Autonomous Transport Routing Algorithm Dominance

1. Introduction

Autonomous transport routing algorithm dominance refers to a situation in which a transport, logistics, mobility, freight, or delivery platform obtains substantial market power through an algorithm that autonomously determines or materially influences routes, vehicle allocation, dispatch, traffic prioritisation, freight matching, delivery sequences, toll choices, or access to transport infrastructure.

The competition-law concern is not simply that an algorithm is sophisticated. The issue arises when control over the algorithm becomes a bottleneck for market participation, allowing the platform to disadvantage rival operators, control access to customers or infrastructure, exploit data advantages, or extend its market power into related transport markets.

Modern competition authorities increasingly examine algorithms as mechanisms through which a platform can exercise market power. For example, the CCI has specifically considered algorithmic pricing in the Ola/Uber context, while the EU's Google Shopping litigation demonstrates that algorithmic ranking can constitute an abuse where a dominant platform systematically favours its own service.

2. Meaning of Autonomous Routing Algorithm

An autonomous routing system can perform functions such as:

• selecting routes for vehicles;
• allocating vehicles to customers;
• determining delivery sequences;
• dynamically changing routes according to congestion;
• allocating scarce road or terminal capacity;
• determining which carrier receives a shipment;
• ranking logistics providers;
• prioritising particular vehicles or fleets;
• controlling access to transport networks;
• calculating estimated arrival times;
• determining toll or fuel-efficient routes;
• integrating real-time traffic information;
• optimising warehouse-to-customer movements; and
• coordinating thousands of independent transport operators.

The competition issue becomes particularly significant when one platform controls the algorithm used by a large portion of the market.

3. How Algorithmic Dominance Can Develop

A. Data advantage

A large platform can collect:

• GPS data;
• traffic data;
• customer demand;
• driver availability;
• delivery histories;
• congestion patterns;
• vehicle performance;
• route profitability; and
• competitor behaviour.

The larger the network, the greater the data available to improve the algorithm.

This can produce a feedback loop:

More users → more data → better routing → better service → more users → still more data.

Eventually, competitors may find it difficult to reproduce the incumbent's routing capability.

B. Network effects

Transport platforms often have two or more sides:

Passengers/shippers ↔ platform ↔ drivers/carriers

More customers attract more drivers, while more drivers make the platform more attractive to customers.

The CCI has recognised that network effects can contribute to platform market power, although it has also emphasised that network effects must be examined together with factors such as multi-homing, entry barriers and competitive constraints.

C. Switching costs

A logistics operator may become dependent upon a platform because:

• its fleet-management software is integrated with the platform;
• drivers are trained around the platform;
• historical route data is stored there;
• customers are acquired through the platform;
• vehicle telematics are connected to it; and
• performance ratings are platform-specific.

Consequently, switching to another routing system may be technically possible but economically difficult.

4. Relevant Competition-Law Issues

A. Abuse of dominant position

Where the platform possesses dominance, competition law may examine whether its autonomous routing system is being used to:

• exclude rivals;
• discriminate between transport operators;
• restrict access;
• impose unfair conditions;
• favour affiliated carriers;
• manipulate ranking;
• degrade rival routes;
• allocate customers selectively; or
• leverage dominance into adjacent markets.

In India, Section 4 of the Competition Act, 2002 is particularly relevant.

B. Self-preferencing

Suppose a dominant logistics platform operates:

1. the routing algorithm;
2. a marketplace connecting shippers with carriers; and
3. its own logistics subsidiary.

The algorithm could theoretically assign the most profitable or efficient routes to the platform's own fleet while giving competing carriers less attractive assignments.

This creates a self-preferencing problem.

The Google Shopping litigation provides an important technological analogy: the EU courts examined Google's preferential treatment of its own specialised search service through its ranking and display mechanisms.

The principle potentially applies to transport algorithms where ranking or routing determines competitive visibility.

5. Refusal of Algorithmic Access

An algorithm can itself become a critical infrastructure layer.

For example:

Freight marketplace → routing engine → carrier allocation → delivery network.

If competing carriers cannot realistically access the routing infrastructure, denial of access may potentially constitute exclusionary conduct where the applicable legal test is satisfied.

This resembles the essential-facilities/bottleneck reasoning developed in telecommunications cases.

