Let’s get something straight first: EU policymaking is fundamentally administrative and precautionary, not adversarial or litigation driven like in the U.S. The Impact Assessment (IA) is not a neutral “jury” evaluating guilt or innocence, and neither is the commission( it’s the executive branch); it’s a structured tool used by the Commission to justify legislative proposals through documented evidence, stakeholder feedback, and legal precedent. It’s part of a multi institutional process where Parliament and Council act as political checks if the IA were merely bureaucratic rubberstamping, the DMA would’ve been enacted instantly. It wasn’t. The commission has zero power to pass legislation. It’s fully within the power of parliament( 720 elected representatives) and council( 27 elected state representatives)
And please read up on how it’s done in EU.
You are correct that Furman Review, Stigler report, ACCC inquiry and ECB/EIB surveys all catalogue rising concentration and complaints from business users. None of them recommend the rulebook the EU adopted. In other words, the IA took a bunch of data that says “something may be wrong” and then assumed its preferred fix is the right one. In fact, the Furman Review warned that over-broad remedies risked “deterring entry and innovation” (emphasis mine).
Evidence of a headache is not evidence that the EU's pill is the cure.
Citing a paper trail back to the 2015 only proves the Commission had been rehearsing the same hypothesis for years. Recycling earlier studies may show policy momentum, but momentum is not evidence that the DMA’s rules will fix the problem.
You’re arguing that identifying structural market failures doesn’t justify regulatory action unless the precise fix is proven in advance to increase innovation with quantifiable certainty. But that’s not how public policy works, especially in digital markets where counterfactuals are notoriously hard to model. And the goal is a healthy competitive environment leads to innovation.
Yes, the Furman Review cautioned against overbroad remedies. It also recommended an ex ante code of conduct for digital platforms, interoperability mandates, and data access obligations remarkably similar to the DMA’s core obligations. Same goes for the ACCC’s recommendations in Australia, which included anti-self-preferencing and stronger merger oversight. You’re cherrypicking their caveats while ignoring their core conclusions.
Sound familiar? That’s basically the DMA’s core obligations. Same with the ACCC in Australia. You’re citing their cautions while ignoring their main recommendations.
You claim the DMA’s IA assumes the policy works because “something must be done.” But that’s a mischaracterization. The IA documents specific bottlenecks( that’s what a gatekeeper is) with control over data, app stores, OS-level integration and justifies each core obligation in response to those mechanisms. It doesn’t assume the DMA works by default; it evaluates a menu of options and discards others (including pure antitrust) as insufficiently timely or effective.
So yes, those reports were evidence of a systemic “headache.”
I'd also point out this is another reason the IA's neutrality is in serious question. By the time it had arrived, the policy train had already left the station, so the it was validating a pre-chosen destination, not acting as a fair and balanced look at all options.
Framing this as “regulation good, innovation bad” misses the point. The real question is "will the DMA increase or decrease innovation?" Paragraph 286 of the IA merely asserts the DMA will fix the problem with no proof to back it up while conceding the benefits are “complex to quantify.”
The studies the IA cites market power (higher mark-ups, rising concentration) but don't show forbidding self-preferencing or anti-steering causes more R&D or start-up entry.
You accuse the IA of “not proving” that forbidding anti-steering or self preferencing will increase R&D. But innovation isn’t only about R&D spend it’s about diffusion, entry, and recombination. And again, the Paragraph doesn’t “merely assert”; it explicitly references studies ( OECD 2018, IMF 2019) that link lower market conmtenstability with declining innovation output.
Claiming the Commission had its mind made up ignores how EU law works. If the policy was predetermined, why did Parliament and Council spend over a year negotiating it? Why were stakeholders consulted multiple times after 2020? This wasn’t regulatory theater. it was the administrative process: propose, assess, amend, legislate. That’s how co-decision works. And The IA isn’t a blankslate scientific experiment, It evaluates a preferred policy option, as that’s its function. And the RSB initially raised issues with the baseline. That’s not a scandal but it’s job description according to good legislation guidelines. The final version was approved after revision, exactly how the process is supposed to work.
You’re demanding an RCT level proof that banning self preferential boosts innovation. That’s not how structural regulation works and it never has. The IA identifies bottlenecks: barriers to market entry, limited interoperability, fee extraction. The removal of those barriers logically increases available capital, platform independence, and R&D allocation downstream.
If you believe gatekeeper lockin stimulates innovation, your burden is to prove it, because the available data doesn’t seem to support it.
