ICLE White Paper

Filling the DMA’s Gaps: Fairness, Contestability, and the Missing Endpoint

Executive Summary

The Digital Markets Act (DMA) directs its regulatory machinery toward “fairness” and “contestability,” yet defines neither goal clearly. Critics argue that this ambiguity gives the European Commission broad discretion, while defenders note that terse competition laws have gained meaning through enforcement and adjudication. This white paper asks whether two years of DMA enforcement have begun to fill those gaps. It reconstructs the Act’s implicit theory from its text, then tests that theory against the Commission’s impact assessment, designation and infringement decisions, specification proceedings, statutory review, and DMA–GDPR guidance.

The record reveals that the DMA’s goals are coherent but redistributive. Fairness means correcting a structural imbalance the Act conclusively presumes between gatekeepers and those who depend on their core platform services. Contestability supplies the remedy by transferring competitive opportunity, resources, and market share toward rivals and business users. User choice serves both goals, but the Act values it primarily when users move away from gatekeepers.

Four concepts explain this logic. The Structural Unfairness Axiom presumes that gatekeeper markets are inherently unfair. The Bidirectional Unfairness Inference allows structural asymmetries and disappointing outcomes to confirm each other as evidence of unfairness. The Scale–Unfairness Paradox treats the same growth that improves a platform as evidence of deeper entrenchment. Dialectical Fairness seeks to level up business users while leveling down gatekeepers. The enforcement record reflects these ideas by treating gatekeeper growth as weak contestability, defining fairness through prohibited conduct, elevating parity without effects analysis, and measuring success through third-party access and uptake rather than price, quality, or innovation.

What the record still lacks is calibration. The DMA identifies who should gain and who should lose without specifying how much, for how long, or against what benchmark. Because its logic assumes that covered markets were never fair, it supplies no baseline or endpoint. Compliance may therefore remain unverifiable and intervention potentially endless. That problem matters for gatekeepers, reviewing courts, consumers, and jurisdictions importing DMA-style rules without fully understanding the program they embody.

I. Introduction

The Digital Markets Act (DMA or Act)[1] has drawn persistent criticism because its twin goals—fairness and contestability—remain ambiguous.[2] Critics raise a familiar objection: Broad objectives allow the European Commission to shift the goalposts and exercise considerable discretion over enforcement and interpretation, weakening the system’s predictability and legal certainty.[3]

Yet the DMA is hardly unique in this respect. Landmark competition laws—the Sherman Act in the United States, most famously, and Articles 101 and 102 of the Treaty on the Functioning of the European Union (TFEU) to a lesser extent—are famously terse. Their underlying logic took shape gradually through litigation and enforcement.[4] Such openness may reflect a deliberate legislative choice to let enforcers, regulated parties, and courts fill the gaps.[5] It may instead, however, reveal poor drafting or a legislature that lacked a clear vision of what it sought to achieve.

Against this backdrop, this white paper asks a simple question: Have two years of enforcement clarified the DMA’s goals and, if so, how?

The question matters to both scholarship and policy. Scholars remain divided over two related issues: whether the DMA’s goals can be meaningfully defined and what those goals require. The literature has debated whether fairness and contestability are coherent, compatible, or even legally cognizable objectives. Few studies, though, have examined how concrete enforcement decisions, interpretive communications, and early institutional practice have given those concepts meaning.

Greater clarity would also provide practical guidance. It would help gatekeepers and firms seeking access to understand what compliance requires and why. It would also help jurisdictions considering DMA-style regulation—including the United Kingdom, Japan, and Brazil[6] —understand the rules they may be importing, including their logic, principles, underlying aims, likely results, and shortcomings.[7]

This white paper argues that the DMA does contain a coherent conception of fairness and contestability, but the Act never states it clearly. That conception emerges only by reading the law alongside the European Commission’s enforcement decisions. Properly reconstructed, fairness and contestability are not free-standing values in tension, as much of the literature assumes. They are, instead, two sides of a single redistributive program.

Fairness supplies a conclusive diagnosis: Gatekeepers and the businesses that depend on them suffer from a structural imbalance that designation conclusively presumes. Contestability supplies the remedy: a continuing transfer of competitive opportunity away from gatekeepers and toward actual or potential rivals. User choice serves both goals, but the DMA only values that choice when (and because) users exercise it away from gatekeepers. This structure explains why the Act can treat “fair and contestable” as a hendiadys—two terms expressing a single idea.

That coherence comes at a price. The DMA’s goals are directional but uncalibrated. The Act identifies who should lose and who should gain, but not how much, for how long, or against what benchmark. Because it presumes that markets for core platform services were never fair, it supplies no earlier equilibrium to restore and no baseline against which to measure progress. Any improvement may therefore prove insufficient, making compliance impossible to verify.

Two years of enforcement have brought us closer to understanding what fairness and contestability mean under the DMA. They have brought us no closer to knowing when, if ever, the Act’s goals will have been achieved.

The remainder of this white paper proceeds as follows. Section II situates the inquiry within the scholarly debate. It explains why the DMA’s goals require an abstract definition, why the Act fails to provide one, and how commentators have divided over what those goals entail. Section III reconstructs the DMA’s implicit theory of fairness and contestability from its recitals and operative provisions. It introduces four analytical constructs that organize the analysis: the Structural Unfairness Axiom, the Bidirectional Unfairness Inference, the Scale–Unfairness Paradox, and Dialectical Fairness.

Section IV tests that reconstruction against the enforcement record, including the impact assessment, designation and infringement decisions, specification proceedings, the Commission’s first statutory review, and joint guidelines governing the interplay between the DMA and the General Data Protection Regulation (GDPR). Section v distills the findings into an explicit reformulation of the DMA’s twin goals. Section VI examines the implications of what remains unknown. Section VII concludes.

II. The Debate Over the DMA’s Undefined Goals

Scholars agree that fairness and contestability must guide the DMA’s interpretation, but the Act defines neither goal nor clearly connects them to its obligations. This ambiguity has produced competing accounts of the DMA’s legal character, aims, and criteria for success.

A. Why the DMA’s Goals Require Definition

The early literature on the DMA’s goals converges on three related conclusions. First, the Act requires an abstract account of fairness and contestability. Their meaning cannot simply be inferred “by doing”—that is, from the prohibitions and obligations in Articles 5 through 7.

Giorgio Monti argues that conceptual clarity would guide courts and help stakeholders understand the Act’s policy objectives.[8] Richard Feasey and Monti likewise treat a clear statement of the DMA’s goals as essential to assessing both compliance and the Regulation’s success. In their account, the DMA creates an “obligation to provide a result on the market,” not merely a set of formal duties.[9] Jacques Crémer et al. similarly argue that courts and regulators need a working definition of fairness and contestability to interpret Articles 5, 6, and 7. The goals express policy aspirations, but they also guide the European Commission’s compliance assessments and any resulting judicial review.[10]

The DMA’s institutional design reinforces this point. The Act allows the Commission to specify what certain obligations require, reopen proceedings when compliance measures prove ineffective, and adapt the rules to new forms of unfairness or weak contestability. Each power presupposes a standard for deciding whether a compliance measure advances the Act’s goals. A list of practices labeled unfair cannot, by itself, supply that overarching standard.[11]

Second, the DMA does not provide one. Friso Bostoen observes that the Act seeks to ensure that markets in which gatekeepers operate “are and remain contestable and fair,” yet defines neither term.[12] Its closest approximation to a general definition of fairness appears in recital 33, which states that unfairness “should relate to an imbalance between the rights and obligations of business users where the gatekeeper obtains a disproportionate advantage.”

As Pinar Akman notes, that formulation leaves basic questions unanswered, including how fairness and contestability apply in markets that have already tipped toward a dominant platform.[13] Nicolas Petit argues that the vague terminology provides little reason to believe the legislature had a clear theory of competition or coherent economic policy in mind. The Act’s structure obscures both its primary objectives and the reasons the legislature selected some obligations over others, raising broader questions about its legitimacy.[14] Giuseppe Colangelo goes further, portraying fairness as a mantra that grants enforcers broad discretion rather than a meaningful legal standard. As he writes, “the endemic uncertainty surrounding the notion of fairness has traditionally made it unsuitable to operate as a stand-alone legal standard.”[15]

Third, the relationship between the DMA’s goals and obligations remains unclear, perhaps because the boundary between fairness and contestability is itself blurred. Monti argues that the Act does not identify which provisions advance which objective. Grouping obligations by policy objective, he suggests, would clarify the Act’s approach and aid judicial interpretation.[16]

Petit similarly finds that “the structure of the DMA makes it very hard to unpack the primary objective of each obligation.” Fairness and contestability do not necessarily reinforce one another. Conduct deemed unfair could, in some circumstances, encourage entry and thereby improve contestability.[17] Akman draws a related distinction: Contestability largely concerns market structure, while fairness primarily concerns conduct. She doubts that this “two-headed” design can deliver greater contestability in markets where gatekeepers already hold entrenched positions.[18]

B. Competing Accounts of the DMA’s Goals

A third conclusion follows from the first two: If the DMA does not adequately define the objectives that guide its interpretation, competing accounts will fill the gap. The literature offers several.

