Competition Law and Competition Governance in Computational Societies

1. Introduction

A computational society is a social and economic environment in which important commercial decisions are increasingly made, assisted, or mediated by algorithms, artificial intelligence, automated decision systems, data infrastructures, digital platforms, machine-learning models, and computational networks.

In such a society, competition is no longer determined only by conventional human decisions. Algorithms may determine:

• prices and discounts;
• ranking and search results;
• allocation of customers;
• access to platforms and APIs;
• advertising placement;
• credit and insurance decisions;
• recommendations;
• matching of buyers and sellers;
• interoperability;
• visibility of competing products;
• allocation of digital resources; and
• acquisition or exclusion strategies.

The central competition-law problem therefore changes from merely asking “Did firms agree to restrict competition?” to also asking:

Who designed the computational system, who controls its inputs and outputs, who can audit it, and whether the system itself structurally prevents effective competition?

This is particularly important because algorithms can facilitate coordination even where competitors do not communicate in the traditional manner. The OECD has recognised that algorithms can facilitate monitoring, signalling and implementation of coordinated outcomes, including circumstances where conventional human interaction is absent.

2. Meaning of Competition Governance

Competition governance is broader than conventional antitrust enforcement.

Traditional competition law generally operates through:

1. prohibition of cartels;
2. control of abuse of dominance;
3. merger control;
4. market investigation;
5. penalties and behavioural remedies.

Computational societies require an additional governance layer involving:

• algorithmic transparency;
• auditability;
• interoperability;
• data portability;
• access to essential digital infrastructure;
• non-discriminatory ranking;
• API access;
• prevention of self-preferencing;
• monitoring of algorithmic pricing;
• governance of automated decision systems;
• technical compliance mechanisms; and
• ex-ante regulation of systemic gatekeepers.

The EU Digital Markets Act illustrates this movement from purely ex-post antitrust toward ex-ante governance. The Commission currently identifies Alphabet, Amazon, Apple, Booking, ByteDance, Meta and Microsoft among designated gatekeepers and regulates specified core platform services.

3. Why Computational Societies Create New Competition Problems

A. Algorithmic collusion

Two competitors may employ pricing algorithms that continuously observe market conditions and adjust prices.

The difficult question is whether:

independent algorithmic adaptation can amount to unlawful coordination?

Traditional cartel law normally looks for an agreement, arrangement, understanding or concerted practice. Autonomous computational systems can complicate that inquiry because the competitive outcome may emerge from algorithmic learning rather than an explicit human agreement.

In India, this issue is particularly significant under Sections 3 and 2(b) of the Competition Act, 2002, because Section 2(b) encompasses an arrangement, understanding or action in concert.

B. Algorithmic self-preferencing

A dominant platform may simultaneously operate:

• the infrastructure;
• the search engine;
• the marketplace;
• the advertising system; and
• competing downstream services.

Its algorithm can therefore determine whether its own service appears first while rivals are demoted.

This was central to Google Shopping, where the European courts considered Google's preferential positioning of its own comparison-shopping service.

C. Data concentration

Computational societies depend heavily upon data.

A dominant undertaking possessing:

• transaction data;
• behavioural data;
• search data;
• location data;
• advertising data; and
• platform interaction data

may acquire advantages that competitors cannot easily reproduce.

Consequently, competition governance increasingly concerns data access and data portability, rather than merely price.

D. Network effects

Digital platforms frequently become more valuable as more users join.

This can produce a feedback loop:

More users → more data → better algorithms → better service → more users → greater market power

Once this cycle becomes entrenched, a technically superior entrant may still face substantial barriers to entry.

4. Computational Market Power

Market power in a computational society can arise from several interconnected resources.

1. Data power

Control over commercially significant datasets.

2. Computational power

Access to advanced computing infrastructure and AI capabilities.

3. Algorithmic power

Ability to determine how markets operate through automated systems.

4. Platform power

Ability to determine who may participate in an ecosystem.

5. Interface power

Control over APIs, app stores, search interfaces and interoperability.

6. Network power

Advantages generated by large user networks.

7. Switching power

Ability to make it difficult or expensive for users and businesses to migrate.

Thus, competition authorities increasingly need to examine architecture as well as conduct.

5. Major Competition-Law Doctrines Applicable to Computational Societies

A. Section 3 — Anti-competitive agreements

Under Indian law, Section 3 can address agreements or concerted practices that cause or are likely to cause an appreciable adverse effect on competition.

