Competition Law And Competition Implications Of Predictive Societies

1. Introduction

A predictive society is a society in which businesses, governments, platforms and other institutions increasingly use large amounts of data, artificial intelligence, algorithms and statistical models to predict behaviour, demand, prices, risks, preferences and future events.

Examples include:

predicting consumer purchasing behaviour;

predicting demand and supply;

personalised advertising;

credit and insurance risk prediction;

predictive pricing;

algorithmic recommendations;

predicting employee or supplier behaviour;

forecasting traffic and logistics;

AI-based market intelligence;

predicting customer switching;

predictive search and ranking;

forecasting competitors' strategies.

From a competition-law perspective, predictive societies create an important economic question:

What happens when a small number of undertakings possess the data, computing infrastructure, algorithms and market access necessary to predict markets better than their competitors?

Competition law is generally not concerned with prediction itself. Predictive technology can produce substantial efficiencies and innovation. The concern arises where predictive capabilities become a source of market power, entry barriers, exclusion, coordination, discrimination or foreclosure.

Thus:

Data → Prediction → Better Decisions → Competitive Advantage → Network Effects → Market Power → Potential Foreclosure

2. Meaning of Predictive Societies

A predictive society has several important characteristics.

A. Data-intensive markets

Businesses collect information concerning:

customers;

transactions;

searches;

location;

purchasing behaviour;

product usage;

prices;

competitors;

suppliers;

market conditions.

B. Algorithmic prediction

Algorithms transform historical and real-time information into predictions concerning:

future demand;

consumer preferences;

prices;

customer churn;

competitor behaviour;

fraud;

supply conditions;

market trends.

C. Continuous feedback

Predictions generate decisions, and those decisions generate additional data.

For example:

Search → Recommendation → Consumer choice → New data → Better recommendation → More users → More data

This creates a feedback loop.

D. Concentration of predictive capability

If one platform has significantly more data, computing resources and user interactions than competitors, its predictive models may become substantially more powerful.

This can create a competitive advantage that is difficult for new entrants to reproduce.

3. Predictive Society And Competition Law

Competition law generally protects the competitive process rather than guaranteeing identical predictive capabilities to all businesses.

Predictive technology may therefore be:

Pro-competitive

It can:

reduce costs;

improve logistics;

reduce waste;

improve product quality;

match consumers with products;

improve forecasting;

reduce fraud;

encourage innovation.

Potentially anti-competitive

Problems may arise where predictive capability is used to:

exclude competitors;

discriminate against rivals;

foreclose market access;

tie customers to an ecosystem;

prevent switching;

facilitate collusion;

exploit data advantages;

strengthen an existing dominant position;

discriminate between business partners;

manipulate rankings or recommendations.

4. Major Competition Concerns Created By Predictive Societies

4.1 Data Concentration

The first major issue is concentration of data.

A dominant digital undertaking may have access to:

search data;

transaction data;

behavioural data;

location data;

advertising data;

customer histories;

product interactions.

A new entrant may technically be able to create an algorithm but lack the data necessary to train it effectively.

Therefore:

Data concentration → Predictive advantage → Better service → More users → More data

This can create a self-reinforcing competitive cycle.

Competition concern

The relevant question is not merely whether the company possesses data, but whether its data advantage:

creates substantial market power;

raises barriers to entry;

prevents effective competition; and

is reinforced through exclusionary conduct.

5. Predictive Capability As A Barrier To Entry

Traditional barriers to entry include:

capital requirements;

patents;

infrastructure;

licences;

distribution networks.

Predictive societies introduce another barrier:

Access to high-quality data and predictive capability.

A new competitor may need:

millions of observations;

historical transaction data;

consumer interaction data;

computing resources;

specialised engineers;

AI models;

feedback from users.

The incumbent may already possess all of these.

Consequently, competition can become increasingly difficult even where the underlying technology is theoretically available to everyone.

6. Network Effects And Predictive Feedback Loops

Predictive markets can exhibit powerful network effects.

Consider an online platform:

More users → More data → Better prediction → Better service → More users

This is a positive feedback loop.

