Training national judges for digital competition law: the DMA, private enforcement, and the infrastructure of judicial capacity

The EU has adopted a dense digitalization ‘acquis’, including the Digital Markets Act (DMA), the Digital Services Act (DSA), Data Act, and Artificial Intelligence (AI) Act. Yet these regimes also create a ‘judicial’ order.1 National courts review National Competition Authorities' (NCAs) decisions, hear injunction and damages actions, manage disclosure and confidentiality, apply the principles of equivalence and effectiveness, and decide whether to refer questions to the Court of Justice. In digital markets, national courts also hear claims framed in contract, tort, or unfair competition law; such disputes may be shaped in substance by platform power, DMA obligations, or competition theories of harm. Digital competition disputes rarely present as clean doctrinal puzzles. They involve ecosystems and network effects, data advantages, switching costs, app-store rules, ranking and recommendation systems, advertising technologies, interoperability, and design choices. They also intensify familiar procedural issues: information asymmetry, protection of trade secrets and personal data, expert evidence, counterfactual reasoning, and quantification of loss. Courts asked to assess abuse, gatekeeper compliance, or traffic-based harm need enough economic and technical literacy to control evidence and provide reviewable reasons. This article advances three interrelated propositions about the future of judicial training in Europe, in light of the expansion of EU digital competition law and the expected growing relevance of disputes in national administrative and civil courts. First, competition law judicial training must be updated to reflect the digitalization acquis without treating all judges as a single audience. Secondly, administrative and civil judges require a shared substantive foundation but different advanced tracks. Thirdly, DMA literacy should be integrated into competition law training because DMA obligations and compliance measures will often frame the background (ie facts, evidence, and remedies) even when the formal cause of action remains Article 101 or 102 TFEU or national private law. Since Regulation 1/2003,2 EU competition law has generated a training ecosystem, and the Commission’s 2016 study remains a key baseline. The main lesson is institutional: competition adjudication remains dispersed across most EU Member States.3 Some judges review NCA decisions; some face quasi-criminal fines; some hear follow-on or standalone damages actions; and many likely private-enforcement judges are generalist civil/commercial judges who encounter competition disputes infrequently. Generic training for ‘competition judges’ therefore risks mis-targeting. A small cohort of specialized administrative/appellate judges can absorb advanced cross-border seminars; a larger and more diffuse cohort needs practical, repeatable, and national language materials. A second lesson is that access and language are part of capacity. Even motivated judges struggle to reuse English-only slides or to locate comparable national judgments, especially when they write in national languages and apply domestic procedure. Training, therefore, needs ‘durable’ artefacts (eg checklists, model orders, short bench notes, and translated summaries) plus easy access to peer decisions. Without visibility of national case law, training becomes episodic rather than cumulative. Five judicial training providers illustrate complementary strengths. The European Judicial Training Network (EJTN)4 is the natural starting point for the institutional mapping. Its strength is not narrow doctrinal specialization but judicial legitimacy, scale, peer exchange, and training methodology. It is the European forum through which judges meet other judges, develop trust, and learn how judicial cooperation operates in practice. In digital competition law, that role is indispensable because national judges increasingly confront questions that recur across Member States: platform access, ranking, self-preferencing, disclosure of technical data, jurisdiction clauses, and the relationship between private proceedings and Commission enforcement. EJTN can diffuse knowledge, organize exchanges, and support training-of-trainers. However, it is not designed, by itself, to provide highly specialized modules on platform economics, algorithms, APIs, DMA-compliant design, or digital remedies. Those subjects require collaboration with more specialized providers.5 The Academy of European Law (ERA)6 occupies a different position. It has long experience in structured EU law training, including competition law training for judges. Its comparative advantage lies in doctrinal organization, repeatable course formats, multilingual accessibility, and the capacity to translate complex EU developments into professional training. Its limitation is that it is not positioned to deliver research-led or residential advanced training; that role falls to institutions such as the EUI. The European Networking and Training for National