6. Algorithmic Discrimination

A dominant routing platform may potentially discriminate by giving different operators:

• different route recommendations;
• different delivery priorities;
• different access to customers;
• different congestion information;
• different algorithmic scores;
• different dispatch probabilities; or
• different access to high-value routes.

The important question is whether the differential treatment has a legitimate operational justification or instead produces exclusionary competitive effects.

7. Algorithmic Coordination

Autonomous systems also raise a different issue: coordination between competing transport operators.

If competing firms use the same algorithm to determine:

• prices;
• routes;
• capacity;
• delivery schedules; or
• allocation,

the system could potentially reduce independent competitive decision-making.

However, algorithmic decision-making alone does not automatically establish a cartel.

This point is particularly important in the Indian Ola/Uber litigation.

8. Major Case Laws

1. Samir Agrawal v. Competition Commission of India & Others — India

This is the most directly relevant Indian authority.

The allegation concerned Ola and Uber's algorithmically determined fares. The argument was essentially that drivers were deprived of independent price competition because the platform's algorithm determined the applicable fare.

The CCI initially rejected the allegation of a hub-and-spoke cartel, explaining that algorithmic pricing based on large datasets did not, without more, establish collusion between drivers.

The matter ultimately reached the NCLAT, which upheld the closure of the case.

Principle

Algorithmic control ≠ automatically unlawful coordination.

For autonomous routing, this means that the existence of a common routing algorithm does not itself establish an infringement. Evidence of an agreement, concerted practice, exclusionary conduct, or abuse of dominance would ordinarily be required depending upon the legal provision invoked.

2. Meru Travel Solutions Pvt. Ltd. v. Uber India Systems Pvt. Ltd. & Others — India

The CCI examined competition in the Delhi-NCR radio-taxi market.

The Commission considered:

• market shares;
• network effects;
• multi-homing;
• entry barriers;
• competitive constraints; and
• the changing relationship between Ola and Uber.

The CCI concluded that Uber did not possess the required dominant position during the relevant period because Ola imposed substantial competitive constraints.

Principle

High usage of an algorithmic platform is not automatically equivalent to dominance.

For autonomous routing, dominance should be demonstrated through the overall competitive structure rather than simply by counting vehicles or algorithmic transactions.

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

The 2024 Grand Chamber judgment concerned Google's treatment of competing comparison-shopping services.

Google's search algorithms affected the visibility of competing services, while Google's own specialised shopping service received preferential treatment.

The Court addressed:

• dominance;
• algorithmic ranking;
• self-preferencing;
• foreclosure;
• causal connection; and
• competitive effects. 

Principle

A dominant digital platform may potentially abuse its position where the architecture or operation of its algorithm systematically disadvantages competing services.

Transport application

A dominant mobility platform could potentially face analogous scrutiny if its routing engine:

systematically gives its affiliated transport operation preferential routes, customers or visibility.

4. Federal Trade Commission v. Surescripts, LLC — United States

This is particularly useful as an analogy because the underlying market involved electronic routing.

Surescripts operated electronic prescription-routing infrastructure connecting healthcare providers and pharmacies. The FTC alleged that Surescripts used exclusivity and loyalty arrangements to prevent customers from using competing routing platforms.

The court found that Surescripts possessed monopoly power in the relevant routing markets; the subsequent settlement restricted certain exclusivity and loyalty arrangements.

The FTC's later discussion explained that Surescripts had approximately 95% shares in the relevant routing and eligibility markets and that network effects created significant barriers to competing platforms.

Principle

A platform controlling a critical electronic routing network can acquire significant market power through network effects and customer lock-in.

Transport application

A transport-routing platform could potentially create similar competitive problems if:

virtually all carriers + shippers → one routing platform → rivals cannot obtain sufficient network participation.

5. MCI Communications Corp. v. AT&T, 708 F.2d 1081 — United States

This is a foundational bottleneck case.

AT&T controlled telecommunications infrastructure required by MCI. The court considered whether refusal to provide necessary interconnection could constitute unlawful monopolisation.

The court identified traditional essential-facilities considerations including:

1. control of the facility by a monopolist;
2. inability of competitors reasonably to duplicate it;
3. denial of access; and
4. feasibility of providing access. 

Principle

Control over an infrastructure bottleneck can enable a dominant undertaking to extend monopoly power into an adjacent market.