A smaller competitor may therefore face a structural disadvantage.

This does not automatically constitute an infringement of competition law. However, where a dominant undertaking uses exclusionary strategies to reinforce the feedback loop, competition authorities may investigate.

7. Predictive Pricing

Predictive systems can be used to determine prices according to:

consumer willingness to pay;

demand forecasts;

competitor prices;

inventory;

time;

location;

customer behaviour.

Dynamic pricing itself is generally not unlawful.

However, competition concerns may arise where algorithms are used for:

price coordination;

personalised exclusion;

discriminatory treatment;

monitoring competitors;

facilitating collusion.

8. Algorithmic Collusion

Predictive societies create a particularly important issue concerning algorithmic coordination.

Suppose competing companies use algorithms that continuously:

observe competitors' prices;

predict their reactions;

adjust prices automatically.

Even without a traditional meeting between executives, sophisticated algorithms may make coordination easier.

Competition authorities must therefore examine whether algorithmic systems:

implement an existing agreement;

facilitate communication;

exchange competitively sensitive information;

coordinate prices;

reduce uncertainty concerning competitors' behaviour.

Important distinction

Algorithmic pricing does not automatically equal cartel conduct.

There must be a legally relevant basis for liability under the applicable competition law.

9. Predictive Market Intelligence

Large firms may possess superior information about:

competitors;

consumers;

suppliers;

prices;

demand;

product launches.

This creates predictive market intelligence.

If the dominant undertaking uses that intelligence to identify emerging competitors and strategically respond to them, competition authorities may examine whether the conduct amounts to exclusionary behaviour.

The information advantage itself is not necessarily unlawful.

The competition concern arises from the use of that advantage.

10. Predictive Self-Preferencing

A vertically integrated platform may use predictive information to identify:

products likely to become successful;

sellers likely to attract customers;

emerging competitors;

profitable market segments.

It may then introduce its own competing products or give its products favourable treatment.

This may raise concerns involving:

self-preferencing;

leveraging;

discriminatory access;

foreclosure.

However, self-preferencing is not automatically unlawful. The legal analysis depends upon the applicable jurisdiction, market power, conduct and effects.

11. Predictive Advertising

Digital advertising systems use predictions about:

consumer interests;

purchasing intentions;

likelihood of conversion;

customer value;

future behaviour.

A dominant advertising ecosystem possessing significantly greater behavioural data may have an advantage over competing advertising providers.

Competition concerns may include:

tying;

self-preferencing;

exclusionary data practices;

discriminatory access;

foreclosure of rival advertising intermediaries.

12. Predictive Consumer Lock-In

Predictive systems become better as consumers use them.

A consumer may therefore become increasingly dependent on:

recommendations;

personalised search;

automated purchasing;

stored preferences;

digital assistants;

ecosystem services.

Switching to another provider may result in a loss of:

historical data;

personalised recommendations;

settings;

reputation;

transaction history.

This creates switching costs.

Competition law may become relevant where a dominant undertaking deliberately creates or exploits such switching barriers through exclusionary practices.

13. Predictive Societies And Refusal To Deal

One of the difficult questions is whether a dominant company should be required to provide competitors with access to:

data;

APIs;

predictive tools;

databases;

infrastructure.

Competition law generally does not impose a general duty to share every commercially valuable resource.

The European essential-facilities/refusal-to-deal cases demonstrate the exceptional nature of compulsory access.

14. Case Law

Case 1: Bronner v Mediaprint

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

Principle

The Court of Justice applied a strict approach to refusal-to-deal claims.

The case concerned access to a newspaper distribution system.

Relevance to predictive societies

A predictive ecosystem may become an important infrastructure for competitors.

For example:

data infrastructure;

cloud systems;

APIs;

search access;

prediction infrastructure.

Bronner demonstrates that importance alone does not automatically create a legal obligation to provide access.

Key lesson

A commercially valuable infrastructure is not automatically an essential facility requiring compulsory access.

15. Case 2: Magill

Cases: Joined Cases C-241/91 P and C-242/91 P, RTE and ITP v Commission

Principle

The Court recognised exceptional circumstances in which refusal to license intellectual property could constitute an abuse of dominance.

