THE ILLEGALITY OF DATA LOCK-IN: AN ANALYSES OF WHETHER EUROPEAN COMPETITION LAW OBLIGES UNDERTAKINGS WHO ARE DOMINANT IN SOCIAL MEDIA TO ENGAGE IN DATA PORTABILITY

(1)

University of Amsterdam Faculty of Law

THE ILLEGALITY OF DATA LOCK-IN: AN ANALYSES OF WHETHER EUROPEAN COMPETITION LAW OBLIGES

UNDERTAKINGS WHO ARE DOMINANT IN SOCIAL MEDIA TO ENGAGE IN DATA PORTABILITY

Jazz Ellis 11007915

Thesis submitted as partial requirement for the degree of LL.M European Competition Law and Regulation

June 2016 (formally submitted October 2016)

Interdisciplinary Research Colloquium conducted under the supervision of Mr. Joris Ruigwaard LL.M

(2)

ii I, Jazz Jethro Ellis, confirm that all works presented in this thesis are my own. Where the origin of information has been procured from other sources, I vouch that this has been indicated within the thesis.

(3)

iii

Abstract

The era of big data has arrived and with-it the lucrative nature of its collection and utilization through the harnessing of (direct and indirect) network effects. Data aggregation is at the forefront of successful online platforms- influencing efficiency, productivity and prosperity. However, such invaluable qualities breed possessiveness, the product of which is a raised risk of abuse through its retention where consumers are locked-in to unfair

purchasing prices and trading conditions. The most appropriate answer given to this dilemma has been endowed in the principle of portability. Portability under competition law concerns the obligation for dominant entities to support and aid the automatic transfer of data in a commonly used format between platforms, thereby mitigating switching-costs and consumer lock-in.

The methodology adopted in this dissertation relates to an analytical and empirical-based assessment in the attempt to answer the numerous theoretical as well as legal questions being contemporarily debated surrounding competition law and portability. The first chapter

addresses the core concept of ‘data’, and outlines the contextual importance of data relative to social media as well as its nature in terms of the two-sided market upon which it operates. The second chapter is devoted to the notions of ‘switching costs’, ‘lock-in’ and ‘portability’, in which these pivotal concepts will be clarified to the extent necessary for the final chapter to provide a coherent examination thereupon. The final chapter encompasses a concise analysis of competition law in relation to big data and portability, in an attempt to provide a qualitative answer to the proposed research question, and an insight into the possible

direction of the soon-to-be established paradigm. Data protection legislation is considered to the extent that it is contextually necessary, and where its normative framework may provide a basis from which competition law may draw semantic as well as circumstantial conclusions.

This paper contributes to existing scholarship in 3 ways. First, through the characterization of data in support of its rivalrous nature in an international setting. Second, it asserts the need for normative frameworks forming the basis of those factors which constitute competition parameters (price-quality substitution accounting for ‘zero-price’ markets). Third, it looks at the relatively neglected notion of an exploitative theory of harm under Article 102 and its relation to data lock-in and portability. KEYWORDS

(4)

iv

Data is often considered a catalyst for economic growth, sparking innovation as well as digitization pertaining to all economic sectors, providing a crucial foundation for SMEs, start-ups and consequentially society as a whole. Faced with increasing scrutiny by antitrust laws, the last decade has seen social networking entities such as Facebook, LinkedIn and Twitter -in the context of Big Data-, as constantly fall-ing short of the European competition

authorities’ watchful eyes, being subsequently assessed conventionally under data protection laws. This reality and possible remnant resembling a past-time may now be about to embark upon a journey of impending change as a result of the uproar of data’s collection,

systematization and utilization. In the wake of contemporary progress in antitrust and Big Data, questions pertaining to the very foundation of what characterizes a relevant market, dominance thereon, abuse thereof, and the extent to which market ‘inputs’ in this context are determinable in their own relevance,- are in search of answers.

The nature of the market for Big Data is two-sided, and as a result; problematic in terms of applying competition law to an ‘asset’ collected ‘free of charge’ on the first side and monetized on the second. Although the schools of thought regarding how best to deal with Big Data and social media on the European market can be described as segregated between regulatory intervention (ex-ante), and antitrust enforcement (ex-post), the line of separation is slowly becoming blurred.

Big Data and the concept of portability has thus started to gain momentum in the realm of competition law, where it was previously seen as reserving itself to regulatory oversight. As a result, complications have arisen due to Competition law’s infant state with regard to Big Data. In order to deal with an abuse of a dominant position on a market which cannot be

1_{Maurice E. Stucke & Allen P. Grunes, ‘No Mistake About It: The Important Role of Antitrust in the Era of Big Data’}

(2015) University of Tennessee Legal Studies Research Paper Series #269, 3.

(8)

viii

determined by conventional methods, and the theory of harm which has not been empirically proven; competition policy and the methodology in its application must adapt fast if its scope is to be extended to lock-in and portability.