Computational applications include:

• algorithmic price coordination;
• exchange of competitively sensitive data;
• platform parity clauses;
• automated allocation arrangements;
• algorithmic resale-price maintenance;
• coordinated bidding systems.

The challenge is attribution.

If an algorithm independently reaches a supra-competitive outcome, authorities must determine whether there is sufficient evidence connecting the outcome to the conduct of the undertaking.

B. Section 4 — Abuse of dominant position

Section 4 becomes particularly significant where computational power is concentrated.

Possible forms include:

• self-preferencing;
• discriminatory access;
• tying;
• refusal of interoperability;
• exclusionary ranking;
• discriminatory API access;
• exploitative data practices;
• leveraging from one digital market into another.

C. Merger control

Computational societies also create difficult merger questions.

A transaction may involve a relatively small company with:

• valuable datasets;
• an important AI model;
• a strategic algorithm;
• a promising technology;
• a critical API; or
• a highly specialised user base.

Traditional turnover-based thresholds may fail to capture the competitive significance of such acquisitions.

Consequently, killer acquisitions and innovation competition become important components of computational competition governance.

6. Important Case Laws

1. Google Shopping — Google and Alphabet v European Commission, C-48/22 P

Court: Court of Justice of the European Union
Year: 2024

This is one of the most important modern competition cases for computational markets.

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

The Court of Justice upheld the €2.4 billion fine and rejected Google's appeal.

Importance

The case demonstrates that an algorithmically controlled ranking system can become an instrument of exclusion.

The significance extends beyond search engines:

Control over algorithmic visibility can itself constitute an important source of market power.

It is therefore relevant to:

• search engines;
• marketplaces;
• app stores;
• social media;
• travel platforms;
• food-delivery platforms; and
• AI-generated search systems.

2. Umar Javeed & Others v Google LLC — CCI

Competition Commission of India
Case No. 39/2018

This Indian case concerned Google's Android ecosystem.

The CCI examined Google's agreements and restrictions involving Android and related services. The CCI ultimately found Google dominant in several relevant markets and identified restrictions affecting competition.

The CCI imposed a penalty of approximately ₹1,337.76 crore in its October 2022 decision.

Computational-society significance

Android demonstrates how an operating system can operate simultaneously as:

technical infrastructure + platform + distribution system + data ecosystem.

Competition governance therefore cannot examine each component in isolation.

The relevant question becomes whether control over one computational layer allows a firm to reinforce dominance in adjacent layers.

3. Matrimony.com Ltd. v Google LLC & Others

CCI Case Nos. 07/2012 and 30/2012
Decision: 31 January 2018

The CCI examined Google's conduct relating to online search and search advertising.

The case is significant because search engines are not ordinary markets: the platform determines the algorithmic presentation and visibility of competing businesses.

The CCI's official case record identifies the matter as an antitrust proceeding under Section 19(1)(a).

Significance

The case illustrates the competition-law importance of:

• search ranking;
• specialised search;
• advertising placement;
• platform neutrality; and
• discrimination between the platform's own services and competitors.

It helped establish an important foundation for later debates concerning algorithmic self-preferencing.

4. Samir Agarwal v Competition Commission of India & Others

Supreme Court of India
2020

The case concerned allegations of price-fixing involving Ola and Uber.

The Supreme Court considered whether the CCI was required to investigate alleged anti-competitive conduct relating to algorithmically influenced pricing. The Court ultimately dealt with the statutory standing and scope of the proceedings and dismissed the appeal in the circumstances of the case.

Computational significance

The case is particularly useful for studying algorithmic pricing.

A platform's pricing algorithm may simultaneously:

• collect information;
• determine prices;
• respond to competitors;
• optimise revenue; and
• influence consumer behaviour.

This creates a fundamental evidentiary question:

When does automated pricing merely constitute legitimate optimisation, and when does it facilitate unlawful coordination?

The case therefore occupies an important place in the Indian discussion of algorithmic competition.

5. United States v Google LLC — Search Distribution Case

United States District Court for the District of Columbia
2024–2025 proceedings

The U.S. government challenged Google's practices concerning search distribution and default arrangements.

The case illustrates how computational markets may involve control over distribution channels and default settings, rather than simply the underlying quality of the product.

The DOJ maintains the case record and subsequent remedy proceedings.

