Algorithmic Innovation Prioritization and Ecosystem Steering

Introduction

Algorithmic innovation prioritization refers to the use of algorithms, artificial intelligence, data analytics, ranking systems, recommendation engines, or automated investment tools to determine which products, technologies, developers, suppliers, applications, or research projects receive visibility, resources, access, or commercial opportunities.

Ecosystem steering occurs when a powerful platform or vertically integrated undertaking uses such systems to influence the direction of an entire commercial ecosystem—for example, by determining which applications receive prominence in an app store, which sellers appear first in search results, which suppliers obtain access to customers, or which technologies become interoperable with a dominant platform.

Competition law becomes relevant where algorithmic prioritization is not merely an efficiency-enhancing innovation tool but is used by a dominant undertaking to favour its own products, exclude rivals, disadvantage complementary innovators, restrict interoperability, exploit data advantages, or make market entry and expansion more difficult.

The legal analysis generally falls under abuse of dominance, self-preferencing, tying and leveraging, discriminatory access, refusal to deal, exclusionary conduct, vertical restraints, and merger/control-of-innovation theories.

I. Meaning and Components

1. Algorithmic prioritization

An undertaking may use an algorithm to decide:

• which innovations receive funding;
• which applications receive ranking prominence;
• which sellers obtain preferred placement;
• which suppliers are recommended;
• which APIs or technical features receive development priority;
• which products are integrated into a platform;
• which third-party innovations are exposed to users;
• which technologies receive access to platform data;
• which developers obtain testing or certification resources.

The algorithm therefore becomes a resource-allocation mechanism.

2. Ecosystem steering

Ecosystem steering is broader. It concerns the ability of a platform or infrastructure provider to influence the development trajectory of complementary markets.

For example:

Platform → data → algorithm → ranking → consumer attention → developer incentives → innovation direction → greater platform dependence.

This can create a feedback loop in which successful complementary firms increasingly depend upon the platform's algorithmic decisions.

3. Competition-law concern

The central question is not whether an algorithm "chooses winners."

The relevant question is:

Does the undertaking possess market power enabling it to use algorithmic control over an ecosystem to restrict competition on the merits or to extend its dominance into adjacent markets?

II. Relevant Competition-Law Theories

1. Self-preferencing

A dominant platform may rank its own products or affiliated services more prominently than competing products.

The concern is particularly strong where the platform controls an important gateway to consumers.

Possible effects include:

• reduced visibility for rivals;
• lower traffic;
• reduced consumer discovery;
• reduced incentives to innovate;
• foreclosure of competing innovators.

The European Commission's Google Shopping decision is particularly important in this context.

III. Key Case Laws

1. Google Search (Shopping) – European Commission, 2017

The European Commission found that Google had abused its dominant position in general search by systematically giving prominent placement to its own comparison-shopping service while applying demotion mechanisms to competing comparison-shopping services.

Relevance

The case demonstrates how ranking algorithms can become instruments of competitive exclusion.

The important distinction is between:

• legitimate relevance-based ranking; and
• preferential treatment that disadvantages competing services.

Algorithmic innovation prioritization can raise a similar issue where a dominant platform systematically gives its own innovations superior placement.

Principle

A dominant undertaking controlling a major digital gateway cannot necessarily rely upon algorithmic neutrality as a justification when the design or application of the ranking mechanism structurally advantages its own service.

2. Google Android – European Commission, 2018

The European Commission found several practices involving Google's Android ecosystem to constitute abuses of dominance, including restrictions concerning search and browser applications and certain contractual arrangements.

Relevance to ecosystem steering

Android illustrates how control over an operating-system ecosystem can affect:

• application distribution;
• search services;
• browser competition;
• default settings;
• developer incentives.

The case demonstrates that competition analysis may extend beyond the immediate product to the architecture of the surrounding ecosystem.

Principle

Control of an ecosystem can provide a dominant undertaking with opportunities to leverage power from one market into neighbouring markets.

3. United States v. Microsoft Corp. (D.C. Cir. 2001)

The Microsoft litigation concerned Microsoft's conduct involving the Windows operating-system platform and competing technologies, particularly web browsers.

The court examined Microsoft's use of its operating-system position to disadvantage competing technologies.

Relevance

Microsoft is foundational for understanding platform control and innovation foreclosure.

A dominant technological platform may possess substantial power to determine which complementary technologies can reach users.

Principle

Competition law may intervene where control over an important technological platform is used to disadvantage competing technologies through exclusionary conduct.

4. European Commission v. Microsoft (Microsoft, 2004)

The European Commission addressed Microsoft's conduct concerning interoperability information and the integration of Windows Media Player with Windows.

The case involved both interoperability and tying-related concerns.

Relevance to algorithmic ecosystem steering

Modern algorithms can perform functions that historically were performed through contractual or technical decisions.

