Experimental Regulation in Digital Markets

Introduction

Experimental regulation in digital markets refers to regulatory approaches that deliberately test new legal, institutional, or technological interventions on a limited, monitored, and revisable basis before applying them across an entire market. It is particularly relevant to digital markets because platform business models, artificial intelligence, algorithms, data ecosystems, app stores, cloud infrastructure, and digital advertising evolve much faster than conventional legislation and judicial precedent.

Traditional competition law is generally reactive: an authority identifies allegedly harmful conduct, investigates it, establishes an infringement, and imposes a remedy. Experimental regulation is more adaptive. Regulators may use pilot schemes, regulatory sandboxes, behavioural experiments, temporary obligations, monitored remedies, interoperability trials, data-access experiments, algorithmic audits, or staged enforcement.

The central idea is:

Regulate, observe, measure, revise, and scale.

Experimental regulation does not mean that competition authorities can experiment without legal limits. Any experiment must remain consistent with statutory authority, procedural fairness, proportionality, due process, judicial review, and fundamental rights.

1. Meaning of Experimental Regulation

Experimental regulation involves a regulatory intervention that has one or more of the following characteristics:

1. Limited duration – the rule or remedy operates for a defined period.
2. Limited scope – it applies to selected firms, products, users, or markets.
3. Monitoring – the regulator collects evidence about the intervention's effects.
4. Measurable objectives – the regulator identifies indicators against which success can be evaluated.
5. Reversibility – the intervention can be modified or withdrawn.
6. Iterative adjustment – regulation changes in response to observed market outcomes.
7. Controlled experimentation – different regulatory approaches may be tested against one another.
8. Technological supervision – compliance may itself be monitored through automated systems.

In digital markets, experimental regulation may therefore occupy a middle position between:

laissez-faire → traditional ex-post antitrust → experimental/ex-ante regulation → permanent structural regulation.

2. Why Digital Markets Require Experimental Regulation

Digital markets create several problems for conventional regulation.

A. Rapid technological change

A legal rule designed for one technological architecture may become obsolete before enforcement proceedings conclude.

For example, regulation concerning:

• search engines;
• app stores;
• AI foundation models;
• cloud computing;
• digital advertising;
• recommendation algorithms;
• online marketplaces

may need continuous adjustment.

B. Network effects

A platform may become more valuable as more users join it.

This creates the possibility of:

users → data → better service → more users → more data → greater market power.

A regulatory intervention that appears modest at the individual-user level may therefore produce significant effects at ecosystem level.

C. Data-driven market power

Traditional market-share analysis may not adequately capture power derived from:

• proprietary datasets;
• behavioural information;
• search histories;
• location data;
• transaction information;
• interoperability advantages;
• machine-learning feedback loops.

D. Algorithmic opacity

A platform may not have a conventional written policy saying:

"Exclude competitors."

Instead, exclusion may arise from an algorithm's design, ranking incentives, default settings, or automated optimisation.

E. Uncertain competitive effects

A digital practice may simultaneously create:

• efficiencies;
• innovation;
• consumer benefits;
• privacy benefits;
• foreclosure risks.

Experimental regulation permits authorities to test these competing effects instead of assuming them in advance.

3. Main Forms of Experimental Regulation

3.1 Regulatory Sandboxes

A regulatory sandbox permits firms to test innovative products or services under controlled regulatory conditions.

A sandbox may involve:

• limited users;
• restricted geographic scope;
• reporting requirements;
• temporary exemptions;
• regulator supervision;
• predefined termination conditions.

In competition law, a sandbox can be particularly useful for:

• fintech;
• AI services;
• digital identity;
• blockchain;
• data-sharing systems;
• algorithmic marketplaces.

However, a sandbox must not become a mechanism through which incumbent platforms receive unjustified regulatory privileges.

4. Experimental Competition Remedies

Competition authorities can also experiment with remedies.

