121] India: Standing Comm. on Fin., Twenty-Fifth Report on the Implementation of the Competition Act, 2002 (Lok Sabha, Aug. 11, 2025) (recommending withdrawal and redrafting of the Digital Competition Bill). Japan: Act on Promotion of Competition for Specified Smartphone Software, Act No. 58 of 2024 (Japan) (enacted June 12, 2024) (adopting a framework significantly narrower than the DMA). South Korea: Korea Fair Trade Comm’n, Legislative Approaches to Promote Competition in E-Commerce Platform Markets (Sept. 2024) (proposing amendments to the existing Monopoly Regulation and Fair Trade Act rather than adopting a standalone ex ante platform law).

[122] Easterbrook, supra note 82, at 3 (arguing that antitrust procedure should minimize the combined costs of anticompetitive conduct that escapes condemnation, competitive conduct that is wrongly condemned or deterred, and the administrative costs of the legal system itself). See also Manne & Radic, supra note 8, at 273–76 (applying the error-cost framework to the DMA and arguing that its per se rules condemn conduct—including self-preferencing, vertical integration, and third-party data use—for which plausible procompetitive explanations remain available).

[123] Rey & Tirole, supra note 41, at 2152; Geoffrey A. Manne & Joshua D. Wright, Innovation and the Limits of Antitrust, 6 J. Competition L. & Econ. 153 (2010).

[124] Geoffrey A. Manne & Dirk Auer, Antitrust Dystopia and Antitrust Nostalgia, 28 Geo. Mason L. Rev. 1279, 1322–26 (2021). On the self-reinforcing nature of false positives, see Easterbrook, supra note 82, at 15 (arguing that erroneous condemnation of efficient conduct can permanently chill the practice because firms abandon the condemned behavior, whereas false negatives are more likely to self-correct because market entry remains possible).

[125] Harold Demsetz, Information and Efficiency: Another Viewpoint, 12 J.L. & Econ. 1 (1969) (coining the “nirvana fallacy” to describe the error of comparing real-world institutions with idealized alternatives rather than with feasible ones. The DMA’s implicit comparison between pre-regulatory digital markets and an idealized post-regulatory counterfactual reflects this logic). See also Herbert Hovenkamp, Antitrust and Platform Monopoly, 130 Yale L.J. 1952, 1955–60 (2021) (“the closer one looks at digital platforms the less they seem to be winner-take-all”).

[126] Friedrich A. Hayek, The Use of Knowledge in Society, 35 Am. Econ. Rev. 519, 524–27 (1945) (arguing that the information necessary to coordinate economic activity is dispersed among millions of individuals, embedded in local and context-specific knowledge, and revealed through the competitive process rather than centralized decision-making). See also Friedrich A. Hayek, Competition as a Discovery Procedure, 5 Q.J. Austrian Econ. 9 (2002) (reprinting a 1968 lecture); Israel M. Kirzner, Competition and Entrepreneurship (1973). For an application of these insights to competition law, see Manne & Radic, supra note 8.

[127] See, e.g., DMA recitals 7 (contestability and fairness), 31 (benefits to end users and business users), 32 (effective compliance and the prevention of imbalance), 33 (overcoming barriers to entry and expansion), and 79 (innovation and consumer welfare as outputs of the regime). Taken together, these recitals present the DMA as simultaneously advancing end-user welfare, business-user welfare, and innovation, without acknowledging the tradeoffs that the literature has long identified among those objectives.

[128] DMA Review Report, supra note 95; Eur. Comm’n Staff Working Document, SWD(2026) 123 final (Apr. 28, 2026). Article 53(1) DMA requires the European Commission to evaluate the regulation and submit a report “by 3 May 2026, and subsequently every 3 years.” The review is conducted by the enforcer itself; the DMA contains no provision for an independent or external audit. SWD(2026) 123 final notes, under the heading “stakeholder feedback,” that trade associations reported “40–50% losses in organic traffic for some retailers” and that hospitality small and medium-sized enterprises reported “20–30% declines in web traffic or direct sales” following Alphabet’s DMA compliance measures, but does not treat those reports as findings requiring substantive analysis.

[129] The Ukrainian alternative app store Setapp Mobile exited the market after concluding that Apple’s DMA compliance terms rendered its business model unviable. See David Snow, Setapp Mobile Shuts Down, Cult of Mac (Jan. 15, 2026), https://www.cultofmac.com/news/setapp-mobile-shuts-down.

[130] The European Commission concluded that “the DMA had thus far effectively contributed to the core objective of making digital markets in the EU fairer and more contestable.” DMA Review Report, supra note 95, at 4. The review evaluates success primarily through enforcement metrics—including designation decisions, noncompliance proceedings, and specification dialogues—rather than through measures of consumer welfare or broader market outcomes.

[131] See Elias Deutscher, The Competition–Democracy Nexus Unpacked: Competition Law, Constitutional Democracy and the Politicisation of Antitrust, 41 Y.B. Eur. L. 197 (2022) (arguing that competition law is inherently political and that the consumer-welfare standard obscures, rather than eliminates, underlying political choices).