For example, instead of explicitly prohibiting a rival technology, a platform may:

• deprioritize it;
• restrict API access;
• reduce interoperability;
• delay certification;
• limit recommendation exposure.

The Microsoft decision therefore provides an important conceptual foundation for analysing technology ecosystems.

5. United States v. Apple Inc. – e-books

The Apple e-books litigation concerned arrangements between Apple and publishers relating to the sale and pricing of electronic books.

The Supreme Court's decision addressed Apple's role in facilitating coordination among publishers.

Relevance

The case illustrates that technology intermediaries can affect competitive conditions not merely through direct pricing but through ecosystem architecture and coordination mechanisms.

Algorithmic systems can similarly create mechanisms through which multiple market participants coordinate or align commercial behaviour.

Competition-law significance

Where algorithmic systems facilitate communication, monitoring, retaliation, or alignment among competitors, authorities may examine whether the technology is facilitating unlawful coordination rather than merely improving efficiency.

6. European Commission – Amazon Marketplace

The European Commission investigated Amazon's use of marketplace data generated by independent sellers and its possible use of that information in competing with those sellers.

The Commission also examined Amazon's marketplace practices concerning the Buy Box and Prime programme.

Relevance

This is highly relevant to algorithmic ecosystem steering because a marketplace operator may simultaneously act as:

1. infrastructure provider;
2. data collector;
3. ranking intermediary; and
4. competitor.

That combination can create a structural conflict.

Algorithmic concern

If marketplace algorithms use third-party seller data to:

• identify successful products;
• imitate successful innovations;
• optimize Amazon's own offerings; or
• determine competitive priorities,

the platform's informational advantage may become a competition concern.

IV. Additional Important Case Laws

7. United Brands v Commission (1978)

The Court of Justice examined abuse of dominance under Article 102 TFEU.

Although the case predates algorithmic markets, its broader importance lies in establishing principles concerning exclusionary conduct and the responsibilities of dominant firms.

Relevance

Algorithmic ecosystem steering should be examined against the broader principle that dominance creates particular responsibilities concerning competitive conduct.

8. Bronner v Mediaprint (1998)

The case concerned access to a newspaper distribution system.

The Court established a restrictive framework for when refusal to provide access to an essential facility can constitute abuse.

Relevance

Modern digital ecosystems can contain technologically important facilities such as:

• APIs;
• app stores;
• identity systems;
• cloud infrastructure;
• interoperability interfaces;
• data access mechanisms.

Algorithmic prioritization that effectively determines which competitors obtain access may therefore intersect with refusal-to-deal and essential-facility principles.

9. Slovak Telekom v Commission (2021)

The case concerned access to telecommunications infrastructure and exclusionary conduct.

Relevance

It demonstrates that competition law can examine the manner in which a dominant undertaking controls access to infrastructure needed by competitors.

In digital ecosystems, algorithmically controlled access can perform a similar economic function.

10. Intel v Commission (2017)

The Intel litigation concerned rebates offered by a dominant undertaking and the assessment of exclusionary effects.

The Court of Justice emphasized the importance of examining the actual or potential exclusionary effects of particular conduct where the undertaking provides evidence challenging the assumption of anticompetitive effects.

Relevance

For algorithmic prioritization, the analysis should therefore not stop at identifying preferential treatment.

Authorities may need to examine:

• coverage of the algorithm;
• duration;
• foreclosure capability;
• affected rivals;
• market shares;
• switching possibilities;
• efficiencies;
• actual effects.

V. How Algorithmic Innovation Prioritization Can Produce Competitive Harm

1. Innovation foreclosure

Suppose a dominant cloud platform develops an algorithm determining which new technologies receive integration resources.

If competing technologies are consistently given lower priority, they may become commercially unattractive even where they are technologically superior.

This creates:

Algorithmic prioritization → reduced adoption → reduced investment → weaker rivals → reduced competitive pressure.

2. Data-driven innovation displacement

A platform may possess data unavailable to competitors.

The platform can use this information to identify:

• emerging consumer preferences;
• rapidly growing products;
• successful third-party features;
• profitable customer segments.

It may then redirect its own innovation resources toward those areas.

This creates a potential data-to-innovation feedback loop.

VI. Algorithmic Steering and Self-Preferencing

A particularly important structure is:

Platform + Marketplace + Data + Algorithm + Own Product

The platform can determine:

1. which products consumers see;
2. which sellers obtain visibility;
3. which innovations receive recommendation;
4. which third parties receive data;
5. which services receive interoperability;
6. which products become default options.

This may create a vertically integrated competitive advantage.

The competition-law inquiry should therefore distinguish:

Legitimate prioritization

• based on objective technical criteria;
• improves quality;
• reduces fraud;
• improves security;
• responds to consumer demand;
• applies consistently.