From Furnans report:
Many stakeholders also submitted evidence to the Panel arguing that such a trend can be harmful to competition. In particular, that this strategy can create barriers to entry, as new firms need to offer an entire ecosystem by competing across a range of related markets to survive. Others argued that where multi-market firms hold a strategic gateway position in one market, they are then able to leverage that position in adjacent markets, give themselves an advantage through self-preferencing, and obtain an unfair advantage through holding of data and imitation of rivals’ innovations.
Sounds like a few people here favorite argument. Just make a new phone and an entire ecosystem just because you wanted to provide a better messaging or browser than the stock…
Europe’s last big tech rule, GDPR, has been linked to a 20-30% fall in VC deals and lower early-stage investment relative to the U.S. That is a massive datapoint that points to my belief that sweeping, one-size rules aimed at a handful of big companies can chill the broader ecosystem. Until the Commission can model the effects (or even better, pilot them) the claim that the DMA will lift “overall innovation” is just wishful thinking.
Blaming the pop-up plague on a missing “technical standard” is confusing cause and effect. GDPR created a legal duty to obtain consent, but the EU launched that rule without issuing a workable UX template and have spent the seven years twiddling their thumbs. The EU imposed a broad obligation, left the implementation details to “sort themselves out,” and have been chasing the fallout ever since. That sounds awfully familiar to their MO with the DMA.
That’s simply incorrect. The cookiebanner chaos is rooted in the ePrivacy Directive, not GDPR. Directives set objectives but leave design and enforcement to Member States so national regulators, not EU, decided how to implement consent flows. The ePrivacy Regulation (which would have harmonized interfaces) has been stalled by the Council, not Brussels.
GDPR never dictated UX it established a horizontal data-protection framework for every actor handling personal data, from Facebook but also your dentist, your school, your landlord or mechanic. It’s technology-neutral, defines specifies legal principles (lawfulness, purpose limitation, transparency), and deliberately avoids prescribing UX design. That was always the job of the proposed ePrivacy Regulation.
The GDPR is not a digital services regulation. It is a horizontal data protection framework that applies across all sectors online and offline, private and public, analog and digital. That’s why:
- It applies to Facebook and Google, but also to your local dentist, hospital, or car rental agency.
- It governs how data is processed (lawfulness, transparency, purpose limitation), not who processes it or in what kind of interface.
- The GDPR is technology-neutral by design. It doesn’t assume digital interfaces or prescribe implementation modalities.
- It’s a framework law, not a design spec. It defines rights and obligations
- The UX/consent interface issue falls under the proposed but still-stalled ePrivacy Regulation, which would have tackled interface-level consistency for online tracking (including cookie banners).
And the Commission couldn’t mandate a universal consent UI even if it wanted to.
The Council sought broader exemptions for businesses and law enforcement, while Parliament pushed for stricter privacy safeguards.
But guess what? Member States blocked it, not the Commission. Blame the Council, not Brussels.
And just because some VC portfolios tanked after GDPR doesn’t mean the law was bad. It could just mean their bets relied on exploitative data models that couldn’t survive basic consent. Either way, that critique doesn’t generalize to the DMA, which targets 10 entrenched gatekeepers,not the entire economy.
Yes or no, had the EU not forced Microsoft to give Crowdstrike kernel access, would the outage have happened?
Again the EU court did not “force” kernel access. That’s Microsoft PR spin. The requirement was nondiscriminatory API access under competition law.
Commission made it very clear: Microsoft is free to secure its systems however it wants, as long as it doesn’t sabotage competitors.
Has they implemented it better it wouldn’t have happened either.
"All the good things that happen are because of the regulation, but the bad stuff has nothing to do with the regulation at all" isn't a serious argument.
Blaming a security lapse on that is a failure of design and testing, not a legal mandate.
That’s not regulatory overreach… it’s basic competition law written before Microsoft was founded.
If the IA was actually an unbiased report as you claim, it wouldn't have originally been rejected because, in part, the baseline was “non-neutral” and said it “offers limited genuine choice to decision-makers.” Then, to fix it, they added an “Option 0” paragraph that predicts regulatory fragmentation and entrenched gatekeeper power, then immediately rules it out. That’s box-ticking, not a serious cost-benefit analysis. Which is why the RSB still said there were "significant shortcomings" in evidence, even while approving it. Their words, not mine.
Just because the report's authors went back after the fact to check a box doesn't change the fact that it WAS WRITTEN TO JUSTIFY THE DMA, not determine whether something like the DMA was needed. It's not me saying that, it's the EU's board that, despite your protestations, is made up of senior Commission officials. They said it was "non neutral". It is hundreds of pages that beg the question without properly analyzing the costs of the rules it was putting into place and imagining benefits without any proof whatsoever that the benefits would arise from the rules. That's not "genuine mixed-methods analysis" - it's cherry picking and making things up.}
And of course the RSB flagged early issues that’s its job. The IA was revised accordingly and resubmitted. You’re cherrypicking quotes from their first opinion while ignoring the final one, which approved the IA. Regulatory scrutiny isn’t proof of failure, it’s the process working.