[132] Herbert Hovenkamp, The Slogans and Goals of Antitrust Law, 25 N.Y.U. J. Legis. & Pub. Pol’y 705, 708–12 (2023) (documenting broad support for the consumer-welfare standard across jurisdictions and political cycles and arguing that pluralistic alternatives have repeatedly failed to produce predictable or consistent outcomes). See also A. Douglas Melamed & Nicolas Petit, The Misguided Assault on the Consumer Welfare Standard in the Age of Platform Markets, 54 Rev. Indus. Org. 741 (2019); Herbert Hovenkamp, Is Antitrust’s Consumer Welfare Principle Imperiled?, 45 J. Corp. L. 65 (2019).

[133] The tension appears in a line of U.S. Supreme Court decisions that prioritized the protection of less efficient competitors over consumer welfare. See Brown Shoe Co. v. United States, 370 U.S. 294 (1962) (blocking a merger between two relatively small shoe companies on the ground that “occasional higher costs and prices” were a price worth paying to preserve a fragmented industry); United States v. Von’s Grocery Co., 384 U.S. 270 (1966) (blocking a merger that would have produced a firm with only a 7.5% market share because the broader trend toward industry consolidation was itself viewed with suspicion). The Court’s subsequent course correction began with Brunswick Corp. v. Pueblo Bowl-O-Mat, Inc., 429 U.S. 477, 488–89 (1977) (“the antitrust laws … were enacted for the protection of competition, not competitors,” quoting Brown Shoe, 370 U.S. at 320), which emphasized that antitrust injury must reflect harm to the competitive process rather than merely harm to a competitor.

[134] The debate is longstanding and extensive. On the protection-of-competitors side, see Eleanor M. Fox, The Modernization of Antitrust: A New Equilibrium, 66 Cornell L. Rev. 1140 (1981); Harry First & Spencer Weber Waller, Antitrust’s Democracy Deficit, 81 Fordham L. Rev. 2543 (2013); Lina Khan, Amazon’s Antitrust Paradox, 126 Yale L.J. 710 (2017). For the contemporary neo-Brandeisian restatement, see Tim Wu, The Curse of Bigness: Antitrust in the Gilded Age (2018); Lina Khan & Sandeep Vaheesan, Market Power and Inequality: The Antitrust Counterrevolution and Its Discontents, 11 Harv. L. & Pol’y Rev. 235 (2017). On the protection-of-competition side, see Bork, supra note 92, at 50–66, 90–106; Easterbrook, supra note 82; Posner, supra note 81, at 9–32.

[135] Manne & Radic, supra note 8.

[136] Louis Kaplow & Steven Shavell, Fairness Versus Welfare, 114 Harv. L. Rev. 961 (2001) (demonstrating that any departure from welfarism, if applied consistently, will in some circumstances require selecting outcomes that leave every individual worse off). The argument applies with particular force to polycentric competition standards, which multiply the dimensions along which such Pareto-inferior outcomes can arise. See also Cass R. Sunstein, Incommensurability and Valuation in Law, 92 Mich. L. Rev. 779 (1994) (arguing that incommensurability expands rather than constrains regulatory discretion because it prevents legal reasoning from resolving the conflicts it generates).

[137] Stigler, supra note 92.

[138] Gordon Tullock, The Welfare Costs of Tariffs, Monopolies, and Theft, 5 W. Econ. J. 224 (1967); Mancur Olson, The Logic of Collective Action (1965) (arguing that concentrated, well-organized interests are more likely than diffuse interests to prevail in political and regulatory processes. In the DMA context, business users with specific grievances against gatekeepers may therefore enjoy a structural advantage over consumers). See also James M. Buchanan & Gordon Tullock, The Calculus of Consent: Logical Foundations of Constitutional Democracy (1962) (arguing that general rules constrain rent-seeking by applying equally across parties, whereas discretionary interventions create opportunities for rent extraction).

[139] Damien Neven & Lars-Hendrik Röller, Consumer Surplus vs Welfare Standard in a Political Economy Model of Merger Control, 23 Int’l J. Indus. Org. 829, 845 (2005).

[140] Mario Monti, quoted in Nicolas Levy, EU Merger Control: From Birth to Adolescence, 26 World Competition 195, 211 (2005). The institutional pathology became apparent in Airtours plc v. Commission, Case T-342/99, 2002 E.C.R. II-2585; Schneider Electric SA v. Commission, Case T-310/01, 2002 E.C.R. II-4071; and Tetra Laval BV v. Commission, Case T-5/02, 2002 E.C.R. II-4381, where the General Court annulled Commission decisions because of speculative evidence, failure to satisfy the burden of proof, and inadequate protection of procedural rights. The 2004 merger-control reforms—including peer-review panels and the creation of the Chief Economist position—responded directly to those failures. See Werner Berg, New EC Merger Regulation: A First Assessment of Its Practical Impact, 24 Nw. J. Int’l L. & Bus. 683 (2004).

[141] See Francisco-José Garcia Sanchez, The Rise of the Authoritarian International Economic Order (forthcoming, Geo. J. Int’l L. 2026) (arguing that European Union measures such as the Carbon Border Adjustment Mechanism, the Foreign Subsidies Regulation, and green industrial policy reflect “regulatory protectionism framed as fairness,” and grouping them with U.S. and Chinese state-capitalist measures as part of a broader shift away from rules-based market governance toward discretionary economic statecraft).