Instead of imposing a permanent remedy immediately, the authority might adopt:

Phase I

Limited behavioural remedy.

Phase II

Monitoring and data collection.

Phase III

Evaluation of competitive effects.

Phase IV

Modification or strengthening.

Phase V

Permanent remedy if necessary.

This can be useful where the authority is uncertain whether:

• interoperability will actually stimulate competition;
• data access will create viable competitors;
• a ranking remedy will eliminate self-preferencing;
• a default-setting remedy will change user behaviour.

5. Algorithmic Regulation as Experimental Regulation

Algorithms create a particularly important experimental-regulation environment.

A competition authority may require:

• algorithmic audits;
• transparency reports;
• independent monitoring;
• access to testing environments;
• periodic compliance reports;
• controlled experiments;
• explanation of ranking changes;
• testing for discriminatory or exclusionary outcomes.

For example, instead of simply ordering a search platform not to favour its own service, a regulator could monitor:

baseline ranking → intervention → user response → competitor visibility → consumer outcomes.

The regulatory decision can then be adjusted according to empirical evidence.

6. Experimental Regulation and Ex-Ante Digital Regulation

Experimental regulation is closely connected to modern ex-ante digital regulation.

Traditional antitrust generally asks:

Has the undertaking abused market power?

Experimental ex-ante regulation may ask:

What regulatory architecture can prevent harmful conduct before it becomes entrenched?

This distinction is particularly important for digital platforms because network effects can make later correction extremely difficult.

7. Case Law

Case 1: United States v. Microsoft Corp. (2001)

The Microsoft litigation is one of the most important foundations for modern digital-market regulation.

Microsoft was found to have unlawfully maintained its operating-system monopoly and engaged in conduct designed to restrict competing technologies, particularly Netscape's browser.

The remedy debate demonstrated the difficulty of designing effective intervention in a rapidly changing technology market.

Relevance to experimental regulation

The case demonstrates that:

• technology markets change during litigation;
• behavioural remedies can be difficult to design;
• structural remedies may have uncertain consequences;
• regulatory intervention must account for technological evolution.

The Microsoft experience therefore supports a more adaptive approach to digital remedies.

8. Case 2: FTC v. Qualcomm Inc. (2020)

The Qualcomm litigation concerned licensing practices involving cellular-standard-essential patents and competition in modem chip markets.

Although the Ninth Circuit ultimately rejected the FTC's theory of antitrust liability, the case is important because it illustrates the difficulty of applying competition principles to complex technology ecosystems.

Experimental-regulation significance

The case demonstrates that intervention in technology markets can have consequences extending beyond the immediate parties.

Regulators must consider:

• innovation incentives;
• licensing structures;
• downstream competition;
• technological standards;
• investment incentives.

An experimental approach can help authorities determine whether a proposed intervention actually improves competitive conditions without undermining innovation.

9. Case 3: Google Shopping – European Commission / General Court

The Google Shopping proceedings concerned Google's treatment of its comparison-shopping service in search results.

The European Commission found that Google had systematically favoured its comparison-shopping service over competing comparison-shopping services.

The General Court largely upheld the Commission's decision, while the litigation also demonstrated the complexity of evaluating self-preferencing in digital ecosystems.

Experimental-regulation significance

The case illustrates the value of continuous monitoring of platform behaviour.

A regulator dealing with self-preferencing could potentially use:

• ranking audits;
• controlled searches;
• competitor-visibility measurements;
• consumer-choice metrics;
• periodic compliance monitoring.

Thus, digital regulation can move beyond a single infringement decision toward continuous assessment of platform conduct.

10. Case 4: Google Android – European Commission

The Google Android case concerned contractual restrictions involving Android devices, including requirements relating to Google Search and the Play Store.

The European Commission identified several practices that restricted competition in mobile search.

Experimental-regulation significance

Android demonstrates why digital regulation can require ecosystem-level monitoring.