Relevance

Predictive societies depend heavily upon:

databases;

information;

software;

analytical systems;

intellectual property.

The case is relevant where control over information prevents the emergence of a new product or competitive opportunity.

Key lesson

Control over information or intellectual property can become a competition concern where exceptional conditions for compulsory access are satisfied.

16. Case 3: IMS Health

Case: IMS Health GmbH & Co OHG v NDC Health GmbH & Co KG, Case C-418/01

Principle

The case developed the exceptional circumstances surrounding refusal to license intellectual property.

Relevance

The case is highly relevant to data-driven predictive markets because commercially valuable information structures can become important competitive inputs.

However, not every proprietary database must be made available to competitors.

Key lesson

Data or information infrastructure may raise competition concerns when control over it satisfies the strict conditions governing exceptional refusal-to-license cases.

17. Case 4: Microsoft

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

Principle

Microsoft was found to have abused its dominant position in relation to interoperability information and tying practices.

Relevance to predictive societies

Modern predictive ecosystems frequently depend upon interoperability between:

platforms;

applications;

APIs;

operating systems;

databases;

cloud systems;

AI services.

A dominant firm may potentially use control over an important technological layer to restrict competitors in neighbouring markets.

Key lesson

Control over an important technological infrastructure can become an abuse-of-dominance concern where it is used to restrict effective competition in related markets.

18. Case 5: Intel

Case: Intel Corp. v Commission, Case C-413/14 P

Principle

The case concerned conditional rebates and the assessment of exclusionary effects.

The Court required attention to the actual or potential ability of the conduct to foreclose competitors.

Relevance to predictive societies

Dominant predictive platforms may offer:

preferential access;

discounts;

rebates;

contractual benefits;

ecosystem incentives.

These practices may become problematic when used to exclude equally efficient competitors.

Key lesson

Competition analysis must examine the actual competitive effects of exclusionary practices rather than relying solely on formal labels.

19. Case 6: Google Shopping

Case: Google and Alphabet v Commission, Case C-48/22 P

Principle

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

The Court upheld the finding of abuse concerning practices that favoured Google's own comparison-shopping service and disadvantaged competing services.

Relevance to predictive societies

Predictive platforms increasingly determine:

rankings;

recommendations;

visibility;

search results;

product placement.

Control over predictive visibility can determine which businesses receive consumer attention.

Key lesson

Control over an important digital access point can create competition concerns where a dominant undertaking uses that position to advantage its own service and restrict competing services.

20. Case 7: Google Android

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

Principle

The case concerned Google's contractual practices involving Android, including requirements concerning app stores, search and browser distribution.

Relevance

Predictive ecosystems often combine:

operating systems;

search;

advertising;

app stores;

data;

user accounts.

A firm controlling one technological layer may be able to reinforce its position in another market.

Key lesson

Bundling and contractual arrangements within a digital ecosystem can reinforce market power across connected markets.

21. Case 8: Deutsche Telekom

Case: Deutsche Telekom AG v Commission, Case C-280/08 P

Principle

The case concerned margin squeeze and the ability of a vertically integrated dominant undertaking to disadvantage downstream competitors.

Relevance

Predictive societies often contain vertically integrated structures:

Infrastructure → Data → Analytics → Platform → Consumer

A dominant undertaking may potentially use control over an upstream infrastructure layer to disadvantage downstream competitors.

Key lesson

Vertical control combined with market power can create foreclosure risks where rivals cannot compete effectively on downstream markets.

22. Case 9: Slovak Telekom

Cases: Slovak Telekom a.s. v Commission, Cases C-165/19 P and C-166/19 P

Principle

The case concerned exclusionary conduct involving access to telecommunications infrastructure.

Relevance

Predictive societies require physical and digital infrastructure such as:

telecommunications networks;

cloud infrastructure;

data centres;

APIs;

computing infrastructure.

Control over such infrastructure can affect competitors' ability to enter or expand.

