Competition Law and Recommendation Interoperability Obligations

1. Introduction

Recommendation interoperability refers to the ability of competing services, platforms, applications, or recommendation systems to exchange, receive, process, or make use of information necessary for their recommendation functions to operate effectively across technological or platform boundaries.

Recommendation systems are increasingly important in search engines, social-media platforms, e-commerce marketplaces, app stores, streaming services, travel platforms, food-delivery applications, digital advertising, and online marketplaces. A platform controlling a large recommendation infrastructure may influence which products, sellers, applications, creators, or services users encounter.

Competition concerns arise where a dominant undertaking:

• prevents competitors from accessing recommendation-related data;
• restricts interoperability with competing recommendation services;
• uses technical interfaces to exclude rival recommendation engines;
• gives preferential treatment to its own recommendation service;
• makes interoperability conditional on unreasonable terms;
• degrades the quality of interoperability for rivals;
• prevents users from choosing alternative recommendation providers; or
• combines recommendation data from several markets to reinforce market power.

The legal issue is therefore not simply whether interoperability exists, but whether a dominant undertaking can lawfully control the technical and informational infrastructure necessary for effective competition between recommendation services.

2. Meaning of Recommendation Interoperability

Recommendation interoperability can involve several layers.

A. Data interoperability

A competing recommendation service may need access to:

• product information;
• user-generated ratings;
• reviews;
• metadata;
• interaction information;
• availability information;
• catalogue information;
• contextual signals; and
• user-authorised preference data.

B. Technical interoperability

The competing system may require:

• APIs;
• SDKs;
• feeds;
• authentication protocols;
• real-time interfaces;
• search interfaces; or
• standardised data formats.

C. Functional interoperability

The competing recommendation engine should be capable of performing substantially equivalent functions, such as:

• ranking;
• filtering;
• personalisation;
• recommendation generation;
• content discovery; and
• product matching.

D. User-choice interoperability

Users may be able to select an alternative recommendation provider rather than being forced to use the platform's own system.

3. Why Recommendation Interoperability Matters to Competition

Recommendation systems can constitute an important distribution and discovery channel.

A rival product may technically be available on a platform but commercially ineffective if the platform's recommendation algorithm systematically prevents users from discovering it.

Consequently:

Access to a marketplace does not necessarily constitute effective access to the recommendation channel through which consumers discover products.

This distinction is particularly important in digital markets.

A platform may therefore possess several forms of competitive power:

1. control over users;
2. control over data;
3. control over ranking infrastructure;
4. control over APIs;
5. control over recommendation algorithms;
6. control over technical standards; and
7. control over the interface through which users discover competing products.

4. Relevant Competition-Law Theories

A. Abuse of Dominant Position

Where an undertaking holds a dominant position, refusal or restriction of interoperability may constitute an abuse where it has exclusionary effects and lacks adequate objective justification.

Depending upon the jurisdiction, relevant theories may include:

• refusal to deal;
• denial of access to an essential input;
• discriminatory access;
• tying;
• self-preferencing;
• exclusionary technical design;
• margin-related exclusion;
• discriminatory interoperability;
• leveraging of dominance; and
• unfair or unreasonable access conditions.

5. Essential-Facility Considerations

Recommendation infrastructure may sometimes resemble an essential facility where:

1. the facility is effectively controlled by a dominant undertaking;
2. competitors cannot reasonably reproduce it;
3. access is necessary for effective competition;
4. refusal substantially eliminates competition; and
5. access can technically and economically be provided.

However, not every valuable dataset, API or algorithm is an essential facility.

Competition authorities and courts generally distinguish between:

• commercially useful access; and
• access that is indispensable for viable competition.

This distinction prevents competition law from becoming a general obligation to share commercially valuable assets.

6. Six Major Case Laws

1. Bronner v Mediaprint — European Union

Case: Oscar Bronner GmbH & Co. KG v Mediaprint Zeitungs und Zeitschriftenverlag GmbH & Co. KG, Case C-7/97.

Facts

Bronner operated a newspaper competing with Mediaprint. Mediaprint controlled an extensive newspaper-delivery network.

Bronner argued that access to the delivery system was necessary for competition.

Principle

The Court of Justice established a stringent framework for refusal-to-supply cases.

Access may be required only where the facility is genuinely indispensable and its duplication is not realistically possible.

Relevance to recommendation interoperability

A dominant platform controlling a recommendation interface cannot automatically be required to expose its recommendation infrastructure merely because competitors would benefit from access.