Competition Enforcers organized by the European University Institute (EUI) provides advanced training specialization rather than mass delivery. Being a research-led institution, the EUI can connect judges with academics, economists, computer scientists, regulators, and practitioners; it can produce durable outputs such as case notes, benchbooks, comparative maps, and policy reports; and it can use residential training to create a high-trust environment for judicial dialogue. The OECD-GVH Regional Centre for Competition in Budapest (RCC),7 a joint initiative of the Organisation for Economic Cooperation and Development (OECD) and the Hungarian Competition Authority (GVH) established in 2005, occupies a hybrid position between regional capacity-building and EU judicial training. Its primary mandate covers competition authorities in South-Eastern Europe and Central Asia. The RCC has also developed a distinct judicial training, the Competition Lab for Judges, co-funded by the European Commission, which offers national judges from EU Member States structured, case-based training on specific competition law topics through hypothetical case exercises and peer dialogue in a residential setting. Yet, similarly to ERA, the RCC is not designed for sustained advanced specialization in digital competition law, and its judicial stream, while high-quality, remains episodic relative to the scale of demand. Finally, national judicial academies remain essential.8 Most judicial work takes place in national languages, under national procedural rules and within national institutional cultures. National academies are therefore best placed to localize training on civil procedure, administrative review, disclosure, confidentiality, costs, appeals, limitation, and remedies. They also understand the structure of domestic courts and the level of specialization available. However, most national academies cannot efficiently develop, on their own, advanced modules on DMA private enforcement, platform economics, or algorithmic evidence. The appropriate model is therefore not centralization but ‘division of labour’: EJTN for judicial networks and diffusion; ERA for structured EU law modules; EUI for advanced research-led laboratories; and national academies for localization and repeat delivery.9 In light of this mapping, the EU already has institutions capable of supporting judicial capacity, but their activities must be better aligned with the demands of digital competition adjudication. The missing element is a sustained training infrastructure that links substantive law, economics, technology, procedure, and transparency of case law. That infrastructure should not replace existing programmes. It should connect them, deliver durable outputs, and ensure that training is designed around judicial functions rather than generic subject labels. Administrative law judges control the legality of public enforcement: competence, procedure, fact-finding, legal characterization, reasoning, and the lawfulness of sanctions and remedies. In digital cases, ‘standard of review’ is not an academic debate. If courts defer too much, complexity shields error; if courts redo the administration, review becomes regulation. The case law of the European Court of Justice provides the constitutional anchor: courts must verify the accuracy, reliability, and consistency of evidence and whether all relevant information was considered (‘Tetra Laval’), and effective judicial protection requires a review capable of assessing law and fact (‘Chalkor’).10 Training should translate those standards into judicial craft: how to reconstruct the authority’s theory of harm and counterfactual; how to test whether the record supports foreclosure capability and effects; how to handle contested expert evidence; and how to review proportionality without substituting the court’s own economic assessment. Leading recent cases illustrate the stakes: evidence-based engagement in ‘Intel’; the need to address alternative explanations in ‘Servizio Elettrico Nazionale’; platform architecture and visibility in ‘Google Shopping’; and feasibility/justification in interoperability disputes such as ‘Android Auto’.11 Administrative review also requires procedural and remedial literacy suited to digital files: voluminous records, internal documents, datasets, technical documentation, confidentiality claims, and privilege. Remedies, especially behavioural remedies involving interoperability, ranking, data access, or default settings, raise feasibility and institutional competence concerns. Judges need structured tools for reviewing remedies (link to infringement, precision, proportionality, monitoring implications) as well as sanctions (gravity, duration, due process). A useful training technique is ‘remedy drafting by constraints’: judges are asked to rewrite a remedy so that it is (i) tied to a proven theory of harm; (ii) expressed in verifiable obligations; (iii) limited by proportionality; and (iv) compatible with the institutional role of courts vis-à-vis regulators. Civil judges see digital competition issues through damages, injunctions, disclosure