Transport application

A comparable question could arise where a dominant autonomous routing platform controls infrastructure that competing carriers cannot reasonably replicate.

6. United States v. Terminal Railroad Association, 224 U.S. 383

Terminal Railroad involved control over strategically important railroad terminal infrastructure.

The Supreme Court treated control over an indispensable transportation bottleneck as capable of restricting competition where competitors could not effectively reach the market without access.

Principle

Transportation infrastructure can constitute a competitive bottleneck.

Modern application

An autonomous routing platform could become a digital bottleneck analogous to physical transport infrastructure where access to the algorithm is indispensable to competing effectively.

7. Otter Tail Power Co. v. United States, 410 U.S. 366

Although concerning electricity rather than transport, Otter Tail is significant for refusal-to-deal principles.

The Supreme Court examined a vertically integrated enterprise's control over infrastructure and its refusal to provide access to competing municipal systems.

Principle

A dominant undertaking may face antitrust scrutiny when control over infrastructure is used to prevent downstream competition.

Transport application

The analogy becomes stronger where a transport platform simultaneously:

• operates routing infrastructure;
• provides transport services; and
• competes against independent carriers using the infrastructure.

8. Bronner v. Mediaprint, Case C-7/97 — European Union

The Court of Justice established a restrictive approach toward imposing compulsory access under Article 102 TFEU.

The refusal to provide access becomes particularly problematic where the facility is indispensable, duplication is not realistically possible, and refusal is capable of eliminating effective competition.

Principle

Not every commercially useful algorithm constitutes an essential facility.

This is important because a transport company ordinarily has no automatic obligation to license its proprietary routing technology merely because competitors would benefit from it.

9. Autonomous Routing as a Digital Essential Facility

The concept can be visualised as follows:

                  TRANSPORT MARKET                         │                         ▼              ┌────────────────────┐              │ Autonomous Routing │              │     Algorithm      │              └─────────┬──────────┘                        │        ┌───────────────┼────────────────┐        ▼               ▼                ▼     Drivers         Shippers         Customers        │               │                │        └───────────────┼────────────────┘                        ▼                 Route Allocation                        │          ┌─────────────┼─────────────┐          ▼             ▼             ▼       Carrier A     Carrier B     Carrier C

If the central algorithm becomes indispensable, control over it can become a digital bottleneck.

10. Forms of Potentially Anti-Competitive Routing Conduct

11. Data as a Source of Routing Dominance

Autonomous routing is especially susceptible to data-driven market power.

Consider:

10 million trips → GPS data → congestion prediction → better routing → more customers → 20 million trips → even better prediction.

A rival beginning with only 100,000 trips may not be able to reproduce the same predictive accuracy.

This creates a potential data-network-effect barrier.

However, large datasets alone should not automatically establish dominance. The relevant questions include:

• Is the data unique?
• Is it difficult to reproduce?
• Is real-time access necessary?
• Can competitors obtain equivalent information?
• Can users multi-home?
• Are alternative routing technologies available?
• Is the data commercially indispensable?

12. Multi-Homing

Multi-homing is particularly important in transport.

A driver may simultaneously use:

• Platform A;
• Platform B;
• Platform C.

Similarly, a freight company may integrate several routing providers.

Multi-homing reduces the ability of a platform to become entrenched.

The CCI expressly considered the possibility of multi-homing in the Ola/Uber market and noted that the apps could coexist on the same smartphone.

Therefore:

Single-homing + network effects + high switching costs = stronger possibility of durable algorithmic dominance.

13. Algorithmic Lock-In

A platform can create lock-in through:

• proprietary APIs;
• historical route databases;
• driver reputation scores;
• vehicle telemetry;
• automated dispatch;
• integrated payment systems;
• fleet-management software; and
• contractual exclusivity.

The resulting problem is not necessarily the algorithm itself.

It is the combination:

Algorithm + data + network effects + contractual restrictions + switching costs.

That combination can create durable market power.

14. Predatory or Exclusionary Routing

Imagine a dominant platform operating its own delivery fleet.

It could theoretically configure the algorithm so that:

• its own vehicles receive profitable routes;
• rivals receive low-margin routes;
• independent carriers experience longer waiting times;
• affiliated vehicles receive priority during congestion.