The Bronner principle supports careful examination of:

• indispensability;
• replicability;
• technical feasibility;
• economic feasibility; and
• elimination of effective competition.

7. Microsoft — European Commission

Case: Microsoft Corp. v Commission, Case T-201/04.

Facts

Microsoft was found to have abused its dominant position by restricting interoperability information needed by competing work-group server operating systems.

Principle

The case is particularly important because interoperability information itself could constitute an important competitive input.

The Commission's intervention concerned Microsoft's refusal to provide sufficient interoperability information to rivals.

Relevance

Recommendation ecosystems may similarly involve interoperability information.

For example, a dominant platform might control:

• API documentation;
• communication protocols;
• metadata;
• system interfaces; and
• technical information necessary for rival recommendation services.

The Microsoft decision demonstrates that competition law may address technical interoperability restrictions, not merely conventional refusals to supply physical products.

8. IMS Health — European Union

Case: IMS Health GmbH & Co. OHG v NDC Health GmbH & Co. KG, Joined Cases C-418/01 and C-7/01.

Facts

IMS Health controlled a copyrighted structure used for pharmaceutical sales-data reporting.

A competitor sought access because the structure had become an important industry standard.

Principle

The Court established strict conditions for compulsory access to intellectual-property-protected infrastructure.

The refusal could amount to abuse in exceptional circumstances where, among other things, access was indispensable and refusal prevented the emergence of a new product for which consumer demand existed.

Relevance

Recommendation platforms frequently involve proprietary:

• algorithms;
• taxonomies;
• datasets;
• classification structures; and
• interfaces.

IMS Health demonstrates that interoperability obligations involving proprietary infrastructure require careful consideration of innovation incentives and the exceptional nature of compulsory access.

9. Slovak Telekom — European Union

Case: Slovak Telekom a.s. v European Commission, Joined Cases C-152/19 P and C-165/19 P.

Facts

The case concerned access to telecommunications infrastructure and alleged exclusionary conduct by a vertically integrated dominant undertaking.

Principle

The judgment examined the relationship between general dominance rules and refusal-of-access situations involving regulated infrastructure.

Relevance

The case demonstrates that competition analysis of access restrictions must consider:

• the nature of the infrastructure;
• the regulatory environment;
• the competitive structure;
• the terms of access; and
• the actual exclusionary effects.

For recommendation interoperability, this means authorities should distinguish between:

complete refusal, degraded access, and discriminatory access.

10. Google Shopping — European Union

Case: Google and Alphabet v Commission, Case T-612/17.

Facts

The European Commission found that Google had favoured its own comparison-shopping service in its general search results while demoting competing comparison-shopping services.

The General Court largely upheld the Commission's decision.

Principle

The case is highly relevant to self-preferencing and ranking discrimination.

The central competition concern was not merely access to Google's search engine but the manner in which Google's search and ranking mechanisms affected competitors' visibility.

Relevance to recommendation interoperability

A recommendation platform could similarly distort competition by:

• giving its own recommendation service privileged placement;
• suppressing competing recommendation providers;
• manipulating ranking interfaces;
• reducing rival visibility; or
• making competing recommendations technically less effective.

Thus, interoperability without non-discriminatory treatment may be insufficient.

11. Android — European Union

Case: Google and Alphabet v Commission, Case T-604/18.

Facts

The European Commission investigated Google's practices concerning the Android ecosystem, including arrangements involving search, browsers and app distribution.

Principle

The case concerned the use of contractual and ecosystem mechanisms to reinforce Google's position in related digital markets.

Relevance

Recommendation interoperability can similarly become problematic where a dominant platform uses control over an ecosystem to restrict alternative recommendation providers.

Examples include:

• mandatory default recommendation engines;
• technical restrictions on alternative engines;
• contractual restrictions;
• preferential access to system resources;
• restrictions on pre-installation; and
• discriminatory treatment of alternative providers.

12. United States v Microsoft

Case: United States v Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001).

Facts

Microsoft's conduct concerning the Windows operating-system ecosystem and Internet Explorer was examined under U.S. antitrust law.

Principle

The case demonstrated that a dominant technology platform can potentially use control over an important platform or interface to disadvantage competing technologies.

Relevance

Modern recommendation platforms can similarly act as technological gateways.

Competition concerns can arise where the platform:

• restricts rival access;
• technically degrades interoperability;
• makes alternative recommendation engines difficult to install;
• controls defaults; or
• uses platform power to reinforce a neighbouring market position.

13. Google AdSense — European Union

Case: European Commission, Google Search (AdSense) decision, 2019.