Key lesson

Infrastructure control can become a competition issue when dominant firms use it to foreclose downstream competition.

23. Case 10: United Brands

Case: United Brands Company v Commission, Case 27/76

Principle

The case is a foundational authority on abuse of dominance and the concept of dominant position.

Relevance

Predictive businesses may acquire significant market power through:

brand strength;

data;

network effects;

customer dependence;

infrastructure;

ecosystem advantages.

United Brands demonstrates that dominance concerns depend upon the economic structure and competitive conditions of the relevant market.

Key lesson

Market power must be assessed through the economic characteristics of the relevant market rather than simply by observing technological sophistication.

24. Predictive Societies And Dominance

Predictive capabilities can contribute to dominance through several mechanisms:

However:

Predictive capability ≠ dominance automatically.

The relevant question is whether the undertaking possesses substantial market power in the relevant market.

25. Predictive Societies And Abuse Of Dominance

Potential abuses include:

1. Self-preferencing

Using predictive systems to favour the dominant firm's own products.

2. Exclusive arrangements

Using predictive insights to secure exclusivity from suppliers or distributors.

3. Refusal to provide access

Restricting essential or exceptionally important predictive infrastructure.

4. Discriminatory access

Giving different businesses different access to:

data;

APIs;

ranking;

advertising;

infrastructure.

5. Tying

Requiring users to adopt one predictive service with another dominant service.

6. Bundling

Combining predictive products with infrastructure or platform services.

7. Margin squeeze

Using upstream infrastructure control to disadvantage downstream competitors.

8. Predatory strategies

Using financial or ecosystem resources to eliminate emerging competitors.

26. Predictive Societies And Merger Control

Predictive markets create important merger-control questions.

A dominant company may acquire:

AI startups;

data companies;

analytics firms;

cloud providers;

predictive software companies;

potential future competitors.

A transaction may raise concerns even when the target currently has relatively low revenue if it possesses:

valuable technology;

unique datasets;

innovative capabilities;

strategic talent;

potential competitive significance.

Therefore competition authorities may examine the future competitive significance of transactions.

27. Killer Acquisitions

A predictive ecosystem may allow a large company to identify promising emerging firms early.

It could potentially acquire a startup before that startup becomes a significant competitor.

This creates a competition concern commonly described as a killer acquisition problem.

The central question is:

Would the target have developed into a meaningful competitive constraint if it had remained independent?

The analysis is particularly important in:

AI;

digital advertising;

fintech;

health technology;

cloud services;

data analytics.

28. Predictive Societies And Innovation

Predictive systems may have two opposite effects.

Positive effect

They can:

improve R&D;

reduce waste;

identify consumer needs;

accelerate innovation;

improve resource allocation.

Negative effect

Concentration of predictive capability may:

discourage entry;

reduce independent innovation;

allow incumbents to identify threats early;

permit rapid imitation;

discourage investment by startups.

Competition authorities therefore need to consider dynamic competition, not merely current market shares.

29. Predictive Societies And Consumer Choice

Prediction can improve consumer choice by:

recommending relevant products;

reducing search costs;

identifying cheaper alternatives.

But excessive dependence on predictive systems may also reduce consumer autonomy.

For competition law, the central issue is whether:

recommendations favour the platform's own products;

rivals are artificially downgraded;

consumers cannot easily switch;

competing suppliers lose access to consumers.

30. Predictive Societies And Small Businesses

Small businesses can become dependent upon predictive platforms for:

advertising;

search ranking;

customer acquisition;

pricing;

logistics;

payments;

cloud computing.

If a platform changes its algorithm, small businesses may experience significant changes in customer access.

This creates an important competition-law question:

Does control over predictive access give the platform the ability to substantially restrict competitors' opportunities to compete?

31. Predictive Societies And India

In India, the principal legal framework is the Competition Act, 2002.

Section 3

Section 3 addresses agreements that cause or are likely to cause an appreciable adverse effect on competition.

Predictive systems may become relevant where competing undertakings use algorithms or data systems to facilitate coordination.

Section 4

Section 4 addresses abuse of dominant position.