A regulator may need to examine:

• default settings;
• pre-installation;
• choice screens;
• app-store access;
• revenue-sharing arrangements;
• switching behaviour;
• competing search-engine adoption.

Instead of treating each contractual arrangement independently, experimental regulation can test whether particular interventions actually increase consumer choice.

11. Case 5: Meta Platforms Inc. / Facebook Data Practices – EU Competition Context

The European Commission and national competition authorities have increasingly examined the relationship between data accumulation and competition in digital platforms.

The Facebook/Meta investigations are particularly important because digital market power can arise from the combination of:

data + network effects + user lock-in + advertising infrastructure.

Experimental-regulation significance

This area demonstrates why regulators may need to test:

• data-access obligations;
• data portability;
• interoperability;
• consent architecture;
• separation of datasets;
• restrictions on combining data from different services.

The regulatory intervention can then be evaluated according to whether competitors actually obtain meaningful opportunities to compete.

12. Case 6: Epic Games v. Apple

The Epic Games litigation concerning Apple's App Store rules is highly relevant to experimental regulation of digital ecosystems.

The dispute concerned Apple's restrictions on alternative payment mechanisms and the structure of its App Store ecosystem.

Experimental-regulation significance

App-store regulation provides a natural setting for controlled regulatory experiments.

Possible interventions include:

• alternative payment systems;
• third-party app stores;
• interoperability requirements;
• reduced commissions;
• anti-steering rules;
• developer choice mechanisms.

The effectiveness of such measures can be empirically assessed through:

• developer entry;
• consumer prices;
• transaction volumes;
• innovation;
• security incidents;
• switching rates.

The case therefore illustrates the challenge of designing remedies that simultaneously preserve legitimate platform functions while reducing exclusionary effects.

13. Case 7: Ohio v. American Express Co. (2018)

The U.S. Supreme Court considered restrictions imposed by American Express on merchants concerning steering customers toward alternative payment systems.

The Court treated the credit-card platform as a two-sided market and emphasised the need to consider effects on both sides of the platform.

Experimental-regulation significance

The case is important because digital platforms frequently operate as multi-sided markets.

A regulatory experiment cannot examine only one side.

For example:

Consumers → Platform ← Advertisers

or

Developers → App Store ← Users

or

Sellers → Marketplace ← Consumers

A remedy that benefits one side may harm another.

Experimental regulation can therefore measure effects across both sides before making a permanent regulatory decision.

14. Case 8: Google LLC v. Oracle America, Inc. (2021)

Although primarily a copyright case, Google v. Oracle is relevant to digital competition because it demonstrates the importance of interoperability and technological reuse in digital ecosystems.

The dispute concerned Google's use of Java API declarations in Android.

Regulatory significance

Digital ecosystems frequently depend upon:

• APIs;
• standards;
• interoperability;
• software interfaces;
• developer ecosystems.

A regulatory system that imposes interoperability obligations should therefore experimentally evaluate whether access:

• increases competition;
• reduces innovation;
• creates security risks;
• facilitates entry.

15. Case 9: AT&T Mobility LLC v. Concepcion (2011)

Although not a competition case in the conventional sense, Concepcion is useful for understanding the contractual architecture through which digital platforms can govern users and counterparties.

Digital platforms increasingly rely upon standard-form contractual arrangements.

Experimental regulatory oversight may therefore examine whether contractual restrictions:

• prevent switching;
• restrict collective action;
• limit alternative platforms;
• impose disproportionate arbitration mechanisms;
• create ecosystem lock-in.

The case illustrates the broader significance of contractual architecture in technology markets.

16. European Union Digital Markets Approach

The EU's modern digital-market framework demonstrates a movement toward continuous ex-ante regulation.

The Digital Markets Act identifies designated gatekeepers and imposes obligations relating to areas such as:

• interoperability;
• data combination;
• self-preferencing;
• steering;
• user choice;
• access to data.