Facts

The Commission examined contractual restrictions imposed by Google on publishers using its advertising intermediation service.

Principle

The case illustrates how contractual restrictions imposed by a dominant digital intermediary can restrict rivals' ability to compete for users or business opportunities.

Relevance

Recommendation infrastructure can involve similar contractual restrictions.

For example, a platform could prohibit participating businesses from:

• using competing recommendation engines;
• exporting recommendation-related information;
• integrating alternative discovery services; or
• providing equivalent recommendation functionality elsewhere.

Such restrictions can be examined for their foreclosure effects.

14. Competition Concerns Arising from Recommendation Interoperability

A. API Refusal

A dominant platform may refuse to provide API access to rival recommendation providers.

The legal questions include:

• Is API access indispensable?
• Can competitors develop alternatives?
• Is the refusal objectively justified?
• Does refusal eliminate effective competition?
• Is the API technically capable of supporting multiple providers?

B. Discriminatory API Access

A platform may technically provide interoperability but give its own recommendation service:

• faster access;
• more data;
• higher API limits;
• richer metadata;
• preferential latency;
• superior update frequency; or
• privileged system permissions.

This may produce de facto discrimination.

C. Self-Preferencing

A platform's own recommendation engine may receive preferential treatment.

For example:

Platform → Product Catalogue → Multiple Recommendation Engines

may become:

Platform → Own Recommendation Engine → Preferred Products → Consumers

Competitors may technically remain interoperable while being commercially disadvantaged.

15. Data Portability and Recommendation Competition

Data portability can increase contestability.

Users may wish to transfer:

• purchase history;
• viewing history;
• ratings;
• playlists;
• follows;
• search history;
• preferences; and
• interaction data

to another recommendation service.

Competition law may therefore interact with broader data-protection and digital-market regulation.

However, competition law does not automatically imply that every category of data must be disclosed to competitors.

Questions concerning:

• privacy;
• consent;
• cybersecurity;
• trade secrets;
• intellectual property;
• personal data;
• proportionality; and
• technical security

must also be considered.

16. Algorithmic Interoperability

Recommendation interoperability becomes more complicated where algorithms are involved.

A platform could theoretically provide raw data but restrict access to information necessary to make that data useful.

For example:

Data access: Yes
Real-time access: No
Metadata: Limited
API rate: Very low
Ranking signals: Unavailable
Technical documentation: Incomplete

Such arrangements can produce formal interoperability without effective interoperability.

Competition authorities may therefore examine the practical quality of access rather than merely whether an API exists.

17. Interoperability and Network Effects

Recommendation systems benefit from network effects.

More users generate:

• more behavioural data;
• better personalisation;
• greater recommendation accuracy;
• more commercial participation; and
• greater attractiveness to additional users.

This can create a feedback loop:

More users → More data → Better recommendations → More users

If competitors cannot obtain sufficient interoperability, they may be unable to reach the scale necessary to compete.

This can create a data-driven entry barrier.

18. Tying and Bundling

A dominant platform may require users to accept its recommendation engine as a condition of accessing another service.

Examples include:

• app store + recommendation engine;
• operating system + search recommendation;
• smart TV + content recommendation;
• e-commerce marketplace + product-ranking engine.

The competition analysis may involve:

1. dominance in the tying market;
2. separate products or services;
3. coercion or technical restriction;
4. foreclosure;
5. competitive effects; and
6. possible efficiencies.

19. Interoperability and Switching Costs

Interoperability can reduce switching costs.

Without interoperability:

User history → Platform A → Recommendation system A

With interoperability:

User history → Transfer/API → Recommendation system B

Reduced switching costs may increase:

• multi-homing;
• entry;
• innovation;
• consumer choice;
• price competition; and
• quality competition.

Conversely, excessive interoperability obligations may potentially reduce incentives to invest in proprietary systems.

Competition law therefore has to balance contestability with innovation incentives.

20. Objective Justifications

A dominant platform may have legitimate reasons for restricting interoperability.

Possible justifications include:

Cybersecurity

Opening APIs may create security vulnerabilities.

Privacy

Recommendation systems can contain highly sensitive behavioural information.

Intellectual Property

Algorithms, databases and technical systems may be protected by IP rights.

System Integrity

Uncontrolled third-party access may interfere with platform stability.

Fraud Prevention

Bad actors could manipulate recommendation systems through automated access.

Capacity Constraints

Unlimited API access may create substantial technical burdens.

Quality Control

A platform may seek to prevent misleading or unsafe recommendations.

However, the justification should generally be assessed against the availability of less restrictive alternatives.