Potentially problematic prioritization

• systematically favours affiliated products;
• discriminates against competing products;
• uses non-public competitor data;
• restricts interoperability;
• reduces rivals' visibility without legitimate justification;
• creates artificial switching costs.

VII. Innovation as a Parameter of Competition

Traditional competition analysis frequently examines:

• price;
• output;
• quality;
• consumer choice.

Digital markets require greater attention to innovation competition.

An algorithm can affect innovation competition by determining which firms obtain:

• users;
• data;
• capital;
• platform access;
• technical integration;
• visibility;
• developer attention.

Thus:

Algorithmic control over visibility can become indirect control over innovation incentives.

VIII. Ecosystem Feedback Loops

A dominant platform can develop a self-reinforcing system:

Stage 1 – Data accumulation

More users generate more data.

Stage 2 – Algorithmic improvement

More data improves ranking and recommendation systems.

Stage 3 – Better targeting

Improved algorithms attract more users and developers.

Stage 4 – Greater dependency

Third parties become increasingly dependent on the platform.

Stage 5 – Innovation steering

The platform gains greater influence over which complementary technologies succeed.

Stage 6 – Further data accumulation

The cycle repeats.

This can produce ecosystem entrenchment even where the platform does not expressly prohibit competitors.

IX. Competition Effects

Algorithmic innovation prioritization may potentially affect:

1. Market entry

New entrants may find it difficult to obtain algorithmic visibility.

2. Expansion

Existing competitors may be prevented from expanding into adjacent markets.

3. Innovation incentives

Competitors may reduce investment if access to users depends upon the dominant platform's ranking decisions.

4. Consumer choice

Consumers may see a narrower range of products.

5. Quality competition

The platform's own products may receive advantages unrelated to objective quality.

6. Investment

Venture capital and research investment may follow platform-controlled signals, reinforcing incumbent dominance.

X. Evidence Required to Establish Competitive Harm

Competition authorities would generally need more than proof that an algorithm favours certain innovations.

Relevant evidence may include:

• source-code or algorithm documentation;
• ranking criteria;
• internal communications;
• A/B testing records;
• product-development documents;
• platform-access statistics;
• changes in traffic;
• click-through rates;
• conversion data;
• developer complaints;
• internal strategic documents;
• treatment of affiliated and unaffiliated products;
• counterfactual ranking results;
• consumer behaviour;
• evidence of rival exclusion.

The counterfactual is especially important:

What would the ecosystem look like if the discriminatory algorithmic prioritization did not exist?

XI. Efficiency Defences

Algorithmic prioritization is not inherently unlawful.

A platform may legitimately prioritize products because of:

• security;
• reliability;
• privacy;
• technical compatibility;
• consumer safety;
• latency;
• fraud prevention;
• quality assurance;
• resource constraints.

Therefore, authorities must distinguish competition on the merits from exclusionary manipulation.

A useful analytical framework is:

Legitimate objective → algorithmic criterion → implementation → differential effect → competitive effect → justification → proportionality.

XII. Remedies

Where algorithmic prioritization produces an established competition problem, possible remedies may include:

1. Non-discrimination obligations

Require equivalent treatment of competing and affiliated products.

2. Transparency

Require disclosure of material ranking criteria.

3. Interoperability

Require technically effective access to APIs or interfaces.

4. Data separation

Restrict the use of competitively sensitive third-party information.

5. Algorithmic auditing

Independent examination of ranking and recommendation systems.

6. Structural separation

In particularly serious circumstances, separate marketplace/platform functions from competing commercial activities.

7. Choice mechanisms

Allow users or developers greater ability to select competing services.

XIII. Comparative Legal Framework

XIV. Distinguishing Innovation Management from Anticompetitive Steering

The distinction must ultimately be established through market evidence, conduct, effects and legitimate justification, rather than simply through the existence of an algorithm.

XV. Conclusion

Algorithmic innovation prioritization is becoming an important competition-law issue because algorithms can influence which technologies receive visibility, investment, platform access, interoperability, consumer attention and commercial success.

The principal competition-law risk arises when a dominant undertaking transforms algorithmic control from an efficiency mechanism into an instrument of ecosystem foreclosure or strategic self-preferencing.

The major lessons from Google Shopping, Google Android, Microsoft, Amazon Marketplace, Intel, Bronner, United Brands and Slovak Telekom are that competition analysis should examine not merely the existence of technological control but how that control affects rivals, market access, innovation incentives, interoperability and competitive conditions.

The central legal proposition can therefore be stated as:

An algorithm does not escape competition law merely because the exclusionary decision is automated. Where a dominant undertaking controls an important digital or technological ecosystem, the competitive consequences of its algorithmic prioritization may be assessed in the same way as other forms of discriminatory, exclusionary, leveraging or self-preferential conduct.