One dispute concerns whether the DMA extends competition law or creates a separate regulatory regime. This may appear to be a formal distinction, but it determines how far established competition-law concepts, benchmarks, and safeguards should constrain enforcement. Some scholars find the DMA’s effort to distance itself from competition law unconvincing because the Act’s recitals connect its intended improvements to traditional competition parameters, including innovation and consumer benefits.[19] Others view the DMA as more than streamlined enforcement of Article 102 TFEU. On this account, the Act looks forward and seeks to shape—and perhaps constrain—the conditions under which gatekeepers exercise power.[20] Still others cast it as a democratic reassertion of rulemaking authority over platforms that competition-law standards should not govern.[21]

Geoffrey A. Manne, Lazar Radic, and Dirk Auer classify the DMA as “digital competition regulation.” In their account, the Act departs from antitrust’s consumer-welfare standard and revives structural concerns and presumptions associated with neo-Brandeisianism and ordoliberalism—traditions the antitrust mainstream has largely rejected. The DMA places conduct that might otherwise fall under Article 102 within a separate framework whose goals, evidentiary standards, and presumptions more closely reflect the regulator’s policy preferences.[22]

Even sympathetic observers acknowledge uncertainty about the Act’s character. Pierre Larouche and Alexandre de Streel describe the DMA as occupying a “difficult and perhaps ominous epistemological position.”[23] Comparative analysis deepens the ambiguity because other jurisdictions have incorporated provisions identical to the DMA’s directly into their national competition laws. The result resembles a “Schrödinger’s regulation” in which the same substantive rules simultaneously are and are not competition law.[24]

A second dispute concerns the relationship between fairness and contestability. Crémer and his coauthors treat them as mutually reinforcing. They describe a “dual-action dial” under which a practice cannot qualify as fair if it undermines contestability.[25] By contrast, de Streel and his coauthors treat the goals as distinct but connected: Contestability lowers entry barriers and enables competitive pressure, while fairness balances rights and obligations within the platform ecosystem.[26] Petit rejects the premise that the two goals necessarily move together.[27] Akman likewise questions whether the same tools can achieve objectives that may pull in different directions.[28]

This disagreement extends to whether the DMA addresses the causes or merely the symptoms of weak competition. A regime focused on preserving contestability would logically target markets before they tip toward a dominant platform. The DMA instead regulates markets in which gatekeepers already exist. That choice leaves the Act’s underlying ambition unclear: Does it accept gatekeepers while regulating their conduct, or does it seek to reshape market structure and ultimately erode their positions?[29]

Because the DMA supplies no clear definitions, scholars have also reconstructed its broader purposes. Manne, Radic, and Auer identify three objectives shared by the DMA and similar regimes: leveling down gatekeepers, assisting rivals, and redistributing economic rents—returns above competitive levels—from gatekeepers to business users. They connect these objectives to the European Union’s broader pursuit of digital sovereignty and strategic autonomy.[30]

After reviewing gatekeepers’ compliance reports, Colangelo and Alba Ribera Martínez connect fairness and contestability to long-term goals of market modeling, ecosystem openness, neutralizing competitive advantages, and transparency. They characterize the fairness-related goals as expressly redistributive.[31] Oles Andriychuk offers a more direct assessment, characterizing the Act as “punitive.”[32]

C. Why Articles 5–7 Are Not Enough

One response is that the DMA needs no overarching definition because Articles 5 through 7 give fairness and contestability their full content. On this view, the two goals begin and end with the obligations themselves. But that account fails both functionally and on the Act’s own terms.

Recital 79 directs the European Commission to “apply a predefined standard” when identifying practices that are unfair or limit contestability. Such a standard would promote predictability and enable the Commission to exercise its other statutory powers. Article 12, for example, allows the Commission to extend the DMA through delegated acts to cover practices that the Act does not enumerate but that nonetheless qualify as unfair or contestability-limiting.

Other provisions likewise require the Commission to look beyond the obligations’ text. Article 18 authorizes remedies when gatekeepers have failed to achieve “fairness and contestability.” Article 19 permits market investigations to identify additional practices that are unfair or restrict contestability. Article 53 requires the Commission to assess every three years whether the DMA has achieved its objective of contestable and fair markets.

Article 8(1) makes the need for an overarching standard clearest. It requires a gatekeeper’s compliance measures to be “effective in achieving the objectives of this Regulation and of the relevant obligation.” This distinction implies that compliance depends on both the specific purpose of an individual obligation and the Act’s broader goals.[33]

Taken together, these provisions presuppose three things: a coherent meaning of fairness and contestability, a benchmark for measuring them, and an endpoint at which the Commission can declare them achieved. The Act supplies none. The question, then, is whether the DMA’s text and enforcement record allow those missing elements to be reconstructed.

III. Reconstructing the DMA’s Goals

Although the DMA never defines fairness or contestability, its recitals and operative provisions reveal a coherent direction. The Act treats gatekeeper scale and user dependence as sources of structural unfairness, seeks to improve business users’ position, and promotes rival entry and expansion by reducing gatekeepers’ competitive advantages.

Four analytical constructs organize this reconstruction. Together, they show that the DMA treats user choice as a tool for shifting activity toward rivals and contestability as market restructuring. The Act’s direction is clear, but it supplies no benchmark for measuring success or determining when intervention should end.

A. The DMA’s Implicit Theory of Fairness

The DMA defines fairness only indirectly, through its opposite. Recital 33 states that “[u]nfairness . . . should relate to an imbalance between the rights and obligations of business users where the gatekeeper obtains a disproportionate advantage.”[34] Recital 62 expands on that definition, holding that pricing or other general access conditions:

. . . should be considered unfair if they lead to an imbalance of rights and obligations imposed on business users or confer an advantage on the gatekeeper which is disproportionate to the service provided by the gatekeeper to the business users or lead to a disadvantage for business users in providing the same or similar services as the gatekeeper.[35]

The Act therefore treats digital markets dominated by gatekeepers as unfair because they distribute rights and obligations unequally across the supply chain. It attributes that inequality to persistent differences in bargaining power that allow gatekeepers to capture a disproportionate share of the value created on their platforms while business users receive inadequate returns on their contributions.[36]

This theory rests on the assumption that users, especially business users, depend on gatekeepers’ core platform services (CPSs). The Act does not distinguish between business users that depend on a gatekeeper and those that do not. Nor does it distinguish firms whose dependence results from their own choices from those that could not reasonably avoid it.[37] This omission implies that the Act treats dependence as an inherent structural feature of CPSs. Bargaining asymmetries increase dependence, which further entrenches those asymmetries and weakens business users’ position.

A closer reading reveals four related themes. First, the DMA treats gatekeeper scale and other structural market features as sources of unfairness that both produce and confirm unfair outcomes. Second, it conceives fairness as a relative redistribution that improves business users’ position while curtailing gatekeepers’ advantages. Third, it values user choice primarily as a means of encouraging switching away from gatekeepers. Fourth, it invokes fair prices, innovation, quality, and other desirable outcomes without providing benchmarks for measuring them. Together, these themes supply the DMA’s implicit theory of fairness.

1. The Structural Unfairness Axiom

The DMA begins with the premise that markets for core platform services (CPSs)—and gatekeepers in particular—are structurally unfair. This white paper calls that conclusive presumption the DMA’s Structural Unfairness Axiom (DSUA). The Act attributes the unequal distribution of rights and obligations between gatekeepers and their business and end users to market features such as network effects, data-driven economies of scale, tipping, and lock-in.[38] On this reading, those features are inherently unfair and create imbalances that only regulation can correct. Unfair inputs necessarily produce unfair outcomes.

The logic is deliberately circular. If an outcome appears unfair, the market characteristics that produced it are presumed unfair. If the market is structurally unfair, its outcomes are necessarily tainted. Competition law disciplines such reasoning through external benchmarks such as consumer welfare and efficiency. The DMA offers no comparable yardstick.

The Act instead asks whether allegedly unfair structural features have been neutralized and whether the resulting distribution is fair. Each inquiry depends on the other. This white paper calls that feedback loop the Bidirectional Unfairness Inference (BUI): Structural features are deemed unfair and therefore presumed to produce unfair outcomes, while outcomes perceived as unfair confirm the structural features’ inherent unfairness.

The DSUA also assumes that the potential for unfairness scales with the size of a CPS. Within the Act’s logic, a larger user base increases the risk that the platform’s structural features will produce unfair outcomes. Network effects and data accumulation can improve a platform’s quality while deepening users’ dependence on it. As the platform becomes more attractive, users become less likely to switch. The same features that make a CPS work well thus make it, in the DMA’s view, inequitable and prone to unfairness.

This white paper calls that dynamic the Scale–Unfairness Paradox (SUP): Greater scale can simultaneously improve the quality and attractiveness of a CPS and heighten the risk of structural unfairness. The result is a partially inverse relationship between DMA-style fairness and service quality. The dynamics that improve the service are the same dynamics that, under the DMA’s theory, undermine fairness.

The DMA’s designation criteria make this logic especially visible. The Act’s use of market capitalization, turnover, and user numbers[39] treat a company’s scale as the ultimate source of unfairness. High turnover might ordinarily suggest that a firm creates substantial value for consumers. The DMA instead reads it as evidence of leveraging potential, market tipping, and latent unfairness.

Once a firm crosses the Article 3 thresholds and becomes a gatekeeper, the Act prohibits specified conduct that would otherwise remain lawful and commonplace. A company does not become a gatekeeper because it acted unfairly. The Act treats its conduct as unfair because the company is a gatekeeper. The recitals extend this logic beyond the quantitative thresholds. They treat the ability to monetize a service, “superior access” to financial markets, and high relative growth as indicators of gatekeeper status and entrenchment—and therefore of unfairness or its potential.[40]

If this structural account is correct, the DMA does not restore fairness so much as construct it. Markets for CPSs are unfair under the Act because of their inherent characteristics, which means they were never fair in the sense the DMA now uses that term. The Act has no prior equilibrium to restore and no historical benchmark against which to measure future success.

This account also yields an implicit operational test. Conduct that increases user numbers, turnover, market capitalization, monetization potential, scale, or access to capital strengthens the structural conditions associated with gatekeeper status. By reinforcing the conditions the Act treats as unfair, such conduct becomes evidence of unfairness itself.

2. Dialectical Fairness: Leveling Up and Down

If the current distribution of rights and obligations is unfair, fairness requires a different allocation. Because the DMA identifies business users as the disadvantaged group, the new allocation must leave them better off relative to gatekeepers. Recital 62 illustrates this logic: Unfairness arises from both a “disadvantage for business users” and an advantage “disproportionate” to the gatekeeper.

Fairness under the DMA is therefore relative and dialectical. It measures each group’s position against the other and seeks movement in both directions: leveling up business users while leveling down gatekeepers. The goal is a new allocation that narrows the gap between them. Fairness must leave business users better off or gatekeepers worse off—and preferably do both.

The Act’s operative provisions reveal how this redistribution works. Article 5(2) restricts a gatekeeper’s ability to process, combine, or cross-use personal data without consent. The provision might appear to address privacy. Recital 36 frames it differently: Cross-use is problematic because it gives gatekeepers an advantage over competitors, allowing them to combine data sources, improve their services, and undermine contestability.[41]

The provision targets gatekeepers’ competitive data advantages rather than data collection itself. By contrast, other provisions—including Article 6(9)’s data-portability mandate—encourage data sharing and movement. The DMA instead treats gatekeepers’ data advantages as a source of unfairness precisely because those advantages have competitive value. On this view, gatekeepers should not compete using resources their rivals lack.

Recital 72 applies the same logic to consumer profiling. It states that greater transparency should discourage gatekeepers from making intensive profiling the industry standard, “given that potential entrants or start-ups cannot access data to the same extent and depth, and at a similar scale.”[42] The DMA thus de facto favors privacy-intensive models over data-driven ones. It both advantages firms without comparable data access and defines the terms on which the Act expects them to challenge incumbents.