Although the DMA is not simply a "sandbox" regime, its gatekeeper architecture is compatible with experimental regulatory techniques because compliance can be:

1. imposed;
2. monitored;
3. investigated;
4. adjusted through enforcement experience.

This represents a major shift from purely retrospective competition enforcement.

17. United Kingdom Approach

The UK has similarly moved toward a more proactive digital-market model through the Digital Markets, Competition and Consumers framework.

The central concept is the use of tailored obligations for firms with substantial and entrenched market power.

This creates the possibility of:

• firm-specific conduct requirements;
• pro-competition interventions;
• monitoring;
• periodic modification;
• consultation with affected stakeholders.

The UK model is particularly suitable for experimental regulation because digital obligations can be designed around the characteristics of individual ecosystems rather than relying exclusively upon uniform rules.

18. Germany and Section 19a GWB

Germany provides another important model.

Section 19a of the German Competition Act permits enhanced intervention against undertakings of paramount significance across markets.

The Bundeskartellamt can examine specific practices involving major digital ecosystems.

This framework is significant because it permits earlier intervention than traditional abuse-of-dominance proceedings.

The approach can therefore be understood as a form of institutional experimentation in digital competition enforcement.

19. Advantages of Experimental Regulation

A. Flexibility

Rules can evolve as technology changes.

B. Evidence-based policymaking

Regulators obtain empirical evidence rather than relying entirely upon predictions.

C. Lower risk of over-regulation

A limited intervention can be tested before becoming permanent.

D. Innovation protection

Regulators can examine whether a proposed remedy unintentionally suppresses innovation.

E. Faster correction

If an intervention fails, it can be modified rather than waiting for a lengthy legislative process.

F. Better understanding of algorithms

Continuous monitoring can reveal effects that traditional investigation cannot detect.

20. Risks of Experimental Regulation

Experimental regulation also creates significant legal concerns.

1. Regulatory uncertainty

Businesses may not know which rules will ultimately apply.

2. Arbitrary experimentation

Authorities cannot treat firms or consumers as experimental subjects without appropriate legal safeguards.

3. Regulatory capture

Large technology companies may influence the design of regulatory experiments.

4. Unequal treatment

A pilot intervention affecting one firm may create competitive distortions.

5. Privacy concerns

Experimental data collection can involve highly sensitive user information.

6. Due process

Businesses must have meaningful opportunities to challenge regulatory decisions.

7. Measurement problems

Digital markets involve numerous variables, making causal attribution difficult.

8. Innovation chilling

Excessive regulatory experimentation can discourage investment in new technologies.

21. Safeguards for Experimental Digital Regulation

A legitimate experimental regulatory system should contain at least the following safeguards:

1. Clear statutory authority

The regulator must have a legal basis for the intervention.

2. Defined objective

The regulator should identify precisely what the experiment seeks to achieve.

3. Time limitation

Experimental measures should ordinarily have a review or sunset mechanism.

4. Transparency

Affected firms and stakeholders should understand the regulatory criteria.

5. Measurable indicators

Examples include:

• entry;
• switching;
• prices;
• quality;
• innovation;
• interoperability;
• consumer choice.

6. Independent evaluation

Regulatory experiments should preferably be assessed independently.

7. Proportionality

The intervention should not exceed what is necessary.

8. Judicial review

Courts must retain authority to examine legality and procedural fairness.

9. Privacy protection

Data collection should comply with applicable data-protection principles.

10. Exit mechanism

Failed experiments should be capable of termination.

22. Experimental Regulation and AI Markets

AI makes experimental regulation even more important.

AI markets may involve:

• foundation models;
• compute providers;
• model marketplaces;
• inference APIs;
• AI agents;
• training datasets;
• cloud infrastructure;
• AI-enabled search;
• autonomous pricing systems.

A regulator might initially impose a limited transparency obligation on a powerful AI platform and then measure:

transparency → competitor access → innovation → consumer outcomes → compliance costs.