Predictive-market issues may arise through:

unfair or discriminatory conditions;

denial of market access;

leveraging;

tying;

discriminatory treatment;

exclusionary conduct.

Section 5

Section 5 is relevant to combinations and merger control.

AI, data and predictive technology acquisitions may raise concerns concerning:

future competition;

innovation;

technology;

data concentration;

ecosystem effects.

32. Predictive Societies And The Concept Of Market Access

One of the most important competition questions is access to customers.

In traditional markets:

Manufacturer → Distributor → Consumer

In predictive digital markets:

Data → Algorithm → Ranking/Recommendation → Consumer Attention → Transaction

The algorithm may therefore become a gateway to market access.

If a dominant platform controls that gateway, exclusion from its predictive system can potentially have major competitive consequences.

33. Predictive Societies And Opportunity Concentration

Predictive societies can create opportunity concentration.

Opportunity concentration occurs where a limited number of undertakings control the important pathways through which businesses can:

reach consumers;

obtain data;

advertise;

distribute products;

obtain infrastructure;

innovate;

access financing;

develop partnerships.

This means competition can be weakened even without complete monopoly over the underlying product market.

34. Predictive Societies And Algorithmic Transparency

Competition authorities may increasingly need to understand:

how algorithms rank products;

how prices are generated;

how recommendations are produced;

how competitors are treated;

how data is collected;

whether ranking systems favour affiliated businesses.

However, competition law does not necessarily require complete disclosure of proprietary algorithms.

The objective is to determine whether the algorithm is being used in a manner that unlawfully restricts competition.

35. Predictive Societies And Algorithmic Discrimination

A dominant platform may theoretically provide different:

prices;

rankings;

access;

visibility;

advertising opportunities

to different businesses.

Discrimination becomes a competition concern particularly where it:

disadvantages competing firms;

favours affiliated businesses;

restricts market access;

exploits dependence on the dominant platform.

36. Predictive Societies And Tacit Coordination

Predictive technology may reduce uncertainty between competitors.

Algorithms can observe:

prices;

capacity;

inventory;

promotions;

consumer demand.

If firms can predict competitors' responses with great accuracy, competitive rivalry may be weakened.

The important legal distinction is:

Independent adaptation

A company independently responds to market conditions.

Illegal coordination

Competitors engage in conduct that satisfies the relevant legal test for an agreement, concerted practice, or other prohibited coordination.

Competition authorities therefore need to distinguish parallel algorithmic behaviour from legally prohibited coordination.

37. Predictive Societies And Privacy

Privacy law and competition law are different areas, but they may interact.

A dominant company may offer services in exchange for extensive personal data.

Competition authorities may examine whether:

data practices strengthen market power;

consumers face reduced choice;

competitors cannot replicate the data advantage;

privacy-related quality competition has deteriorated.

Therefore, quality can be an important competitive parameter in data-driven markets.

38. Predictive Societies And Consumer Welfare

Consumer welfare may involve more than price.

Relevant competitive dimensions include:

price;

quality;

privacy;

innovation;

choice;

security;

convenience.

A predictive service may be free in monetary terms while competition concerns arise through reduced:

privacy;

choice;

quality;

innovation.

39. Predictive Societies And Essential Facilities

A predictive resource could potentially become an essential input where it is:

indispensable;

practically impossible to duplicate;

necessary for effective competition;

controlled by a dominant undertaking.

But competition law generally applies a high threshold before compelling access to privately controlled resources.

Bronner, Magill and IMS Health illustrate different aspects of this principle.

40. Competition Risks At Different Stages

41. Regulatory Approach

Competition authorities examining predictive societies should consider:

Step 1: Define the relevant market

Determine:

product/service market;

geographic market;

digital ecosystem boundaries.

Step 2: Assess market power

Examine:

market shares;

data;

network effects;

switching costs;

entry barriers;

infrastructure.

Step 3: Identify the predictive capability

Determine:

what is being predicted;

what data is used;

who controls it;

whether competitors can replicate it.

Step 4: Identify the conduct

Possible conduct:

tying;

bundling;

exclusivity;

discrimination;

self-preferencing;

refusal to deal;

predatory conduct.