Recital 46 and Article 6(2) follow the same pattern. Accumulating business-user data is not unfair per se. Unfairness arises when a gatekeeper uses that data in competition with the business users that generated it.[43] The provision prohibits such use in any market, even when the resulting competition might benefit end users.

Here, “unfair” means that business users could receive more. All else equal, they would benefit if gatekeepers exerted less competitive pressure against them. Fairness therefore operates as a mandate to relax the competition business users face by preventing gatekeepers from deploying competitively valuable resources against them.

Article 6(5)’s self-preferencing prohibition offers the clearest example. Gatekeepers may not favor their own products, but the Act does not prevent them from favoring third-party offerings. Article 6(7)’s interoperability mandate and Article 6(10)’s real-time data-access requirement likewise compel gatekeepers to share the products of their investments with third parties and competitors. These obligations reduce gatekeepers’ advantages and place the two sides on more equal footing.[44]

The underlying premise is that a gatekeeper should share its competitive advantages because its exclusive possession of them is inherently unfair. More generous access terms and lower prices for business users qualify as fair because they move the parties’ positions in both directions. They improve business users’ position, curtail the gatekeeper’s “disproportionate advantages,” and narrow the imbalance the DMA seeks to correct.

3. Choice as Directed Switching

Several DMA recitals suggest that practices become unfair when they restrict user choice. Recital 70, for example, instructs gatekeepers not to “subvert or impair user autonomy, decision-making, or choice.”[45] That language might suggest that choice is an independent value under the Act. A closer reading reveals a directional preference: The DMA encourages users to choose away from gatekeepers’ products and core platform services and toward rival offerings.

The Act’s asymmetries are systematic. Article 6(3) requires gatekeepers to make changing default settings easy, but only when those defaults direct users toward the gatekeeper’s own services. The Act does not prohibit defaults that steer users toward third-party products, even though such steering could also “subvert” user autonomy within the meaning of recital 70.[46]

Article 6(4) likewise requires gatekeepers to permit third-party apps and app stores and allows those third parties to prompt users to make their products the default. EU law imposes no corresponding obligation on other firms to open their systems to gatekeepers or anyone else. Defaults that might otherwise lock in users become unobjectionable—and perhaps desirable—when they benefit a firm other than a gatekeeper.

Article 5(3) prohibits parity clauses, which prevent business users from offering better terms through other channels, because they “unfairly” restrict business users’ freedom. The gatekeeper’s freedom to adopt such clauses receives no comparable weight, even when the clauses prevent businesses from free-riding on the platform.[47] Articles 5(4), 5(7), and 5(8) follow the same pattern by reducing reliance on gatekeepers and easing users’ path to alternative products and services. In each case, the Act promotes choices that move users away from gatekeepers.

If user choice were an overarching objective, the Act would express it as a general rule. All firms would have to make software easy to uninstall, and no firm could steer users toward its preferred channels. The DMA instead values choice only when it advances the Act’s other goals. It seeks to level down gatekeepers by directing users toward rivals and creating room for those rivals to challenge the gatekeeper’s position. The Act may even subordinate convenience to that goal, most visibly through mandatory choice screens that require users to take additional steps.

The DMA therefore values switching more than choice itself. Switching demonstrates contestability because it suggests that users have overcome entry barriers. It advances fairness by redirecting users and traffic away from gatekeepers. Under this logic, regulators may measure the effectiveness of choice mechanisms through observed switching rates and treat low switching as evidence of inadequate implementation. Choice becomes “meaningful” only when it leads away from the gatekeeper.

Taken to its logical conclusion, the Structural Unfairness Axiom implies that choices made before the DMA took effect were not genuinely meaningful. Structural unfairness tainted those choices, making them normatively suspect and unworthy of acceptance at face value.

4. Outcomes Without Benchmarks

Some commentators argue that the DMA seeks to create rights and opportunities rather than particular market outcomes.[48] Yet the Act does contain several outcome-oriented goals. Recital 5 refers to “ensuring fair economic outcomes” when market processes cannot do so. Recital 107 connects the Act’s objectives to “promoting innovation, high quality of digital products and services, fair and competitive prices, as well as a high quality and choice for end users.” Recital 45 asserts that advertising costs are “likely to be higher than they would be in a fairer, more transparent and contestable platform environment.”[49]

The Act provides no benchmark for these outcomes. It does not explain what constitutes a fair price, adequate innovation, or fair product quality. Competition law, the closest neighboring framework, relies on market power and the competitive counterfactual—a comparison with conditions absent the challenged conduct—to identify suboptimal outcomes.

The DMA expressly separates its analysis from those tools. Recital 23 makes market definition and efficiencies irrelevant to designation. Recital 11 states that the DMA pursues goals distinct from those of competition law and operates “without prejudice” to anticompetitive effects, objective justifications, and efficiency considerations.[50]

By excluding those considerations, the Act discards competition law’s principal method for determining when prices, output, or innovation are inadequate. Competition law asks whether market power distorts competition and harms consumers. The DMA supplies no equivalent test. Competition-law principles therefore cannot define “fair” outcomes under the Act, except perhaps at the margins.

B. The DMA’s Implicit Theory of Contestability

The DMA treats contestability as rivals’ ability to enter, scale, and take market share from gatekeepers. Although some provisions recognize competition on the merits, the Act primarily seeks to restructure markets by reducing gatekeeper advantages and promoting third-party growth.

1. Contestability as Market Restructuring

The DMA also leaves contestability undefined. Mirroring its negative account of fairness, recital 3 suggests that contestability is lacking when a gatekeeper’s position remains difficult to challenge “even by more innovative or efficient market operators.” It attributes that difficulty to high barriers to entry or exit and limited access to key inputs such as data.[51] Recital 5 adds that certain platform markets are “prone to tipping” and that gatekeepers hold entrenched and durable positions.

Two features of this account stand out. First, contestability extends beyond the formal freedom to enter a market. It asks whether rivals can attract users, achieve scale, and exert meaningful pressure on gatekeepers. The emphasis on effectiveness matters. Without observable entry, expansion, or shifts in market structure toward challengers, it becomes difficult to say that rivals can contest the market effectively. The DMA thus appears to infer inadequate contestability from rivals’ failure to gain a meaningful foothold. As with the Bidirectional Unfairness Inference, the absence of the desired outcome becomes evidence of a structural defect.

Second, contestability entails reducing entry barriers to induce actual entry and expansion by third parties, regardless of the barriers’ source. This includes advantages produced by otherwise legitimate business conduct, such as data accumulation. Recital 44, for example, objects to tying because it gives gatekeepers “a potential advantage in terms of accumulation of data” that “is liable to raise barriers to entry.”[52]

The treatment of online search follows the same logic. The DMA links contestability to gatekeepers’ data-driven quality advantages, then requires them to share query data with rivals. The premise is that access to those data will help rivals improve their services, attract users, and gain market share.[53]

Under this account, contestability requires reducing a critical advantage on one side so that the other can compete more effectively and take market share from incumbents. The Act treats access to data—and, by extension, scale and a large user base—as an illegitimate advantage when concentrated in gatekeepers, yet affirmatively provides that access to their rivals. That asymmetry reflects a broader objective of restructuring markets by dispersing market share among more firms.

2. Gatekeeper Growth as Evidence of Weak Contestability

If contestability requires entry and expansion by third parties, continued gatekeeper growth points in the opposite direction. The DMA operationalizes this implication of the Scale–Unfairness Paradox through Article 18.

In deciding whether to impose additional behavioral or structural remedies, Article 18 asks whether a gatekeeper has “maintained, extended or further strengthened its impact in the internal market.”[54] The inquiry does not consider whether that increased impact resulted from superior performance or a better product, even though growth in turnover and user numbers may reflect precisely those advantages.

Recital 75 confirms this approach. A gatekeeper “should be deemed to have maintained, extended or strengthened its gatekeeper position where, despite the enforcement actions taken by the Commission, that gatekeeper still holds or has further consolidated or entrenched its importance as a gateway for business users to reach end users.”[55]

The Act likewise presumes an “entrenched and durable position” when a core platform service reaches a very high number of users across several Member States for three years. By the same logic, contestability improves when a gatekeeper’s user numbers decline. If network effects, scale, and data advantages restrict contestability, reducing those advantages should enhance it.

Gatekeeper growth therefore cuts against the Act’s objectives even when it results from success on the merits. Taken to its logical conclusion, the DMA could treat a gatekeeper’s loss of market share as evidence that its compliance measures are working.

3. Limits of the Restructuring Account

Not every DMA provision fits the market-restructuring account. Article 6(5) permits a gatekeeper’s products to rank first when that result follows from transparent, fair, and nondiscriminatory criteria. This provision therefore supports an efficiency-based reading: A gatekeeper may prevail on the merits but may not leverage its position to distort competition in adjacent markets.

Other passages point in the same direction. Recital 53 states that offering better services or lower prices does not create a prohibited barrier to switching. Recitals 55 and 57 frame interoperability as a way to ensure that rival hardware competes on its merits.[56] These provisions support a consumer-oriented conception of contestability.

Other parts of the Act are difficult to reconcile with that conception. Article 5(2) restricts data combination without asking whether integration improves service quality. Article 6(11) requires search gatekeepers to share query and click data with rivals because the competitive advantage generated by legitimate scale and user engagement creates a structural obstacle. Article 6(2) bars gatekeepers from using business-user data even when doing so would produce a genuinely superior product. Recital 64 similarly treats the integration of communication services into an ecosystem as an entry barrier regardless of whether users actually multihome.[57]

This broad conception makes many consumer-facing integrations suspect. It also casts doubt on claims that the Act primarily seeks to benefit end users or increase innovation overall, rather than promote innovation by firms other than gatekeepers. If consumer welfare and innovation were controlling objectives, the DMA would distinguish efficient from inefficient entry and efficient from inefficient exclusion, as competition law does. It makes neither distinction.

The Act retains a residue of consumer-oriented contestability, but its broader structure subordinates consumer welfare to effective third-party entry and market restructuring.

C. The DMA’s Known Unknowns

A close reading reveals a more coherent account of fairness and contestability than the DMA expressly provides. Fairness rests on the Structural Unfairness Axiom (DSUA): Markets for core platform services are presumed structurally unfair because of their inherent characteristics. That presumption eliminates the need to prove anticompetitive conduct or harm.

The Bidirectional Unfairness Inference (BUI) reinforces the premise. Market structure is presumed to produce unfair outcomes, while outcomes deemed inadequate confirm that structural defects persist. The Scale–Unfairness Paradox (SUP) follows because growth and scale heighten the risk of unfairness even when gatekeepers earn them through competition on the merits. Dialectical Fairness supplies the remedy by leveling up business users while leveling down gatekeepers. The Act treats user choice as an instrument of that redistribution and directs it away from gatekeeper ecosystems.