If the experiment produces positive results, the obligation could be expanded.

23. Experimental Regulation and Algorithmic Pricing

Suppose competing firms use autonomous pricing algorithms.

Traditional antitrust asks whether there is:

• an agreement;
• concerted practice;
• unilateral conduct;
• exclusionary intent.

But autonomous systems may generate parallel pricing without explicit communication.

An experimental regulatory response could involve:

1. monitoring pricing algorithms;
2. collecting historical price data;
3. identifying algorithmic convergence;
4. testing alternative algorithmic constraints;
5. measuring consumer effects;
6. imposing additional safeguards if necessary.

This is particularly important where algorithms can adapt faster than conventional enforcement.

24. Experimental Regulation and Data Access

Data-access remedies may also be experimentally designed.

For example:

Stage 1: limited competitor access to a dataset.

Stage 2: monitor innovation and privacy effects.

Stage 3: expand access if competitive benefits appear.

Stage 4: introduce stronger safeguards if privacy or security risks emerge.

This avoids the false choice between:

unrestricted data access

and

complete data enclosure.

25. Experimental Regulation vs Traditional Antitrust

The two approaches should not be viewed as competitors. Experimental regulation can supplement traditional antitrust.

26. Decision Framework for Regulators

A regulator considering experimental regulation can use the following framework:

Identify systemic digital problem

↓

Determine legal authority

↓

Define competitive harm

↓

Identify measurable objectives

↓

Design limited intervention

↓

Set monitoring indicators

↓

Conduct regulatory experiment

↓

Collect market evidence

↓

Assess consumer + competitor + innovation effects

↓

Modify / terminate / expand intervention

↓

Adopt permanent regulatory framework if justified

27. Key Legal Principles

Experimental digital-market regulation should be guided by:

Proportionality

The intervention must be suitable and necessary.

Legal certainty

Affected firms should understand their obligations.

Procedural fairness

Affected parties should have an opportunity to respond.

Accountability

Regulators should explain why the experiment was adopted.

Non-discrimination

Comparable firms should not be arbitrarily treated differently.

Innovation neutrality

Regulation should not unnecessarily favour one technological model.

Consumer welfare

Consumer choice, quality, privacy, security, and innovation should be considered.

Competitive neutrality

The experiment should not unintentionally create advantages for incumbents.

28. Overall Assessment

Experimental regulation represents a significant evolution in digital competition governance. Digital markets are characterised by rapid technological change, network effects, data accumulation, algorithmic decision-making and ecosystem dependence. These characteristics make purely retrospective enforcement increasingly difficult.

The major contribution of experimental regulation is that it transforms regulation from a one-time legal decision into a continuous learning process.

The strongest model is therefore not:

regulate first and never change the rule.

It is:

regulate → measure → evaluate → revise → institutionalise.

The Microsoft, Google, Qualcomm, American Express, Epic Games and other technology-related cases demonstrate why digital competition remedies must account for technological complexity, multi-sided markets, interoperability, innovation and rapidly changing business models.

At the same time, experimentation cannot become an excuse for uncontrolled administrative discretion. Legality, proportionality, transparency, due process, privacy, judicial review and measurable objectives are essential safeguards.

Conclusion

Experimental regulation in digital markets is best understood as adaptive competition governance. It combines traditional legal authority with empirical testing, continuous monitoring and reversible interventions.

Its greatest value lies in situations where regulators face high uncertainty and potentially irreversible digital market effects. Instead of waiting until network effects, data advantages or ecosystem lock-in make competition impossible to restore, authorities can test targeted interventions earlier.

The future of digital competition law is therefore likely to involve a hybrid model:

traditional antitrust + ex-ante regulation + regulatory experimentation + continuous market monitoring.

That combination allows regulators to remain sufficiently flexible for rapidly changing digital markets while preserving the fundamental legal constraints that prevent experimentation from becoming arbitrary regulation.