applications, interim measures, contract invalidity, or collective actions. Their task is not to police administrative discretion but to adjudicate private disputes while ensuring the effective protection of EU-law rights. ‘Courage’ and ‘Manfredi’ establish the right to compensation as an implication of effectiveness, within national procedural autonomy constrained by equivalence and effectiveness.12 Private enforcement training should be ‘procedure-heavy’. First, jurisdiction and forum clauses: platforms’ standard terms often attempt to steer litigation. Brussels Ia enables exclusive jurisdiction clauses in many B2B settings, but competition law-based claims may proceed under the tort head; ‘Wikingerhof’ is particularly relevant because it allows an abuse-of-dominance claim to fall under Article 7(2) even where a contract exists, if the claim relies on an obligation independent of the contract.13 Secondly, interim measures and case management. Digital harms can be time-sensitive (ranking demotion, delisting, blocked Application Programming Interface (API) access). Civil judges need training on how to structure urgency assessments, balance of convenience and proportionality, and how to craft interim injunctions that are technically implementable and reviewable. This includes managing ‘design’ remedies (eg temporary access, neutral ranking parameters, interoperability access) without freezing innovation or imposing broad product redesign. Thirdly, disclosure and confidentiality. The Damages Directive requires proportionate disclosure while protecting leniency and settlement materials; pre-Directive case law (‘Pfleiderer’, ‘Donau Chemie’)14 remains instructive on balancing private claims with public enforcement. In digital markets, this balancing applies to technical material (ranking logs, API documentation, experiments), trade secrets, and personal data. Civil judges need practical ‘toolkit’ instruments: staged disclosure, targeted keywords, confidentiality rings, expert-only access, redactions plus narrative summaries, and reasoned proportionality tests to avoid both fishing expeditions and evidence denial. Fourthly, liability, pass-on, and binding effects. The EU concept of ‘undertaking’ shapes private enforcement, including succession and intra-group attribution (‘Skanska’, ‘Sumal’). Regulation 1/2003 and the Damages Directive constrain contradictory outcomes and strengthen follow-on actions via binding effects; ‘Otis’ illustrates the weight of Commission findings in national litigation.15 Training should also address pass-on and standing questions in platform contexts (eg business users vs end users), including how to manage expert evidence on incidence and avoid double recovery within domestic procedural categories. Finally, causation and quantum. Digital harms often involve diverted traffic, reduced visibility, impaired access, or exclusion from a platform rather than higher prices. EU law limits national rules that make compensation practically impossible (‘Kone’) and permits estimation where quantification is excessively difficult once harm is established (Damages Directive Article 17(1); ‘Tráficos Manuel Ferrer’).16 Follow-on ‘Google Shopping’ litigation exemplifies the need for counterfactual reasoning about visibility, traffic, and monetization.17 The DMA reinforces this civil track. Even absent an explicit general damages provision, parties will invoke DMA obligations directly or indirectly, and Commission DMA decisions or compliance measures may structure evidence and remedies in private disputes. Civil judges therefore need functional DMA literacy: what obligations require in practice, what Commission procedures produce (designations, non-compliance findings, compliance reports), and how cooperation mechanisms affect stays, evidence, and injunctive relief. The DMA changes what competition law training must address. It highlights platform ecosystems and compliance design, and increases interaction among public enforcement, regulatory dialogue, and private litigation. Judges need DMA literacy to understand context while avoiding doctrinal conflation, that is, treating the DMA as ‘Article 102 by another name’ or importing DMA obligations mechanically into competition analysis.18 Four cross-cutting competencies should structure advanced training modules on the DMA: (1) Platform literacy: judges should understand multi-sidedness, network effects, switching costs, default bias, ecosystem leverage, and data accumulation, because these concepts affect market definition, dominance, causation, and remedies. Training should show how to reason from these concepts without turning judgments into economics treatises, using the 2024 Market Definition Notice as a practical reference point for non-price parameters and platform settings.19 (2) Technical evidence literacy: judges need a working understanding of APIs, interoperability, ranking/recommendation systems, and experimentation (A/B tests): what such evidence can prove, its limits, and common strategic uses of ‘technical complexity’. Increasingly, disputes also involve automated decision systems (eg ranking or moderation tools) and claims about explainability or auditability. Training should therefore cover how to frame precise questions to experts (inputs, objectives, constraints, testing, error rates), and how to evaluate whether a party’s ‘black box’ claim is an evidential problem, a trade-secret claim, or a feasibility argument. (3) Remedial literacy: Digital remedies are dynamic and often engineering-heavy. Judges need criteria for designing or reviewing injunctions that are precise and enforceable and for reviewing authority-imposed behavioural remedies for proportionality and feasibility. The classical refusal-to-supply/interoperability doctrine remains relevant (‘Bronner’, ‘IMS Health’, ‘Microsoft’, ‘Slovak Telekom’), but digital cases show that its application depends on platform architecture and implementation constraints.20 (4) Regulatory coherence and parallel proceedings: DMA disputes intersect with General Data Protection Regulation (GDPR)/ePrivacy, DSA transparency, the Data Act, and consumer law. Courts must reason across regimes while preserving distinct legal tests. For example, a defendant’s GDPR data-minimization argument may conflict with a claimant’s DMA interoperability claim: both are legally cognizable, but the applicable standards, burden allocation, and remedial logic differ and must not be collapsed into one another. Article 39 of the DMA enables judicial cooperation with the European Commission; training should cover when stays are justified, how to use Commission observations, and how to prevent the routine suspension of private enforcement on speculative references to ongoing regulatory dialogue.21 The unifying aim is disciplined scrutiny that preserves effective judicial protection: intelligible theories of harm, verifiable evidence assessment, proportionate and enforceable remedies, and respect for defence rights, fast enough for digital markets but not ‘thin’ in review. Training will remain episodic unless judicial experience becomes visible and reusable. Regulation 1/2003 obliges Member States to forward judgments applying Articles 101/102 to the Commission; Article 39(5) DMA adds a parallel duty for national judgments applying the DMA. These duties support coherence and supply the raw material for comparative learning. Yet notification and accessibility have historically been patchy. Invisible judgments undermine legal certainty and inhibit peer learning, particularly in digital markets where similar disputes arise across jurisdictions. A practical response is a public database of national digital competition decisions: judgments applying Articles 101/102, the Damages Directive, significant procedural orders on disclosure/confidentiality, DMA-related judgments, and Article 102 cases materially shaped by DMA concepts. The database should include original-language texts, English summaries, European Case Law Identifier (ECLI) links, and structured metadata: legal basis, platform service, theory of harm, key economic concepts, evidence types (datasets, experiments, source-code descriptions), remedy type and monitoring approach, procedural posture (interim, merits, appeal), and any use of cooperation mechanisms (requests for information, Commission observations, stays). It should align with the DigitalJustice@2030 objectives on online access and structured legal data.22 The database should feed training. Cases should be converted into tools: model disclosure orders, confidentiality-ring templates, expert-question protocols, remedy-review checklists, and judgment-writing exercises (including traffic-based damages, interoperability injunctions, and stay decisions under Article 39 DMA). Funding should mix grants and with the modules and produce should to such as notification reuse of by national reasoning on and The EU digital competition acquis will if the courts that apply and review it have capacity to The DMA increases demands on national courts by concepts, technical evidence, and remedial and by between public enforcement and private litigation. The appropriate response is not generic but training by durable tools and case law A judicial capacity for EU digital competition law should therefore existing providers rather than EJTN for networks and diffusion; ERA for structured EU-law modules; for advanced and durable and national academies for localization and repeat delivery. Training should administrative review and civil private-enforcement and technical evidence, and remedial literacy across Finally, a public database of national judgments is to make judicial experience and reusable. in this a multilingual and practical judicial tools the between digital and national adjudication. They also national courts to as in and enforcement. In with the the that the article from the European Training and Networking and Training for National Competition Enforcers a training in competition law and economics organized by the European University Institute and co-funded by Competition of the European The expressed in this article are those of the and not reflect the or of the European Commission or the European University