The DMA’s conception of contestability likewise departs from the standard economic concept. Conventional contestability relies on the credible threat of entry to discipline incumbents. The DMA seeks to induce actual entry and expansion by transferring gatekeepers’ competitive advantages—including access to data, distribution, and interoperability—to rivals. It also treats continued gatekeeper growth as evidence of inadequate contestability regardless of the growth’s cause.

Taken together, these elements define fair and contestable markets as markets in which competitive advantages and market share move away from gatekeepers and toward rivals and business users. Consumer welfare remains an expected downstream benefit rather than the Act’s primary objective.

The Act nevertheless leaves five questions crucial to interpretation and enforcement unresolved. These are its “known unknowns.”

First, the DMA provides no benchmark for deciding when fairness or contestability has been sufficiently achieved. Article 8(8) presupposes an equilibrium without defining it. Article 18 authorizes remedies when the goals “have not been achieved” but supplies no standard of achievement. Articles 12, 19, and 53 assume that the goals can be measured without providing a yardstick. The “predefined standard” contemplated by recital 79 appears nowhere in the Act.

This measurement problem compounds the missing-baseline problem. If markets for core platform services were never fair, the DMA has no prior state against which to measure progress. Without a baseline, the Commission can always deem any improvement insufficient.

Second, the operative terms governing enforcement remain undefined. The Act does not explain what constitutes “equally effective interoperability,” “effective provision,” “effective use,” or “effective implementation.” Regulators cannot apply those terms consistently without first clarifying the goals that give them meaning.

Third, the Act does not identify which business or end users its obligations should benefit or how the Commission should manage tradeoffs among them. Neither group is monolithic. Small and large business users may have different interests, as may developers that use
“freemium” and paid models. A compliance measure may benefit one group at another’s expense, but the DMA supplies no principle for choosing among them.

Fourth, the DMA does not establish whether compliance turns on structural inputs, such as access, switching mechanisms, and data sharing, or distributional outputs, such as prices, market shares, quality, and switching rates. Under the BUI, disappointing outcomes after full structural compliance support an inference that structural defects remain. Structural compliance can therefore never guarantee protection from further intervention.

Fifth, the Act leaves consumers in an ambiguous position. Article 9(4) distinguishes the interests of third parties, including consumers, from “the objectives of this Regulation.” That distinction suggests that consumer interests do not themselves constitute an objective. Consumers instead appear to be indirect beneficiaries whose welfare the DMA expects to improve as a byproduct of market restructuring.[58]

The text therefore establishes a clear direction but no method of calibration. Section IV asks whether the enforcement record supplies what the text lacks.

IV. Testing the DMA’s Goals Against the Record

This section tests the preceding reconstruction against the DMA’s institutional and enforcement record, including the impact assessment, designation and infringement decisions, specification proceedings, first statutory review, and joint DMA–GDPR guidance. Across these sources, the Commission consistently treats gatekeeper advantages as structural problems and rival access, growth, and uptake as evidence of success, while supplying no benchmark for measuring achievement or resolving tradeoffs. Pinpoint references to institutional documents appear parenthetically in the text.

A. The Impact Assessment’s Built-In Conclusions

The Commission’s impact assessment is among the most revealing sources on the DMA’s goals.[59] It exposes the Regulation’s underlying chain of reasoning: assumptions about how digital markets work, harms inferred from those assumptions, and rules designed to remedy them.

That chain begins with a selective reading of the literature. The impact assessment cited almost no studies counseling against prescriptive platform regulation. It instead prioritized work supporting intervention, creating the appearance of consensus about both the harms of the digital economy and the urgency of addressing them.[60]

The Commission’s Regulatory Scrutiny Board (RSB) identified this weakness even after the Commission had “substantially redrafted” the assessment in response to an earlier rejection. The RSB concluded that the assessment:

. . . does not appropriately describe the shortcomings the initiative intends to address and does not provide a proper evidence base for them.

It also asked the Commission to:

. . . consider the negative consequences of curtailing the size advantages following from network economies and economies of scale for consumers [and] . . . more convincingly demonstrate for each of the selected core platform services that the identified weak contestability has negative effects in terms of higher mark-ups, lower quality of service, or reduced innovation (pp. 3–4).[61]

The Commission’s own quality-control body thus identified two central gaps before the DMA’s adoption: The assessment assumed harms rather than demonstrating them, and it discounted tradeoffs involving innovation.

The analysis also reflects what economist Harold Demsetz called the Nirvana fallacy—the mistake of comparing real institutions with idealized alternatives while ignoring the alternative’s imperfections and costs.[62] Echoing the mission letter’s promise to make markets “work better for consumers, business and society” (para. 8), the impact assessment seeks to make digital markets “better” without identifying either a baseline or an endpoint. Any market could conceivably perform better.

The assessment “fully recognise[s] the benefits that online platforms bring to the economy and society” (para. 21). It nevertheless argues that digital markets would perform even better if policymakers addressed “the most salient incidences of unfair practices and weak contestability” (para. 6). The assessment does not base its conclusion that markets function poorly on an observed decline in output, innovation, or economic performance. It instead infers poor performance from its starting assumptions. Because some ecosystems are “essentially impossible to contest . . . irrespective of how innovative and efficient [competitors] may be” (para. 27), “the likelihood increases that these markets do not function well—or may soon fail to function well” (para. 27).

Theoretical premises thus establish the imperfection of digital markets, while contrary views receive little consideration. The command to make markets work “better” supplies no stopping point. Identifying harm becomes both impossible, because no counterfactual exists against which to measure it, and unnecessary, because the theory presumes it. The DMA carries that method forward by dispensing with effects analysis and an efficiency defense.

The assessment draws an equally direct link between unfairness and size. It identifies gatekeepers’ “incomparable economic strength” (para. 88) through revenue, markups “amongst the highest in the world” (paras. 56, 129), profits that outperform the market (para. 96), and pandemic-era stock valuations (para. 95). Paragraph 88’s principal evidence for unfairness consists of app-store and social-network revenue figures.

The assessment then treats economic strength as de facto conclusive evidence of unequal bargaining power, which transforms ordinary business conduct into a potentially unfair practice (para. 41). Yet it never measures bargaining power or compares gatekeepers’ size with that of supposedly dependent business users. Some of those users rank among the world’s largest companies and include other gatekeepers. The assessment also fails to ask whether platforms depend on their business users or to identify a comparative benchmark.

A survey supplies the principal corroboration. It found that 88% of businesses reported “unfavourable trading conditions on large platforms,” which respondents attributed mostly to “a perceived imbalance of bargaining power” (para. 61). The assessment substantiates that perception only through proxies.

Its logic is straightforward. Gatekeepers’ size and economic success produce power. That power creates a bargaining imbalance, and the imbalance produces unfair outcomes. Curbing the growth, profitability, and revenue that provide gatekeepers with “economic strength” therefore becomes one of the Act’s principal objectives, if not the principal one. This reasoning supplies the foundation for the Structural Unfairness Axiom and Dialectical Fairness.

The impact assessment also portrays fairness and contestability as projects for redistributing wealth and restructuring digital markets. It measures anticipated improvements through a projected decline in the Herfindahl-Hirschman Index (HHI), a standard measure of market concentration (RSB Opinion, Annex I, Table 1.1). The stated aim is to make it easier for third parties to dislodge gatekeepers (para. 74). The assessment would achieve that aim by reducing competitive advantages it recasts as entry barriers, including data access, ecosystems, vertical integration, and zero prices (paras. 75–76).

Those measures would enable rivals to “[steal] market shares from” gatekeepers (para. 77) and reduce their “supra-normal profits” (para. 302). According to the assessment, more effective regulation produces a “more optimal . . . redistribution of profits” (para. 355).

Its treatment of scale is expressly asymmetric. The assessment acknowledges that the rules “would lead to curtailing the size of network effects and economies of scale thus reducing associated advantages for consumers” (para. 318). It does not deny that cost. Instead, it seeks to extend those advantages to “also non gatekeeper platforms benefitting from such advantages” (para. 318). Scale is desirable when challengers obtain it (paras. 29, 69, 308) and problematic when gatekeepers retain it.

The same logic governs innovation. The assessment treats the concentration of research and development (R&D) among a few firms as a “gap” for regulation to close rather than evidence of those firms’ contribution (para. 282). It accepts the possibility of reduced gatekeeper innovation on the premise that greater aggregate innovation, especially “amongst smaller businesses” (para. 104), will more than offset the loss (paras. 283–285). Yet the observable evidence available at the time showed higher R&D investment and greater innovation—the opposite of the predicted harm.

Finally, the impact assessment presumes the effects of the practices it regulates and writes those presumptions into law. It states that gatekeepers “monopolise data” (para. 38) without addressing data’s nonrivalrous character, meaning one firm’s use does not prevent another’s. It treats high markups as uniform evidence of inadequate competition across all core platform services (para. 56). It infers business-user “dependence” from revenue-share figures without explaining what would demonstrate independence (para. 58). It also assumes that gatekeepers will not pass on compliance costs (paras. 301–302) and excludes consumer-side costs from its calculations (para. 350).

The exercise was also quantitative. The assessment projected €13 billion in consumer surplus, €12 billion to €23 billion in income gains from increased R&D, and a 0.25-point decline in the HHI (Part 2, pp. 59–61). The DMA is therefore effects-based in a distinctive sense. The Commission calculated its effects once, across the entire regulatory scheme, and converted those estimates into statutory premises exempt from case-by-case verification. The benefits later invoked in the preamble and enforcement decisions were assumed before enforcement began, and subsequent decisions need not establish them again.

B. Designation Decisions: Contestability as Status

The DMA’s first enforcement acts designated gatekeepers under Article 3.[63] On their face, the seven decisions reveal little about the Act’s goals. Designation turns almost entirely on Article 3(2)’s quantitative presumptions, while fairness and contestability appear mostly in boilerplate. The decisions’ omissions and exclusions nevertheless reveal how the Commission understands those goals.

The principal omission concerns fairness. None of the seven decisions defines or applies the term. Four mention it only in titles and citations. The Apple decision mentions neither fairness nor contestability.

Fairness makes its sole substantive appearance in the Meta decision, paired with its statutory twin. The Commission states that multihoming across WhatsApp and Messenger “reinforce[s], rather than mitigate[s], the contestability and fairness issues” that led the legislature to include messaging services in the Act (Meta designation, para. 230). Fairness therefore does not function as an independent standard at the designation stage. Designation establishes whatever unfairness gatekeepers embody through classification alone.