The legal inquiry would focus on evidence of exclusionary purpose/effect and whether there are legitimate operational explanations.

An unexplained algorithmic disparity could therefore become important evidence, but the disparity itself would not automatically establish an infringement.

15. Autonomous Routing and Essential Facilities

A useful analytical test is:

Question 1

Does the platform possess dominance?

Question 2

Is the routing system genuinely indispensable?

Question 3

Can competitors reasonably reproduce it?

Question 4

Does the platform refuse or restrict access?

Question 5

Is access technically and economically feasible?

Question 6

Does the restriction eliminate or substantially weaken effective competition?

MCI Communications illustrates the traditional bottleneck framework, while Bronner demonstrates the more restrictive European approach to compulsory access.

16. Autonomous Routing and Section 3 of the Indian Competition Act

Section 3 becomes relevant where autonomous systems facilitate agreements or concerted practices between competitors.

Potential scenarios include:

• common routing algorithms used by competing carriers;
• exchange of competitively sensitive information;
• coordinated capacity allocation;
• algorithmically coordinated prices;
• common software providers transmitting sensitive data;
• hub-and-spoke arrangements.

But Samir Agrawal demonstrates an important limitation:

An algorithm setting prices or operational parameters does not by itself prove an agreement among competing operators.

Evidence of coordination remains important.

17. Section 4 and Autonomous Routing

Where dominance exists, Section 4 could potentially become relevant to:

Section 4(2)(a)

Unfair or discriminatory conditions.

Section 4(2)(b)

Limiting or restricting:

• goods/services;
• technical development; or
• market access.

Section 4(2)(c)

Denial of market access.

Section 4(2)(d)

Tying/leveraging.

Section 4(2)(e)

Using dominance in one relevant market to enter or protect another market.

These provisions provide a useful framework for analysing a dominant autonomous-routing platform.

18. Algorithmic Transparency

Competition authorities may increasingly examine:

• algorithmic audit trails;
• route-allocation criteria;
• API access;
• ranking mechanisms;
• data inputs;
• discrimination between operators;
• changes in algorithmic parameters;
• model training data;
• historical allocation patterns.

A platform should therefore be capable of explaining why a particular carrier received a particular route.

19. Remedies

Potential competition-law remedies could include:

Structural remedies

• separation of routing and transport operations;
• divestiture of a logistics subsidiary;
• separation of marketplace and carrier businesses.

Behavioural remedies

• non-discriminatory access;
• interoperability;
• API access;
• prohibition of exclusivity;
• transparent ranking criteria;
• independent algorithmic audits.

Data remedies

• data portability;
• interoperability;
• access to essential datasets;
• restrictions on combining datasets.

Contractual remedies

• prohibition of loyalty clauses;
• limits on exclusivity;
• restrictions on tying;
• multi-homing protections.

The Surescripts settlement illustrates how restrictions on loyalty and exclusivity can be used to protect multi-homing and competitive entry in a network platform.

20. Key Legal Distinction

The most important distinction is:

21. Emerging Legal Theory: The Digital Bottleneck

Autonomous transport routing creates a possible evolution of the traditional essential-facilities doctrine.

Traditional bottleneck

Railway track → terminal → telecommunications network

Digital bottleneck

Data → algorithm → routing infrastructure → customer allocation

The critical asset is no longer necessarily physical.

It can be:

a computational system controlling access to the market.

This makes autonomous transport-routing platforms particularly important to future competition-law enforcement.

22. Conclusion

Autonomous transport-routing algorithm dominance is fundamentally a question of control over market access through computational infrastructure.

The principal competition concerns are:

1. algorithmic self-preferencing;
2. data-driven barriers to entry;
3. network effects;
4. multi-homing restrictions;
5. algorithmic discrimination;
6. refusal of algorithmic access;
7. exclusive routing arrangements;
8. leveraging into adjacent transport markets;
9. algorithm-facilitated coordination; and
10. digital bottleneck/essential-facility problems.

The Indian Samir Agrawal and Meru decisions are particularly useful for understanding algorithmic platforms and dominance, while Google Shopping, Surescripts, MCI, Terminal Railroad, Otter Tail, and Bronner provide broader principles concerning algorithmic foreclosure, routing networks, bottlenecks, refusal to deal and access to essential infrastructure.