Computational significance

Defaults can determine:

• which search engine consumers use;
• which applications receive traffic;
• which datasets are generated;
• which advertising ecosystem benefits; and
• which competitors obtain scale.

Thus:

Default position → user acquisition → data accumulation → algorithmic improvement → greater scale

can become a reinforcing competitive cycle.

6. United States v Google LLC — Digital Advertising

U.S. District Court for the Eastern District of Virginia
Decision: 2025

The court found Google liable for monopolising parts of the open-web digital advertising market. The DOJ described the judgment as involving Google's control over digital advertising markets and its effect on publishers and the competitive process.

Subsequent proceedings have focused heavily on remedies and interoperability with competing advertising technologies.

Computational significance

Digital advertising is an archetypal computational market because automated systems determine:

• advertising auctions;
• ad placement;
• pricing;
• publisher revenue;
• matching of advertisers and consumers; and
• allocation of information.

The case demonstrates that market architecture itself can become the object of competition-law scrutiny.

7. Meta Platforms / Bundeskartellamt

Court of Justice of the European Union — C-252/21
2023

The case concerned Meta's collection and combination of personal data across services and the interaction between competition law and data protection law.

It is important for computational societies because personal data may constitute a competitive resource.

Significance

The case demonstrates the increasing relationship between:

competition law + data governance + privacy + platform power.

A computational society cannot treat data merely as a privacy issue. In certain circumstances, data access and data accumulation can also influence competitive conditions.

8. Epic Games v Apple

U.S. federal litigation

The dispute concerned Apple's App Store ecosystem, including distribution restrictions, payment mechanisms and Apple's control over app distribution.

The broader significance is the concept of ecosystem governance.

An app-store operator can simultaneously determine:

• who enters the ecosystem;
• which payment mechanisms are available;
• which applications can reach users;
• what fees developers pay; and
• what technical rules developers must follow.

This illustrates why computational competition governance increasingly examines private rule-making by platforms.

7. Algorithmic Collusion

Algorithmic collusion can be divided into four categories.

Type I — Human-directed algorithmic collusion

Human competitors agree to coordinate and use software to implement their agreement.

This is comparatively straightforward.

Type II — Hub-and-spoke algorithmic coordination

A common platform or pricing intermediary supplies algorithms to competing firms.

The question becomes whether the common algorithm facilitates coordinated behaviour.

Type III — Algorithmic monitoring

Competitors independently use algorithms to monitor one another and automatically punish deviations.

Type IV — Autonomous emergent coordination

Independent AI systems learn that coordinated behaviour produces higher returns and converge toward similar strategies without an express human agreement.

The fourth category presents the greatest doctrinal difficulty.

8. The Attribution Problem

Competition law generally regulates undertakings and their conduct, not machines as independent legal persons.

Therefore:

AI chooses price → who is legally responsible?

The answer may depend upon:

• who designed the algorithm;
• who trained it;
• what objectives were programmed;
• what data were supplied;
• whether managers knew of the outcomes;
• whether the firm monitored the system;
• whether corrective mechanisms existed;
• whether the algorithm was intentionally designed to coordinate; and
• whether the undertaking benefited from the resulting conduct.

This suggests a developing principle of:

Algorithmic responsibility through organisational control.

9. Competition Governance of AI Systems

A computational competition framework should require appropriate governance throughout the AI lifecycle.

Stage 1 — Design

Competition risks should be assessed before deployment.

Stage 2 — Training

Training data should be examined for:

• competitively sensitive information;
• discriminatory inputs;
• exclusionary datasets;
• competitor information.

Stage 3 — Deployment

The system should be monitored for:

• coordinated pricing;
• exclusion;
• discriminatory ranking;
• self-preferencing.

Stage 4 — Continuous learning

Machine-learning systems can change behaviour after deployment.

Consequently, compliance cannot stop at initial certification.

Stage 5 — Audit

High-risk systems may require:

• independent audits;
• logging;
• explainability;
• documentation;
• testing;
• red-team exercises; and
• competition-impact assessments.

10. Data as a Competitive Infrastructure

Data should increasingly be analysed through three questions.

A. Who owns or controls the data?

B. Who can access it?

C. Can competitors realistically reproduce it?

Where data is:

• unique;
• commercially valuable;
• difficult to replicate; and
• necessary for effective competition,

refusal or discriminatory access may raise competition concerns.