And again:
neutrality is not the same as passivity. IAs are designed to assess the impact of proposals, not neutrally weigh all possible futures.
The Impact assessment isn’t written in a vacuum, but to examine the viability of different options against a set of policy objectives already adopted by co-legislators. The fact that the Commission had already acknowledged structural asymmetries in digital markets back in 2016 isn’t a flaw it’s institutional memory. You’re treating that as evidence of bias when it’s just responsible policymaking over time.
surferfb said:{It's not Teece & Kahwaty's job to write a version of the DMA for the Commission; it simply points out that the DMA’s Impact Assessment never connects documented market power and the specific, per-se prohibitions the law imposes. In a cost-benefit exercise the burden of proof sits with the regulator, not the critic: if the EU can't demonstrate that the obligations will do what they say, the proper response isn't “well, the critics didn’t supply their own spreadsheet.”
I agree the formal sequence is "propose 👉 review 👉 revise 👉 legislate", but when the same institution drafts the proposal, supplies the evidence, and participates in the trilogue that finalizes the text, it is hardly proof that the underlying analysis received a truly independent reality-check.}
That isn’t a simple “you say, I say” exercise it’s a mixed methods approach combining empirical studies, economic theory, stakeholder feedback, and scenario modeling. If you demand a row-by-row spreadsheet with predicted euro impacts for every Article, you’re holding the IA to a U.S style rulemaking standard (APA/OMB Circular A-4) that the EU never adopted. The Impact assessment has met its mandate under the Better Regulation Guidelines. it provides reasonable justification for each obligation, and it flags where quantification is infeasible rather than pretending false precision.
And the European Commission has a monopoly on legislative initiative in the EU. That’s how the system is designed under the treaties. Independence is ensured through the Council and Parliament, not through artificial separation between analysis and drafting. If your model of neutrality is U.S style independent agencies or court-driven rulemaking, you’re looking at the wrong system.
Again, Evidence of Market failure (which I would dispute, but giving it to you for purposes of argument) is not proof that the DMA's draconian regulations will fix anything. In fact, I'd argue they're going to make the problem worse and just act as a drag on innovation for everyone. Economically, the burden of proof for an ex-ante, per-se regime should be higher than for a case-by-case approach. And I'd argue they didn't even meet the burden of proof for a case-by-case approach.
I am not saying that stakeholder testimony is useless; the point is that self-selected complainants are evidence of grievance, not of the systemic effects of the cure.
You’re conceding for argument’s sake that there’s a market failure. entrenched dominance, tipping effects, user lock-in, suppressed innovation at the ecosystem level. Good. So what problem does the DMA supposedly “make worse”? Because without defining that, your claim has no traction and no meaning.
If your answer is “innovation,” then back it up:
- Innovation for whom? Incumbents who’ve scaled by leveraging cross-service bundling and self preferences?
- Or innovation from competitors, rivals, startups that has to face the gatekeeper bottlenecks?
The IA and accompanying studies (see Annex 11, Section 7.3) make it pretty clear the innovation drag is on the non-gatekeepers, not the gatekeepers themselves. If you want to argue that removing structural barriers, like exclusive defaults, ranking manipulation, or anti-steering clauses somehow reduces innovation, you’ll need a theory that explains how gatekeeper control stimulates competition at the margins. Where is that theory? Where is that evidence?
- Which of those harms is being intensified?
- By what mechanism?
- For whom?
Because vague appeals to “innovation drag” aren’t analysis, but again just vibes. Vague vibes aren’t a rebuttal to documented market power and persistent enforcement failures under Articles 101/102.
As for your “burden of proof” argument:
Yes, ex-ante regimes carry a higher upfront burden but not an impossible one. The DMA doesn’t need 100% certainty that every Article will optimize innovation output in every market. It needs to show that:
- Certain entrenched practices ( tying, anti-steering, default preinstallation) routinely distort market access.
- Remedies that restore contestability, even if blunt, are proportionate to the structural harm.
- The administrative cost is lower than the social and economic cost of delay (as we saw post-Android and Shopping)
The DMA clears that bar as it’s built on article 101 and 102 though decades of legal precedent. If you think it doesn’t, then you need to spell out:
- What alternative mechanism would work better?
- Why the case-by-case model: which was already tried and stalled for over two decade is more effective despite being slower, reactive, and often toothless?
The per-se prohibitions mirror problems already identified in case-by-case rulings that took years and had limited effect. The point isn’t to rerun the litigation, it’s to preempt known bottlenecks before they harden further.