Step 5: Assess foreclosure

Ask:

Are competitors prevented or materially disadvantaged from competing?

Step 6: Examine efficiencies

Consider:

innovation;

cost savings;

improved quality;

consumer benefits.

Step 7: Examine dynamic effects

Consider:

future competition;

innovation;

entry;

technological development.

42. Important Legal Principle

The central competition-law distinction is:

Predictive superiority is not unlawful merely because it is superior.

A firm may lawfully become successful because it:

collects data efficiently;

develops better algorithms;

innovates;

offers better products;

invests in infrastructure.

Competition concerns arise where predictive power is combined with conduct that unlawfully:

excludes competitors;

restricts market access;

facilitates prohibited coordination;

reinforces dominance;

forecloses innovation.

43. Summary Of Major Case Laws

44. Key Competition-Law Implications

The principal implications of predictive societies can be summarised as follows:

1. Data concentration

Large datasets can strengthen existing market power.

2. Predictive advantage

Better prediction can create a competitive advantage.

3. Entry barriers

New firms may lack sufficient data and infrastructure.

4. Network effects

More users generate more data and better predictions.

5. Switching costs

Personalised ecosystems can make switching more difficult.

6. Self-preferencing

Predictive systems can potentially favour affiliated products.

7. Foreclosure

Control over predictive infrastructure can restrict rivals.

8. Algorithmic coordination

Algorithms may increase the ability of competitors to observe and respond to each other.

9. Merger concerns

Acquisitions can concentrate data, technology and future competitive capabilities.

10. Innovation concerns

Excessive concentration may reduce incentives for independent innovation.

45. Short Exam Answer

Predictive societies are societies in which data, algorithms and artificial intelligence are extensively used to predict consumer behaviour, demand, prices, risks and market developments.

From a competition-law perspective, predictive systems can produce efficiencies and innovation but can also create market-power concerns. Large firms may obtain significant advantages from accumulated data, network effects, computing infrastructure and superior predictive algorithms. These advantages can create entry barriers and switching costs.

Competition concerns may arise where dominant undertakings use predictive capabilities for self-preferencing, discriminatory access, tying, bundling, exclusive dealing, refusal to provide essential access, margin squeeze or foreclosure. Predictive algorithms may also facilitate coordination between competitors, although algorithmic parallel behaviour does not automatically establish an unlawful agreement.

Important cases include Bronner, Magill, IMS Health, Microsoft, Intel, Google Shopping, Google Android, Deutsche Telekom, Slovak Telekom and United Brands.

The central principle is that competition law does not prohibit predictive technology, data advantages or successful innovation as such. It intervenes where market power is acquired or maintained through legally prohibited exclusionary conduct or anti-competitive coordination.

46. Conclusion

Predictive societies represent an important development in modern competition economics.

The competitive advantage of the future may depend not only on owning factories, stores or physical infrastructure but also on owning:

data + algorithms + computing power + predictive capability + consumer access.

This creates a potential competitive feedback loop:

Data → Prediction → Better Service → More Users → More Data → Greater Predictive Power → Greater Market Power.

Competition law therefore needs to examine whether this cycle results from competition on the merits or is reinforced by exclusionary conduct.

The major legal questions concern:

data concentration;

predictive infrastructure;

algorithmic coordination;

self-preferencing;

market access;

switching costs;

interoperability;

refusal to deal;

vertical foreclosure;

acquisitions of emerging competitors;

innovation;

ecosystem power.

The most important conclusion is that prediction itself is not an antitrust violation. The competition-law issue is whether predictive capability becomes a source of durable market power and whether that power is used to prevent effective competition.

Quick Revision Formula

Predictive Societies → Data → Algorithms → Prediction → Network Effects → Entry Barriers → Switching Costs → Market Power → Potential Foreclosure/Coordination → Competition-Law Scrutiny.

One-Line Revision Point

Predictive societies create competition concerns when concentrated data, algorithms and predictive infrastructure give firms durable market power that is used to restrict entry, access, innovation or effective competition.