This is particularly important in:

• AI;
• fintech;
• healthcare;
• autonomous vehicles;
• cloud computing;
• digital advertising;
• e-commerce; and
• smart infrastructure.

11. Interoperability

Interoperability is another major component of computational competition governance.

A dominant platform can potentially limit competition by preventing rivals from interacting with:

• APIs;
• operating systems;
• payment systems;
• messaging networks;
• cloud environments;
• identity systems;
• databases.

The policy logic is:

Interoperability → lower switching costs → easier entry → greater contestability.

However, interoperability requirements must also consider:

• cybersecurity;
• privacy;
• intellectual property;
• system integrity; and
• legitimate technical restrictions.

12. Computational Gatekeepers

A computational gatekeeper is an undertaking capable of controlling access to an important digital ecosystem.

Typical examples include operators of:

• search engines;
• app stores;
• operating systems;
• marketplaces;
• social networks;
• cloud infrastructure;
• advertising exchanges;
• payment networks.

The EU's DMA represents a major example of this governance model. The Commission designated six gatekeepers in 2023 and has subsequently expanded and modified the designated services.

13. Ex-Ante and Ex-Post Governance

Ex-post model

Competition authority acts after harmful conduct occurs.

Examples:

• investigation;
• infringement finding;
• fine;
• damages;
• behavioural remedy.

Ex-ante model

Rules apply before competitive harm materialises.

Examples:

• interoperability requirements;
• restrictions on self-preferencing;
• anti-steering rules;
• data portability;
• access obligations;
• merger notification obligations.

The computational economy increasingly requires a hybrid model.

14. Competition Governance and Automated Public Systems

Computational societies extend beyond private platforms.

Governments increasingly use algorithms for:

• procurement;
• allocation of public resources;
• infrastructure;
• taxation;
• welfare;
• healthcare;
• transport;
• public-sector purchasing.

Competition concerns may arise where public computational systems:

• favour incumbent suppliers;
• exclude smaller firms;
• create discriminatory access;
• lock government into one technology;
• prevent interoperability.

Therefore competition governance should apply not only to private markets but also to digital public infrastructure where procurement and market access are affected.

15. Remedies for Computational Competition Problems

Traditional fines may be insufficient where the underlying problem is architectural.

Possible remedies include:

Structural remedies

• divestiture;
• separation of business units;
• separation of infrastructure and downstream services.

Behavioural remedies

• non-discrimination;
• prohibition of self-preferencing;
• fair ranking;
• anti-steering obligations.

Technical remedies

• API access;
• interoperability;
• data portability;
• access protocols.

Governance remedies

• independent compliance officers;
• algorithmic audits;
• technical monitoring committees;
• reporting obligations;
• algorithmic impact assessments.

Recent U.S. digital-advertising proceedings illustrate this movement toward technically detailed remedies, including interoperability and non-discrimination requirements.

16. Competition Compliance for Computational Enterprises

A computational enterprise should maintain a Competition-by-Design Framework.

Step 1

Identify markets affected by algorithms.

Step 2

Identify algorithms capable of influencing competitive outcomes.

Step 3

Map data flows.

Step 4

Identify competitors' data received by the system.

Step 5

Test for algorithmic coordination.

Step 6

Test self-preferencing.

Step 7

Test discriminatory access.

Step 8

Test interoperability restrictions.

Step 9

Create audit logs.

Step 10

Establish human oversight.

Step 11

Conduct periodic competition audits.

Step 12

Maintain emergency intervention procedures.

17. Role of Competition Authorities

Competition authorities in computational societies require capabilities beyond traditional legal investigation.

They increasingly need:

• data scientists;
• economists;
• AI specialists;
• software engineers;
• forensic technology experts;
• cybersecurity specialists.

An effective investigation may require examination of:

• source code;
• model architecture;
• training datasets;
• API logs;
• pricing records;
• recommendation outputs;
• ranking changes;
• system instructions;
• internal communications.

Thus, digital evidence becomes central to competition enforcement.

18. Evidentiary Challenges

Computational competition cases may involve the problem of black-box decision-making.

A regulator may observe:

Competitors' prices repeatedly converge.

But convergence alone does not establish a cartel.

The authority may need to determine:

1. what the algorithms were instructed to do;
2. what information they received;
3. whether they communicated indirectly;
4. whether managers knew of the behaviour;
5. whether the outcome was predictable;
6. whether the algorithm rewarded coordination;
7. whether the system punished deviation.