If you think this still doesn’t meet a “case-by-case” threshold, show your standard. Because “I don’t like it” isn’t one.
The 2020 public consultation drew fewer than 600 business-user responses across the entire EU, and over 70% came from firms that already had an open dispute with at least one “gatekeeper.” That is not a representative sample of the millions of developers, merchants and advertisers that actually trade on the platforms. End-user input was tiny (under 200 replies) and largely ignored in the IA’s option scoring, while entire downstream sectors like hotels, publishers, consumer-goods brands are almost completely absent. That is in part why the European Parliamentary Research Service said the IA “does not provide policy-makers with a neutral baseline and offers limited genuine choice.”
You’re cherry-picking one early consultation to imply the DMA was rushed through based on a handful of noisy complainants. That doesn’t hold up.
The 2020 consultation and inception impact assessment were just early stages of a multi-phase evidence-gathering process. Since then, the Commission ran:
- A full legislative proposal consultation (2020-2021)
- Stakeholder feedback during 2022,2023,2024 and now
- Feedback from stakeholders with experts before mandating specific actions.
- Hearings and workshops in Parliament,
- Feedback through trilogues involving national competition authorities, trade associations, and even the gatekeepers themselves,
- Ongoing input via the High-Level Group and enforcement consultations.
So no, this wasn’t a policy scribbled on Spotify’s napkin.
That grievance was backed by antitrust precedent and confirmed by the Commission. You’re cherry-picking again. Spotify wasn’t the only voice; app developers, publishers, and advertisers raised identical concerns. The Commission didn’t write DMA obligations based on Spotify’s press releases it is as inhale stated based on decades of legal procedures and years of enforcement experience under Articles 101 and 102, plus documented platform behavior across markets.
Spotify telling regulators “Apple’s 30% fee hurts me” is a data point; it does not prove that the DMA will raise innovation rule-by-rule cost-benefit analysis, which again, the EU concedes it could not produce.
That’s a legal data point. And it’s related to the market competition. And again not supposed to
You keep insisting the IA failed because it didn’t operate like a U.S. court brief. But that’s a category error. The EU Commission is not a judge or a court, it’s a policymaking body operating under administrative law. The DMA isn’t a sentence but policy proposal that was made, built on the best available evidence, wide consultation, and public interest logic.
If you want to argue that status quo platform dominance is net good, or that innovation flourishes under gatekeeper control, that’s your burden. But don’t claim the EU failed to justify its action just because it didn’t follow your preferred epistemology. The DMA was passed after extensive scrutiny and democratic input. You’re free to critique it, but don’t pretend the process was illegitimate just because it didn’t follow U.S. policy procedures. They aren’t written with the same intention or according to the same guidelines.
Understanding the DMA debate requires understanding how the EU actually works, not assuming it mirrors U.S. government structure. Here’s a quick breakdown:
- EU Directives ≠ binding federal law. They’re framework instructions to EU Member States, who must implement them via national legislation (e.g. the ePrivacy Directive). Think: federal guidelines with state-level implementation leeway.
- EU Regulations = directly binding law across all EU Member States without national transposition. The DMA is a Regulation—it’s enforced as-is, uniformly, like a federal statute applied across all states.
- EU Law = a hierarchy. Regulations > Directives > National Law, and subject to interpretation by the Court of Justice of the EU (CJEU). Supremacy is well-established.
- European Commission = executive + civil service. It proposes legislation, enforces EU law, and represents the EU internationally. Think of it like the U.S. Executive Branch + Federal Agencies, but with no legislative power by itself.
- European Parliament = directly elected chamber, akin to the U.S. House of Representatives, though elected by proportional representation. It co-legislates all EU laws.
- Council of the EU (aka Council of Ministers) = the Member States’ national governments, with rotating presidencies. Comparable to a hybrid of the U.S. Senate + state governors’ conference—co-legislator with Parliament.
- European Council = heads of state/government (e.g. Macron, Scholz), sets high-level political direction. Not a legislative body.
- Administrative vs Legislative: The Commission is administrative—it prepares, justifies, and implements. It doesn’t pass laws. Laws only pass via trilateral (three-party) negotiation between Commission, Council, and Parliament—called a trilogue.
- Separation of Powers vs Shared Powers: The U.S. model separates executive, legislative, and judicial functions strictly. The EU model features shared responsibilities and institutional checks. Think co-decision, not strict opposition.
- Trilogues = informal 3-way legislative negotiations to finalize texts: Commission (proposer), Parliament (amender), Council (national ministers). No equivalent in U.S. politics—closest analogy is bicameral reconciliation with executive guidance.