The Commission gives contestability more content, but only by identifying what does not demonstrate a contestable market. Multihoming does not. As the ByteDance decision explains, “the fact that there is (some degree of) multi-homing does not preclude the possibility of weak contestability and unfair practices” (ByteDance, para. 136).

The presence of several large competitors also does not suffice. Contestability “can also be limited if there is more than one gatekeeper for a core platform service” (Meta designation, fn. 287, citing recital 32). The Commission also excludes conventional economic evidence. Rebuttal arguments “should relate directly to the quantitative criteria,” while “any justification on economic grounds such as those related to market definition or to efficiencies should be discarded, because it is not relevant to the designation as a gatekeeper” (Apple designation, para. 23, applying recital 23).

The ByteDance decision goes further by converting classic evidence of contestability into evidence of its absence. ByteDance argued that TikTok was “a recent entrant and challenger of several established ecosystems” and therefore could not hold an entrenched and durable position (para. 111). The Commission responded that Article 3(1)(c) “does not refer to an ‘unassailable’ position, but to an ‘entrenched and durable position’” (para. 156). It then presumed limited contestability from three years of high user numbers (para. 157) and concluded that “TikTok’s significant scale and growth . . . with an upward trajectory further supports the finding” of entrenchment (para. 158).

TikTok is the only successful large-scale entrant into a core platform market in the past decade. Its entry and growth would ordinarily provide strong evidence that the market remains contestable. The Commission treated them as evidence of the opposite based on the DMA’s formal, irrebuttable assumptions about core platform-service markets. The decision illustrates the Scale–Unfairness Paradox—the same entry and growth that gave TikTok scale also made it a gatekeeper.

Under the DMA, contestability operates as a status attached to firms rather than as a property of markets. The Commission infers that status from scale and duration, and firms cannot rebut it by pointing to entry, growth, switching, or multihoming—the very conduct that demonstrates contestability in its ordinary economic sense.

This approach reverses William J. Baumol’s conception of contestability, under which the threat of potential entry disciplines incumbents because incumbency offers no protection from competition.[64] Under the DMA, successful entry leads to incumbency, and incumbency triggers regulation.

C. Infringement Decisions: Prohibitions Without Endpoints

The Commission issued its first DMA infringement decisions against Apple’s anti-steering rules and Meta’s “consent or pay” model on April 23, 2025. These decisions provide the first applications of fairness and contestability to specific conduct. Both follow the same structure: They define the goals negatively through the elimination of practices labeled unfair or contestability-limiting, without identifying when a market becomes fair or contestable.

In the Apple decision, the Commission found that the App Store’s terms unlawfully restricted developers’ rights under Article 5(4) to communicate with users, promote offers, and conclude contracts outside the App Store.[65] Two findings stand out.

First, the Commission found that a disclosure sheet shown when users attempted to leave the App Store was “not neutral and objective” and could “deter end users from exercising their right under Article 5(4)” (para. 103). The sheet warned that Apple “is not responsible for the privacy or security of purchases made on the web.” According to the Commission, that warning implied that transactions outside Apple’s ecosystem were unsafe and therefore “unfairly reinforce[d] Apple as a gatekeeper” (para. 103).

Second, the decision prevents Apple from charging for “gatekeeper value,” defined as “the value of the core platform service provided by a gatekeeper, on top of its market value, because of its position as an important gateway” (para. 197). Apple may charge for facilitating a developer’s initial acquisition of a new user. It may not monetize the positional advantage associated with its gatekeeper status. The decision offers no criteria for determining when a fee crosses the line between legitimate compensation and payment for gatekeeper value.

The Commission adds that fairness may justify structural changes to a gatekeeper’s business and charging models when those restrictions are lawful, proportionate, and necessary to achieve “objectives of general interest recognised by the Union” (para. 212). Yet it does not specify how far those changes must go before the market becomes fair. Fairness remains an open-ended standard. A gatekeeper can eliminate conduct the Commission has labeled unfair without knowing whether it has reached the required endpoint.

The decision treats contestability in much the same way. Allowing Apple to charge fees on contracts concluded after steering would “reinforce developers’ dependence” and reduce contestability (para. 160). Contestability therefore means reducing dependence and increasing openness. The decision provides no criteria for determining when dependence has fallen far enough. Its enforcement analysis verifies that Apple removed the restrictions (para. 312). It does not ask whether developers can reach users in practice or whether alternative distribution channels have grown.

The Meta decision follows the same architecture.[66] Meta’s “consent or pay” model required users either to permit the combination of personal data across Meta services or to pay a subscription for an alternative. The Commission found that model inconsistent with Article 5(2).

The decision approaches fairness through imbalance and coercion. Designation, it states, “is a strong indication of the data controller’s power to impinge on the freedom of choice of the data subject.” Recital 4 links that power to “a serious imbalance in bargaining power and, consequently, unfair practices and conditions” (para. 183). The binary choice prevented users from withholding consent “without suffering a detriment” (para. 189), particularly given Meta’s “strong lock-in and network effects” (para. 190).

Contestability again turns on data as an input. The restriction on combining data seeks to prevent gatekeepers from exploiting “potential advantages in terms of accumulation of data” that “raise barriers to entry and hinder contestability” (para. 32). Article 5(2)’s choice mechanism therefore serves as a means of limiting gatekeeper data accumulation rather than as an independent end.

The decision does not explain what equivalence between paying with money and paying with data would eliminate coercion. Nor does it specify how much user control or data separation would make the market sufficiently open. A gatekeeper can identify what the Commission has prohibited yet cannot determine when its practices have become fair or contestable enough.

The Commission added a third data point on July 23, 2026, when it fined Google €890 million in two noncompliance decisions. It imposed a €460 million fine for favoring Google’s services in Google Search in violation of Article 6(5) and a €430 million fine for restricting steering from Google Play in violation of Article 5(4). The Commission gave Google 60 days to comply and backed that deadline with potential periodic penalty payments.[67]

The Commission had not published the full decisions at the time of writing. Based on its announcement, they appear to follow the Apple and Meta template: conduct-specific prohibitions, reduced dependence as the operative theory, and no articulated endpoint.

D. Specification Proceedings: Parity Without Effects Analysis

If the infringement decisions show what fairness and contestability prohibit, the specification proceedings show what they require. The DMA’s redistributive logic is clearest here.

In its preliminary findings on Alphabet’s search-data-sharing obligation under Article 6(11), the Commission introduced a “principle of parity.”[68] Under that principle, Alphabet may retain no advantage from data it collects unless eligible third-party search providers receive the same data. The Commission defines eligible providers to include chatbots with search functions (paras. 2–3).

Alphabet must therefore share all query, view, click, and ranking data that it collects to optimize its services. It must provide those data as frequently as it accesses them itself (para. 15). This parity principle goes beyond Article 6(11)’s text, which requires access only on “fair, reasonable and non-discriminatory” (FRAND) terms.

The broader standard follows from Section III’s reconstruction. If collecting and “hoarding” data is inherently unfair, forcing a gatekeeper to share those data on equivalent terms corrects the imbalance. A narrower interpretation could have required meaningful access without equalizing inputs. That approach might have helped rivals challenge Google Search while preserving some of Alphabet’s advantage. The Commission’s conception of fairness instead demands parity.

The preliminary findings extend that parity in three ways. First, Alphabet must anonymize the shared data, monitor third parties’ compliance with privacy restrictions, and enforce retention and encryption requirements (paras. 35–50). The gatekeeper thus bears the cost and risk of policing the rivals the obligation seeks to empower.

Second, Alphabet may charge only the incremental costs of preparing, storing, and distributing the dataset, plus a return capped at its weighted average cost of capital. It cannot recover the costs of collecting the data. Small and medium-sized access seekers must receive the data below cost even when Alphabet demonstrates that it cannot otherwise cover its costs efficiently (paras. 71–78).

Third, the post-FRAND standard requires prices that allow beneficiaries to optimize their services “without Alphabet maintaining or obtaining an unfair advantage.” That test does not explain whether a rival’s inefficiency may account for a continuing performance gap or whether any remaining Alphabet advantage is presumptively unfair. The latter interpretation would make the obligation perpetual and impossible to satisfy conclusively.

The Apple interoperability decision under Article 6(7) separates contestability even more clearly from competitive effects.[69] The Commission repeatedly states:

[It] is not required to demonstrate that each of the measures are necessary to enable contestability, i.e. that (some) third parties are able to provide a “competitive offering” or an “alternative solution” which is enough for achieving contestability.

Such an inquiry, the Commission concludes, “would re-import the effects analysis, which the legislator explicitly rejected” (paras. 77, 330).

The Commission neither measures contestability before intervening nor tests it afterward. Contestability instead operates as a legal category that follows from the Act’s assumptions. The decision then translates that category into strict parity. Article 6(7) requires access to the “same,” rather than merely “similar,” features Apple uses (para. 61). Parity extends across the user experience, including ease of use, setup, data-transmission speed, and energy consumption.

Rivals’ mere ability to compete does not suffice. They must compete using identical inputs. The reason a third party cannot offer a competitive product is irrelevant, whether the obstacle arises from connectivity limitations or the product’s inferior quality.

Interoperability is a “results-based concept” for which the gatekeeper bears sole responsibility (para. 92). “Allowing” interoperability means proactively enabling it. Apple may have to develop application programming interfaces (APIs), which allow software systems to communicate, test for bugs, and provide third parties with “adequate and timely assistance” (para. 80). Its solution may impose no “undue costs—including development costs” on beneficiaries (para. 81). Apple must provide all of this without charge.

Apple must also build interoperability into new and updated features from inception—that is, by design—and release it to third parties at launch. This obligation applies even when third parties do not know that Apple is developing the feature (paras. 71, 146, fn. 208).

Two aspects of the decision are especially revealing. First, although the Commission rejects effects analysis, it repeatedly asserts beneficial effects. Equal access “fosters contestability and fairness,” which “in turn, improve[s] the innovation potential of the wider online platform economy.” The Commission also asserts that “more contestability for complementary or supporting services or hardware will bring more contestability for operating systems” and that mandated sharing “will also indirectly increase Apple’s incentives to innovate within iOS itself” (paras. 95–98).

The decision offers no evidence for those propositions. The Act’s preamble already assumes the benefits, so the Commission sees no need to demonstrate them. Those assumed effects then exclude contrary evidence. Examining whether forced sharing might weaken Apple’s incentives to innovate “would introduce the requirement to investigate on a case-by-case basis the effects on competition of gatekeepers’ given conduct, which the legislator explicitly rejected” (para. 66).