21. Possible Competition-Law Remedies

Where anticompetitive interoperability restrictions are established, remedies may include:

1. API Access

Ordering access to specified technical interfaces.

2. Non-Discrimination

Requiring equivalent treatment of internal and external recommendation providers.

3. Data Portability

Allowing users to transfer relevant data.

4. Technical Standards

Requiring adherence to interoperable standards.

5. FRAND-Type Access

Access may potentially be required on fair, reasonable and non-discriminatory terms where appropriate.

6. Choice Screens

Users may be offered a genuine choice among recommendation providers.

7. Monitoring

An independent monitoring mechanism may verify whether interoperability is actually effective.

22. Key Legal Tests

A competition authority examining recommendation interoperability can consider the following framework:

Step 1 — Relevant Market

Identify the relevant:

• product market;
• technology market;
• recommendation market; and
• geographic market.

Step 2 — Dominance

Determine whether the undertaking possesses substantial market power.

Relevant indicators may include:

• market share;
• network effects;
• switching costs;
• data advantages;
• entry barriers;
• ecosystem control; and
• vertical integration.

Step 3 — Interoperability Dependency

Ask whether competitors genuinely depend upon the platform's infrastructure.

Step 4 — Conduct

Identify whether the undertaking:

• refuses access;
• delays access;
• degrades access;
• discriminates;
• self-preferences; or
• imposes exclusionary contractual conditions.

Step 5 — Foreclosure

Determine whether competitors are actually or potentially excluded.

Step 6 — Consumer Effects

Examine effects on:

• choice;
• prices;
• quality;
• innovation;
• privacy;
• product variety; and
• switching.

Step 7 — Justification

Consider legitimate technical, security, privacy and efficiency explanations.

Step 8 — Remedy

Select the least restrictive remedy capable of restoring effective competition.

23. Comparative Case-Law Principles

24. Indian Competition-Law Perspective

In India, recommendation interoperability can potentially be examined primarily under the Competition Act, 2002, particularly the provisions concerning abuse of dominant position.

Relevant forms of conduct may include:

• denial of market access;
• discriminatory conditions;
• unfair conditions;
• leveraging dominance;
• tying or bundling; and
• exclusionary contractual arrangements.

The Competition Commission of India can also consider the characteristics of digital markets, including:

• network effects;
• data advantages;
• ecosystem effects;
• switching costs;
• multi-sided platforms; and
• technological barriers.

The CCI's digital-market enforcement involving major technology platforms provides an important background for analysing recommendation systems, even where the precise factual issue is not labelled "recommendation interoperability."

25. Distinguishing Interoperability from Mandatory Algorithm Sharing

An important legal distinction is:

Interoperability ≠ compulsory disclosure of the entire algorithm.

An interoperability obligation could potentially require:

• API access;
• technical protocols;
• data portability;
• standardised interfaces;
• functional access; or
• non-discriminatory technical conditions.

It does not necessarily require disclosure of:

• source code;
• proprietary algorithms;
• confidential business information; or
• commercially sensitive ranking parameters.

This distinction is particularly important because compulsory disclosure can undermine innovation incentives and cybersecurity.

26. Emerging Competition Issues

Future disputes are likely to concern:

AI recommendation systems

AI models may determine:

• what consumers see;
• which products are recommended;
• which sellers receive visibility; and
• which content is promoted.

Foundation-model recommendations

Large AI models may become intermediary recommendation layers between consumers and suppliers.

Smart-device ecosystems

Smart TVs, vehicles, phones and wearable devices may control recommendation interfaces.

Retail media

Retail platforms may use first-party purchasing data to favour their own advertising and recommendation products.

Generative AI assistants

AI assistants may replace traditional search and become a major discovery channel.

This could make interoperability between AI assistants, search indexes, marketplaces and recommendation providers a significant competition-law issue.

27. Conclusion

Recommendation interoperability represents an emerging intersection between dominance, refusal to deal, essential facilities, self-preferencing, data portability, platform regulation and technological neutrality.

The central competition question is:

Does control over recommendation infrastructure allow a dominant undertaking to prevent rival recommendation services from competing effectively, and is any restriction objectively justified and proportionate?

The leading cases demonstrate different dimensions of the problem. Bronner and IMS Health establish caution around compulsory access; Microsoft illustrates the importance of technical interoperability; Google Shopping demonstrates the competitive significance of ranking and self-preferencing; Android illustrates ecosystem-based leveraging; and US v Microsoft shows how control over technological platforms can affect adjacent competition.