Therefore, computational competition law requires sophisticated causal and technical evidence.

19. Relationship Between Competition Law and AI Governance

Competition law should not become a general AI-regulation statute.

Nevertheless, the two fields increasingly intersect.

20. Emerging Doctrine: Computational Essential Facilities

Traditional essential-facilities reasoning may become relevant to certain computational infrastructure.

Potential examples could include:

• critical APIs;
• cloud infrastructure;
• dominant app stores;
• interoperability protocols;
• digital identity infrastructure;
• important datasets;
• payment rails.

But not every important digital resource should automatically qualify as an essential facility.

Authorities must carefully examine:

• indispensability;
• duplication feasibility;
• economic viability;
• legitimate business justification;
• technical feasibility;
• competitive effects.

21. Computational Society and Consumer Welfare

Consumer welfare must also be understood differently.

Consumers may receive a service at zero monetary price while paying through:

• personal data;
• attention;
• reduced privacy;
• switching costs;
• reduced choice;
• behavioural manipulation.

Accordingly, computational competition analysis may need to examine quality, privacy, innovation and choice, in addition to price.

22. Innovation Competition

Computational markets are frequently characterised by rapid innovation.

A dominant company may harm competition not by raising prices but by:

• acquiring potential competitors;
• restricting access to its ecosystem;
• preventing interoperability;
• controlling distribution;
• limiting innovation by rivals.

Competition law therefore needs to protect future competitive possibilities, not merely existing market shares.

23. Six Core Principles of Competition Governance in Computational Societies

A coherent framework can be built around six principles:

1. Contestability

Markets must remain open to effective entry.

2. Algorithmic neutrality

Algorithms should not unlawfully discriminate against competing services.

3. Data fairness

Control over essential competitive data should not unnecessarily foreclose rivals.

4. Interoperability

Dominant systems should not use technical architecture to exclude competitors without legitimate justification.

5. Accountability

Businesses remain responsible for competition-relevant automated systems.

6. Auditability

High-impact computational systems should be capable of meaningful regulatory examination.

24. Key Case-Law Principles — Consolidated

These cases collectively demonstrate that computational competition problems arise at several different levels: pricing, ranking, distribution, data, infrastructure, interoperability and ecosystem governance.

25. Future Challenges

The next generation of competition law will face increasingly autonomous systems.

Particularly difficult questions include:

A. Autonomous AI cartels

Can independently learning systems create legally attributable coordination?

B. AI agents negotiating with AI agents

If autonomous purchasing agents interact with autonomous seller agents, who is responsible for the resulting market outcome?

C. Foundation-model concentration

Could control over compute, data and models create new forms of structural dominance?

D. AI-powered self-preferencing

Can an AI assistant systematically recommend its owner's products over competitors?

E. Algorithmic exclusion

Can an automated system gradually eliminate competitors without a traditional exclusionary strategy?

F. Computational mergers

Should acquisitions of strategically important datasets or AI capabilities be reviewed even where conventional turnover thresholds are not triggered?

G. Machine-speed competition

Can conventional investigation procedures keep pace with markets operating in milliseconds?

26. Conclusion

Competition law in computational societies is ultimately moving from regulation of individual business conduct toward governance of competitive systems.

The central transformation can be expressed as:

Traditional economy

Human decision → business conduct → market effect → competition-law response

Computational society

Data → algorithm → automated decision → network effect → ecosystem power → market effect

The Google Shopping litigation demonstrates the importance of algorithmic ranking and self-preferencing. The Indian Android proceedings demonstrate how an operating-system ecosystem can be used to reinforce power across connected markets. The Uber/Ola litigation illustrates the difficult relationship between algorithmically influenced pricing and traditional cartel concepts. Digital-advertising litigation further shows that competition remedies may increasingly need to address technical architecture and interoperability, rather than merely impose monetary penalties.

Accordingly, the future of competition governance in computational societies is likely to depend upon a combination of:

antitrust law + algorithmic accountability + data governance + interoperability + technical auditing + ex-ante platform regulation + conventional competition enforcement.

The fundamental legal principle remains that automation does not remove the responsibility of the undertaking controlling the computational system. The more deeply algorithms govern markets, the more important it becomes for competition law to govern not merely what firms sell, but also how computational infrastructures structure the conditions under which everyone else can compete.