The decision simultaneously declares that “[a] main goal of Article 6(7) . . . is to enable innovation by third parties” (para. 95). The source of innovation therefore matters more than its substance.

Second, the Commission concedes that full compliance cannot eliminate Apple’s advantage. Apple will retain an “intrinsic advantage” because “only Apple decides which iOS features are being prioritized, planned and developed” (para. 74). A fully level playing field is therefore unattainable. Apple will always manage its own operating system, and that unavoidable advantage can justify continuing intervention.

The gatekeeper ultimately becomes both a competitor and the custodian of its rivals’ competitiveness. The DMA preserves the platform’s “dual role”—owner and participant, platform operator and commercial actor—but repurposes it. Recast as a quasi-public utility, the gatekeeper must exercise control over its platform for third parties’ benefit and at its own expense.

E. The First DMA Review: Redistribution as Success

The Commission’s first statutory report offers its own assessment of what two years of DMA enforcement have achieved.[70] It also provides a revealing account of how the Commission applies fairness and contestability. Three themes stand out.

First, the report defines “competition on the merits” in a way that discounts gatekeepers’ competitive advantages. It states that Article 6(5) will “curb gatekeepers’ power from vertical integration, promoting merit-based competition and innovation” (p. 24). The implication is that vertical integration represents a source of exclusion rather than competition on the merits.

The report applies the same logic to data. Article 6(2) allows market participants “to compete on the merits” by preventing gatekeepers from using business-user data to compete against those users (p. 11). The provision targets that use precisely because the data can improve gatekeepers’ offerings and increase competitive pressure. Competition by gatekeepers becomes unmeritorious “exploitation.”

This reasoning extends to artificial intelligence (AI). The report recounts trade associations’ requests that the Commission apply Article 6(5) to Google’s AI Overviews because their integration into search may entrench Alphabet’s position and reduce traffic to third parties (p. 26). It also notes that Article 6(2) “may apply to the provision of AI services if gatekeepers rely on non-public data of business users to compete with them” (p. 12).

The report’s conception of competition and innovation is therefore source-specific. It favors competitive pressure from third parties while discounting pressure from gatekeepers. This approach echoes the Apple specification proceeding, which described a principal goal of Article 6(7) as enabling innovation by third parties.

Contestability likewise means “strengthen[ing] the ability of alternative search engines to compete” (p. 15) by improving their position relative to gatekeepers. That relative improvement counts as progress even if it worsens gatekeepers’ products or reduces the overall quality of available offerings. The report accordingly treats greater uptake of services offered by firms other than gatekeepers as evidence of success (pp. 6, 14). Greater uptake of gatekeeper services implies failure.

The report identifies products such as Firefox, Aloha, Opera, and Vivaldi as beneficiaries because they are “smaller,” without defining smallness beyond their size relative to a gatekeeper (p. 14). It even counts Alphabet’s decision to permit account registration without an email address as progress because it “further reduc[es] user dependency on Gmail” (p. 23). That change does not create competition with Gmail as an email service. It simply reduces the number of Gmail users.

Third-party access, uptake, and presence also substitute for evidence of market outcomes. The availability of third-party alarm-clock apps on iOS is presumed to increase innovation without examining their quality or consumer demand (p. 17). The report similarly treats the launch of interoperable messaging services as confirmation of demand for interoperability without supplying usage figures. At the time of writing, one of the two cited services had not launched in the European Union. The other had few downloads and an average App Store rating of 3 out of 5 stars.

The report thus resolves Section III’s input-output ambiguity in favor of inputs. It assumes that favorable outcomes will follow from access, interoperability, and third-party presence because the Act defines those inputs ex lege as fair and contestability-enhancing.

Second, the report exposes a tension among user choice, contestability, and consumer welfare. It describes Article 5(2) as “giving users back control over their data” through a “real and unbiased choice” (pp. 6, 8). It also presents the provision as a means of “limiting gatekeepers’ accumulation of personal data” (p. 7). The first objective can be satisfied regardless of what users choose. The second can be satisfied only if users choose against data combination.

The report correctly states that “Art 5(2) of the DMA does not require any target threshold of how users exercise their choice” (p. 8). That assurance sits uneasily with the report’s framing and the Act’s broader effort to reduce gatekeepers’ accumulation of a crucial competitive resource. The asymmetry remains: Data accumulation is desirable when Articles 6(9) and 6(10) facilitate it for third parties and undesirable when gatekeepers undertake it.

Third, the report casts gatekeepers as financiers of the DMA’s redistributive project. Article 6(10) requires them to provide commercially valuable data free of charge, even though Amazon charges for portability solutions outside the European Union (p. 10). Steering beyond initial customer acquisition must be free (p. 15). Gatekeepers must also fund alternative dispute-resolution mechanisms (p. 29).

The report applies the same approach to consumer profiling. Gatekeepers must disclose their profiling techniques to facilitate contestability by allowing rival platforms to distinguish themselves through stronger privacy guarantees (p. 39). The obligation requires gatekeepers to reveal information resembling business secrets without imposing a comparable duty on third parties.

The Commission goes further when compliance creates conflicts among business users. Suppliers—including hotels, airlines, and merchants—complained that Alphabet’s Article 6(5) compliance measures reduced their visibility in Google Search. The Commission responded that any solution must “consider the interests of all participants in the market, including merchants, train operators, hotels, airlines, as well as vertical search services and content creators” (pp. 25–26).

The DMA thus makes gatekeepers responsible for managing tradeoffs among the groups the Act seeks to benefit. Yet neither the Act nor the report provides a principle for resolving those conflicts.[71]

F. Joint DMA–GDPR Guidance: Choice as Data Redistribution

The final source is the draft joint guidance from the Commission and the European Data Protection Board (EDPB) on the relationship between the DMA and the General Data Protection Regulation (GDPR).[72] It offers the first joint institutional account of how the DMA’s goals interact with a neighboring and potentially competing regulatory regime.[73]

The Guidelines present the two regimes as harmonious. They describe them as “complementary in terms of goals and in terms of the protections provided.” They also posit a reinforcing cycle in which “[g]reater fairness in and contestability of digital markets lead to more choice for individuals, which in turn should increase incentives for gatekeepers and their business users to develop and implement data protection and privacy features.”

The Guidelines offer no evidence for that relationship or explain why greater contestability should improve privacy. More firms might instead have stronger incentives to monetize data to compensate for their lack of scale. The Guidelines then devote substantial attention to managing frictions that their harmony narrative obscures, including consent requirements for business-user data access under Article 6(10), anonymization under Article 6(11), and security exceptions under Article 7.

The Guidelines make the instrumental role of choice explicit. Article 5(2)’s specific-choice requirement “seeks to address the enhanced access to personal data that provides gatekeepers with potential advantages in terms of data accumulation, which in turn raises entry barriers and hinders contestability.” Offering that choice “is a means to address the accumulation of personal data by gatekeepers and the correlated erosion of market contestability.” Choice is therefore a mechanism for reducing gatekeepers’ data holdings.

The treatment of data portability reveals the same asymmetry. The Guidelines criticize data accumulation by gatekeepers while praising continuous portability under Article 6(9) for “addressing data accumulation by gatekeepers and providing a stable source of data for the improvement and creation of new services.” Data accumulation creates a structural harm when data remain with a gatekeeper and an innovation input when third parties receive them. The gatekeeper becomes, in the Guidelines’ words, a “stable source of data” for its rivals. This is Dialectical Fairness in practice.

The Guidelines also make an important concession. Under Article 5(2), gatekeepers must offer users who withhold consent a “less personalised but equivalent alternative” that is “not . . . different or of degraded quality.” The Guidelines do, however, permit degraded quality when “a degradation of quality is a direct consequence of the gatekeeper not being able to process such personal data.”

That exception recognizes that personal data can improve product quality and that restricting their use can make a service worse. This tension runs throughout the enforcement record. The DMA seeks to reduce gatekeepers’ data-driven advantages even though those same advantages may support product improvements.

The Guidelines complicate the analysis in one respect. They frame choice as protecting end users’ informational self-determination, meaning their control over personal data. That framing gives consumers’ interests independent weight that the First DMA Report rarely affords them. The redistributive flows are also conditional. Article 6(10) requires user consent before business users receive personal data, while Article 6(11) requires anonymization before search data may be shared.

Yet if end-user self-determination controlled the analysis, the direction of users’ choices would not matter. The Guidelines’ own account of Article 5(2) shows that it does. The choice mechanism aims specifically to reduce gatekeepers’ data accumulation and the competitive advantages it creates.

The Guidelines therefore reinforce the pattern developed across the enforcement record. Fairness operates as a structural presumption against gatekeepers, while contestability calls for redistributing advantages toward their rivals. The official rhetoric assumes harmony among privacy, competition, and innovation even as the detailed rules must manage tradeoffs among them. Theoretical premises, rather than demonstrated effects, continue to determine the expected outcomes.

V. Reformulating Fairness and Contestability

Reading the Act alongside its enforcement record permits a clearer account of its twin goals. Fairness emerges as gatekeeper-financed structural redistribution, while contestability directs competitive opportunity toward rivals and business users. Both rest on unrebuttable premises, assume broad benefits despite unresolved tradeoffs, and use consumer choice to advance a single redistributive program.

A. Unrebuttable Premises and Assumed Benefits

The DMA rests heavily on theory, but the law insulates that theory from contradictory evidence in two ways.

First, the Act converts its assumptions into statutory premises that enforcement cannot revisit. As the Commission repeatedly explains, contestability is a legal category rather than an empirical condition open to dispute. It follows from the definition of a gatekeeper, the list of core platform services, and the recitals. If market developments disprove those assumptions, a gatekeeper cannot rebut them during enforcement. Only legislative change can correct them.

Second, the DMA assumes that achieving fairness and contestability benefits everyone except gatekeepers. It does not account for tradeoffs among different business users or between business users and consumers. Nor does it assign weight to gatekeepers’ compliance costs, which the Act treats as effectively limitless.

B. Fairness as Structural Redistribution

Fairness under the DMA means correcting a presumed structural imbalance between gatekeepers and those who depend on their core platform services. Properly reformulated, it has four features:

It is axiomatic. The Structural Unfairness Axiom presumes unfairness from gatekeeper status and establishes it conclusively at designation. The Commission does not infer unfairness from specific conduct, and gatekeepers cannot rebut it with evidence of superior products, consumer benefits, or economic efficiency. The designation decisions expressly exclude such evidence.

It is dialectical. Fairness operates by leveling down gatekeepers while leveling up business users and rivals. It does not seek to restore a neutral baseline or prevent discrete harms.

It is distributive rather than commutative. It redistributes value and opportunity among groups rather than policing the fairness of particular exchanges. Advantages rooted in scale, integration, or data accumulation may therefore be neutralized or reassigned even when they improve products and services.

It is gatekeeper-financed. Gatekeepers bear the costs of correction, including free access, free portability, compliance systems, and reconciling data utility with privacy. The Act implicitly assumes that this redistribution leaves everyone else no worse off.

C. Contestability as Competitive Redistribution

Contestability under the DMA means transferring competitive opportunity from gatekeepers to actual or potential rivals. It also has four features:

It is status-based. The Commission infers limited contestability from a gatekeeper’s scale and duration. Contestability attaches to firms rather than markets, and evidence of entry, growth, switching, or multihoming cannot rebut it. Those same facts would demonstrate contestability in its ordinary economic sense.

It is source-dependent. Competition, innovation, and data accumulation support contestability when they come from firms other than gatekeepers and undermine it when they come from gatekeepers. Under the Scale–Unfairness Paradox, gatekeeper innovation and growth reinforce rather than rebut concerns about weak contestability.

It is produced rather than diagnosed. Legal obligations seek to create contestability rather than identify an existing market condition. This reverses Baumol’s conception, under which the credible threat of entry makes a market contestable by disciplining incumbents.

It is input-measured. The Commission measures success through third-party access, uptake, and presence. It assumes rather than demonstrates the resulting effects on price, quality, and innovation.

D. One Redistributive Program

Under this reformulation, fairness and contestability form a single program. Fairness supplies the diagnosis—a structural imbalance presumed at designation. Contestability supplies the remedy—an ongoing redistribution of competitive opportunity toward rivals and business users. User choice serves both goals and gains value when it moves activity away from gatekeepers.

This relationship explains why the DMA can treat “fair and contestable” as a hendiadys, or two terms expressing one idea. Fairness and contestability represent distinct stages of the same redistributive logic.

VI. Implications of the DMA’s Unresolved Goals

Three implications follow from the gaps that remain. The DMA supplies no benchmark or endpoint for compliance, lacks an internal mechanism for detecting and correcting regulatory error, and risks exporting these features to jurisdictions that adopt its language without fully embracing its redistributive aims.

A. No Benchmark or Endpoint

The most fundamental gap is the absence of any benchmark for determining when fairness or contestability has gone far enough. The Structural Unfairness Axiom treats core platform-service markets as structurally distinct and presumes they were never fair. The DMA therefore has no prior equilibrium to restore and no historical baseline against which to measure progress.

The enforcement record supplies no substitute. Infringement decisions define both goals negatively through the elimination of prohibited practices, without identifying an endpoint. Specification proceedings elevate parity into the governing standard while conceding that full parity is unattainable because only the gatekeeper decides which features of its platform to develop. The First DMA Report measures success through third-party presence, which can always increase, rather than through any outcome-based metric.

The Bidirectional Unfairness Inference compounds the problem. Unsatisfactory outcomes after full structural compliance support an automatic inference that structural defects remain. Compliance therefore cannot protect a gatekeeper from further intervention, and the gatekeeper can never demonstrate conclusively that it has done enough.

That indeterminacy undercuts defenders’ claim that the DMA offers greater predictability than case-by-case antitrust enforcement.[74]

B. No Mechanism for Detecting or Correcting Error

The DMA’s architecture forecloses the usual mechanisms for detecting and correcting regulatory error. It excludes effects analysis and efficiency defenses by design. The legislature quantified the Act’s projected benefits once and converted them into statutory premises that enforcement need not verify.

The enforcement record shows the result. Claims about innovation move from the impact assessment to the preamble and then into individual decisions without gaining evidentiary support. The Act excludes by law the possibility that compliance could reduce product quality or weaken innovation incentives. A regime that treats incumbent growth as failure and discounts quality improvements by gatekeepers cannot distinguish successful regulation from harmful regulation. Both appear as a reduction in gatekeeper advantage.[75]

The framework can distinguish them only if reducing gatekeeper advantage is itself the objective, regardless of cost. Yet the tradeoffs the Act refuses to consider do not disappear. It resolves them silently and in advance against gatekeepers and, potentially, against consumers who value the integrated products and data-driven quality the Act targets.

C. Implications for DMA Importers

The reconstruction also matters beyond the European Union. The United Kingdom, Japan, Thailand, Vietnam, Türkiye, and Brazil have adopted or are considering DMA-style regimes, often copying the EU law’s operative language and twin goals.[76]

These jurisdictions would import an internally coherent redistributive program rather than a neutral toolkit for creating fair and contestable digital markets. That program includes a conclusive presumption of structural unfairness, a remedial strategy that levels down incumbents and levels up rivals, an instrumental conception of user choice, and a source-dependent, input-measured account of contestability with no defined endpoint.

Legislatures that want this program should adopt it knowingly. Those seeking consumer-oriented contestability or greater innovation regardless of its source should not assume the DMA will deliver those outcomes. The Act was not designed to do so.

VII. Conclusion: A Direction Without a Destination

This white paper argues that the DMA contains a coherent but unstated conception of fairness and contestability. Fairness seeks to correct a presumed structural imbalance. It is axiomatic, dialectical, distributive, and gatekeeper-financed. Contestability transfers competitive opportunity toward rivals. It is status-based, source-dependent, legally produced, and measured through inputs rather than market outcomes. User choice serves both goals by directing activity away from gatekeepers.

The enforcement record confirms this redistributive logic. The impact assessment embedded its predicted harms and benefits in the Act’s premises. Designation decisions treat contestability as a status inferred from scale. Infringement decisions identify prohibited conduct without defining an endpoint. Specification proceedings elevate parity while excluding effects analysis. The First DMA Report and joint DMA–GDPR guidance treat rival access, uptake, and data accumulation as progress, largely without testing effects on price, quality, innovation, or consumer welfare.

Two years of enforcement have therefore clarified the DMA’s direction but not its calibration. The Act identifies who should gain and who should lose without saying how much, for how long, or against what benchmark. It assumes away tradeoffs among business users, gives consumer welfare no controlling weight, and offers no internal mechanism for distinguishing beneficial intervention from harmful intervention. Full compliance cannot guarantee finality because disappointing outcomes can always support an inference that structural defects remain.

These omissions matter within the European Union and wherever lawmakers borrow the DMA’s language. Jurisdictions seeking consumer-oriented contestability or greater innovation from any source should recognize that the Act embodies a more specific redistributive program.

We now understand more clearly what fairness and contestability mean under the DMA. The Act tells markets which direction to move, but it still cannot say when they have arrived.

VIII. Glossary of Terms

This white paper uses several original terms to describe recurring patterns in the DMA’s text and enforcement. These terms are analytical tools rather than statutory categories. The following glossary defines each and explains its role in reconstructing the Act’s goals.

A. BUI (Bidirectional Unfairness Inference)

The Bidirectional Unfairness Inference (BUI) describes reasoning used without objective or agreed benchmarks for market performance. Under the BUI, outcomes that depart from a regulator’s preferred state support an inference that the market is structurally unfair. Conversely, asymmetries in rights, obligations, or constraints imply that the resulting prices, economic rents, market structure, or distribution must also be unfair. Market structure and outcomes thus serve as mutual evidence of unfairness.

The BUI is autopoietic, meaning self-generating, and it is self-referential. Neither unfairness nor the relevant counterfactual—the outcome expected under alternative conditions—has an independent definition. Structural asymmetries and deviations from regulatory expectations therefore validate each other. Structure explains the outcomes, while the outcomes confirm the diagnosis of an unfair structure.

Consider Apple’s 30% commission on in-app purchases. The fee is presumed excessive because Apple controls the in-app payment system and restricts steering, practices characterized as unfair. Its perceived excessiveness then serves as evidence that the market itself is structurally unfair. The diagnosis rests on reciprocal inferences from market design and observed outcomes rather than an external benchmark.

B. Dialectical Fairness

Dialectical Fairness describes fairness as a relational and directional process that simultaneously reduces gatekeepers’ advantages and improves the position of rivals or business users. Gatekeepers are made worse off so that others may be made better off. Fairness lies in the relative shift between those positions rather than in reaching a fixed equilibrium.

Constraining gatekeepers and empowering rivals are therefore linked. Any persistent gatekeeper advantage remains a potential target for correction.

Dialectical Fairness is dynamic and nonterminal, with no built-in endpoint. It supports continuing, asymmetric intervention to limit the renewal or expansion of gatekeeper advantages and rebalance competitive positions, rather than restoring a neutral baseline or remedying discrete harms.

C. DSUA (DMA Structural Unfairness Axiom)

The DMA Structural Unfairness Axiom (DSUA) is the Act’s foundational assumption that digital markets suffer from endemic unfairness and weak contestability. The premise is axiomatic because evidence cannot rebut it during enforcement. Changing it would require legislative amendment or repeal. The DMA’s obligations and interventions flow from—and find justification in—this axiom.

D. Source-Dependent Competition and Innovation

Source-Dependent Competition and Innovation describes a view of competitive activity whose value depends on its source. Conventional competition law protects the competitive process without favoring which firm succeeds. The DMA instead credits competition and innovation from rivals and entrants while discounting the same activity from gatekeepers, regardless of their respective effects on consumers.

Gatekeeper competition and innovation may therefore count against fairness and contestability. Under the Scale–Unfairness Paradox, no gatekeeper can innovate or grow its way out of the presumption of unfairness.

E. SUP (Scale–Unfairness Paradox)

The Scale–Unfairness Paradox (SUP) arises because greater scale can simultaneously improve a core platform service’s quality and attractiveness while increasing the risk of structural unfairness. The DMA treats the same features that make gatekeepers successful—network effects, data, and user growth—as sources of bottleneck power and weak contestability.

The SUP produces a second paradox. By constraining gatekeeper scale while helping rivals grow, the DMA may eventually turn successful rivals into gatekeepers subject to the same restrictions.

Paragraph 84 of the Impact Assessment illustrates this logic:

The bigger the platform, the stronger the indirect network effects, the larger the amount of data and the higher its quality. This leads to increased insight into user profiles and preferences, allowing gatekeepers to offer them more personalized services and advertisements, thus attracting even more users and reinforcing consumer lock-in, single-homing and rendering switching to alternative platforms more difficult.

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

[2] See, e.g., Giuseppe Colangelo, In Fairness We (Should Not) Trust: The Duplicity of the EU Competition Policy Mantra in Digital Markets, 68 Antitrust Bull. 618 (2023).

[3] See Nicolas Petit, The Proposed Digital Markets Act (DMA): A Legal and Policy Review, 12 J. Eur. Competition L. & Prac. 529 (2021).

[4] On the Sherman Act as a “common-law statute” whose content courts developed through adjudication, see Leegin Creative Leather Products, Inc. v. PSKS, Inc., 551 U.S. 877, 899 (2007). On the incremental development of Articles 101 and 102 of the Treaty on the Functioning of the European Union, see Pablo Ibáñez Colomo, The Shaping of EU Competition Law (Cambridge Univ. Press 2018).

[5] See, e.g., Belle Beems, The DMA in the Broader Regulatory Landscape of the EU: An Institutional Perspective, 19 Eur. Competition J. 1, 27 (2023); 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. 201 (2025).

[6] See Digital Markets, Competition and Consumers Act 2024, c. 13 (UK); Act on Promotion of Competition for Specified Smartphone Software, Act No. 58 of 2024 (Japan); Projeto de Lei No. 4675/2025, Câmara dos Deputados (Braz.).

[7] See Manne, Radic & Auer, supra note 5 (arguing that the stated goals of DMA-style regulations diverge from their actual objectives).

[8] Giorgio Monti, The Digital Markets Act: Improving Its Institutional Design, 5 Eur. Competition & Regul. L. Rev. 90, 91–92 (2021).

[9] Richard Feasey & Giorgio Monti, DMA Process and Compliance, in Implementing the DMA: Substantive and Procedural Principles 94, 120 (Alexandre de Streel coord., Ctr. on Regul. in Eur. 2024) [hereinafter CERRE].

[10] Jacques Crémer et al., Fairness and Contestability in the Digital Markets Act, 40 Yale J. on Regul. 973, 975–77 (2023).

[11] Id. at 975–76; see also DMA, supra note 1, arts. 12, 18–19, recital 65.

[12] Friso Bostoen, Understanding the Digital Markets Act, 68 Antitrust Bull. 263 (2023).

[13] 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).

[14] Petit, supra note 3.

[15] Colangelo, supra note 2, at 620.

[16] Monti, supra note 8, at 91–92.

[17] Petit, supra note 3.

[18] Akman, supra note 13.

[19] See Bostoen, supra note 12 (noting that the DMA’s recitals connect its anticipated improvements to innovation and consumer benefits, reinforcing the regime’s proximity to competition-policy benchmarks despite its ex ante structure).

[20] Monti, supra note 8.

[21] Rupprecht Podszun, Philipp Bongartz & Sarah Langenstein, Proposals on How to Improve the Digital Markets Act (working paper, Feb. 18, 2021), https://ssrn.com/abstract=3788571.

[22] Manne, Radic & Auer, supra note 5.

[23] Pierre Larouche & Alexandre de Streel, The European Digital Markets Act: A Revolution Grounded on Traditions, 12 J. Eur. Competition L. & Prac. 542 (2021).

[24] Manne, Radic & Auer, supra note 5, at 217 (describing a “Schrödinger’s [digital competition regulation]” scenario in which “the same substantive rules simultaneously are and are not competition law”).

[25] Crémer et al., supra note 10, at 1006–08.

[26] CERRE, supra note 9, at 14.

[27] Petit, supra note 3.

[28] Akman, supra note 13.

[29] Id.

[30] Manne, Radic & Auer, supra note 5.

[31] Giuseppe Colangelo & Alba Ribera Martínez, The Metrics of the DMA’s Success, 16 Eur. J. Risk Regul. 1017 (2025).

[32] Oles Andriychuk, Do DMA Obligations for Gatekeepers Create Entitlements for Business Users?, 11 J. Antitrust Enf’t 123, 123–32 (2023) (characterizing the DMA as “punitive”).

[33] DMA, supra note 1, recital 79 & arts. 8(1), 12, 18–19 & 53.

[34] DMA, supra note 1, recital 33.

[35] Id. recital 62.

[36] See id. recitals 2, 13, 20 & 32–33.

[37] The Act likewise does not distinguish companies whose dependence on a CPS resulted from a lack of foresight or diversification from those whose dependence was not self-inflicted.

[38] See DMA, supra note 1, recitals 2–5, 13 & 32–33 (attributing imbalances to features such as extreme economies of scale, strong network effects, data-driven advantages, lock-in, and a lack of multihoming).

[39] Id. art. 3(2).

[40] See id. recitals 15 & 17 (identifying monetization potential and access to capital markets as indicia of gatekeeper status); id. recital 25 (identifying high relative growth as an indicator of entrenchment).

[41] Id. recital 36.

[42] Id. recital 72 (emphasis added).

[43] Id. recital 46 (asserting further that it is unfair for a gatekeeper to benefit from its “dual role” as both provider and user of a core platform service).

[44] Id. arts. 6(5), (7) & (10).

[45] Id. recital 70; see also id. recital 37.

[46] Id. art. 6(3) & recitals 49 & 63.

[47] Id. art. 5(3) & recital 39.

[48] See, e.g., Christophe Carugati, The Digital Markets Act Is About Enabling Rights, Not Obliging Changes in Market Conditions, Bruegel (Sept. 6, 2023), https://www.bruegel.org/analysis/digital-markets-act-about-enabling-rights-not-obliging-changes-market-conditions.

[49] DMA, supra note 1, recitals 5, 45 & 107; see also id. recitals 6–7 (linking unfair practices to “negative societal and economic implications” and the “proper functioning of the internal market”).

[50] Id. recitals 11 & 23; see also id. recital 5 (stating that the DMA does not address the forms of market power covered by Articles 101 and 102 TFEU).

[51] Id. recital 3.

[52] Id. recital 44.

[53] Id. recital 61 & art. 6(11).

[54] Id. art. 18(3).

[55] Id. recital 75.

[56] Id. recitals 53, 55 & 57; see also id. art. 6(5) (permitting rankings based on transparent, fair, and nondiscriminatory conditions).

[57] Id. recital 64.

[58] See id. recital 79 & arts. 8(8), 9(4), 12(2)(b), 18–19 & 53.

[59] Eur. Comm’n, Commission Staff Working Document: Impact Assessment Report Accompanying the Proposal for a Regulation of the European Parliament and of the Council on Contestable and Fair Markets in the Digital Sector (Digital Markets Act), SWD(2020) 363 final (Dec. 15, 2020). Parenthetical paragraph references in this subsection refer to this document unless otherwise indicated.

[60] See Manne, Radic & Auer, supra note 5; Lazar Radic & Dirk Auer, A Europe Fit for the Age of Startups: Rhetoric and Reality in the EU’s Digital Package (Int’l Ctr. for L. & Econ. 2025), https://laweconcenter.org/wp-content/uploads/2025/07/Is-a-Europe-fit-For-the-Digital-Age-also-Fit-for-the-Start-up-Age.pdf.

[61] Eur. Comm’n Regul. Scrutiny Bd., Opinion on the Proposal for a Regulation of the European Parliament and of the Council on Contestable and Fair Markets in the Digital Sector (Digital Markets Act), SEC(2020) 437 (Dec. 10, 2020).

[62] Harold Demsetz, Information and Efficiency: Another Viewpoint, 12 J.L. & Econ. 1 (1969).

[63] Eur. Comm’n Decisions of Sept. 5, 2023, designating Alphabet, Amazon, Apple, ByteDance, Meta, and Microsoft as gatekeepers under Article 3 of Regulation (EU) 2022/1925; Eur. Comm’n Decision of May 13, 2024, designating Booking.com as a gatekeeper under Article 3 of Regulation (EU) 2022/1925. Parenthetical references in this subsection refer to the respective designation decisions.

[64] See William J. Baumol, Contestable Markets: An Uprising in the Theory of Industry Structure, 72 Am. Econ. Rev. 1 (1982); William J. Baumol, John C. Panzar & Robert D. Willig, Contestable Markets and the Theory of Industry Structure (Harcourt Brace Jovanovich 1982).

[65] Eur. Comm’n Decision of Apr. 23, 2025, Case DMA.100109—Apple (noncompliance with Article 5(4) of Regulation (EU) 2022/1925) [hereinafter Apple Anti-Steering Decision].

[66] Eur. Comm’n Decision of Apr. 23, 2025, Case DMA.100055—Meta (noncompliance with Article 5(2) of Regulation (EU) 2022/1925); see also Regulation (EU) 2016/679 of the European Parliament and of the Council of Apr. 27, 2016 (General Data Protection Regulation), 2016 O.J. (L 119) 1 [hereinafter GDPR] (providing the consent standard incorporated into Article 5(2) of the DMA).

[67] Press Release, Eur. Comm’n, Commission Fines Google €890 Million for Breaches of the Digital Markets Act (July 23, 2026), https://digital-markets-act.ec.europa.eu/commission-fines-google-eur890-million-breaches-digital-markets-act-2026-07-23_en.

[68] Eur. Comm’n Preliminary Findings of Apr. 16, 2026, Case DMA.100209—Alphabet (specification proceedings under Article 6(11) of Regulation (EU) 2022/1925), https://ec.europa.eu/competition/digital_markets_act/cases/202616/DMA_100209_1986.pdf.

[69] Eur. Comm’n Decision of Mar. 19, 2025, Specifying the Measures to Be Adopted by Apple Pursuant to Article 6(7) of Regulation (EU) 2022/1925 (interoperability with connected physical devices).

[70] Eur. Comm’n, Report on the Review of Regulation (EU) 2022/1925 (Digital Markets Act), COM(2026) 178 final (Apr. 28, 2026). Parenthetical page references in this subsection refer to this document.

[71] The report does not address whether a compliance solution must leave every affected business user no worse off—a Pareto improvement—or may trade some users’ interests against others’ and, if so, under what principle. See also id. at 48 (acknowledging “the concerns raised by the media industry, including in relation to traffic losses”).

[72] Eur. Comm’n & Eur. Data Prot. Bd., Joint Guidelines on the Interplay Between the Digital Markets Act and the General Data Protection Regulation (version for public consultation, Oct. 9, 2025), https://www.edpb.europa.eu/public-consultations/joint-guidelines-on-the-interplay-between-the-digital-markets-act-and-the_en. Quotations in this subsection are from this document.

[73] See GDPR, supra note 66.

[74] See DMA, supra note 1, recital 79 (invoking predictability as the rationale for a “predefined standard”).

[75] On error-cost analysis in this setting, see Manne, Radic & Auer, supra note 5; Colangelo & Ribera Martínez, supra note 31 (arguing that identifying the DMA’s goals is a prerequisite to measuring its success).

[76] See supra note 6. On the incorporation of DMA-style provisions directly into national competition laws, see Manne, Radic & Auer, supra note 5.