The concepts of ‘lock-in’ and ‘portability’ can be regarded as some-what antonyms. Lock-in concerns the inability of consumers to switch between competing providers of a service. This is because it entails manually duplicating the inputs (information) which had been invested into the first ‘seller’; a ‘costs’ which exceeds the benefits of migrating to a competing

platform. Portability circumvents such lock-in by obliging dominant undertakings to transfer (automatically) to the competing entity, that data pursuant to which the consumer is locked-in. The concept of portability, although not entirely new in terms of data, is a new

development in terms of competition law. Although the only contemporary normative legal framework pertaining to portability regards that of the soon-to-be implemented General Data Protection Regulation (GDPR), the question is open as to whether portability is

implementable as a remedy outside of data protection, and if-so, how this would be achieved in a theoretical as well as practical sense.

Therefore, this dissertation aims at contributing to the current debate and attempts to further the discussion of how best to adapt a system of legal standards, in a regime renowned for its slow evolution.*3

* The following terms will be referred to interchangeably throughout this dissertation; undertaking, company, entity, business, enterprise, start-up, corporation, platform

While multiple undertakings in the social media sector will be made reference-to, the online platform known as ‘Facebook’ will be regarded as the overarching paradigm.

(9)

1

Chapter 1: Big Data

1.1 Defining Big Data

Having been referred to as (among other things) the 'new oil' of the 21st century,4

there's no surprise that reluctance ensues at the thought of sharing something that has been compared to a multi-trillion euro industry. The question however, pertains to the appeal that consumer data has, in that it would compel an undertaking to largely forego demand for remuneration and accept something that until only recently has been referred to as an ‘asset’.5

The definition of Big Data as a concept can be seen to change depending on the purpose and subjective understanding held by the individual attempting to define it. A constant however, is evident wherein ‘Big Data’ is frequently characterized by the “four Vs”; volume, variety, velocity (speed of transfer), and value.6 These ‘classes’ delineate an abundance of

information collected by an array of mechanisms; social media platforms being of primary importance for the purpose of this paper, pertaining to capabilities of continuous retrieval of ‘large-scale quantities of data’, including: video, audio, image and text.7_{In this respect, the}

most concise definition of Big Data characterizes it as “large volumes of high velocity, complex and variable data”, which require “advanced techniques and technologies” in accomplishing its “capture, storage, distribution, management” and “analysis”.8

Big data has also been characterized as encompassing much more than just aggregated data, but involves the transcription of “all aspects of life and turning them into data”.9_{As such, and}

as a result of the large amounts of different types of data being produced at high speed from multiple sources, the handling of said data requires ever-updated and ever-powerful

4_{Meglena Kuneva, ‘Keynote Speech’ (Roundtable on Online Data Collection, Targeting and Profiling, Brussels, 31 March}

2009) <http://europa.eu/rapid/press-release_SPEECH-09-156_en.htm> accessed 08 June 2016.

5_{‘Big Data and Competition – merger control is not the remedy for data protection issues’ (TaylorWessing, July 2014)}

<http://united-kingdom.taylorwessing.com/globaldatahub/article_big_data_competition.html> accessed 08/06/2016.

6_{Ibrahim Abaker Targio Hashem, Ibrar Yaqoob, Nor Badrus Anuar, Salimah Mokhtar, Abdullah Gani, Samee Ullah Khan,}

‘The rise of “big data” on cloud computing: review and open research issues’ (2015) Information Systems 98, 100-102.

7_{Maryam M Najafabadi, Flavio Villanustre, Taghi M Khoshgoftaar, Naeem Seliya,}

Randall Wald, Edin Muharemagic, ‘Deep learning applications and challenged in big data analytics’ (2015) Journal of Big Data, 1, 8.

8_{TechAmerica Foundation’s Federal Bid Data Commission, Demystifying Big Data: A practical Guide to Transforming the} Business of Government, 10.

(10)

2

processing mechanisms. ‘Big Data’ in this sense, is a term illustrating the ever increasing volume of a category of data which is not only difficult to process and analyze with respect to conventional data-base technologies, but also to comprehend; primarily due to its raw state. Considerable resources are required in converting a mass of individualized, uncategorized and un-systematized data into usable information. The final asset of inferred data and its rich insight into customer opinion and behavioral statistics10 is the ‘gold bar’ entities have forged out of individual flakes; an effort arguably not easily replicated by everyone. The collection, storage and analysis of consumer data allows for the identification of trends which would otherwise not be visible when examined through the lens of individual data collections or smaller data-aggregates.

1.2 Data as a Currency for Online Services

As a result of entering the era of Big Data and behavioral targeting,11 It can be said that ‘but for Big Data collection operations, the internet would be much more expensive’. Conventional businesses operate on the basis of selling products and services; however, their products will not appeal to everyone and as such they appeal only to a certain demographic. The ‘hit-and-miss’ approach of advertising to the ‘largest candidate group’ is a thing of the past, replaced by the insight of aggregate data accurately informing businesses as to the types of people to which an offer will be a ‘hit’, and those to whom it will result in a ‘miss’.12_If

Big Data collection operations were inexistent, revenue-creating services in industries such as advertising would be less personalized due to the inexistence of targeted advertisements, and as such- less profitable. It can be said that the personal data (the information “relating to an identified or identifiable natural person”)13_{collected by undertakings –who are (also)}

represented online- is the price paid by consumers for the use of the services offered.14

10_{Hsinchun Chen, Roger H. L. Chiang, Veda C.Storey, ‘Business Intelligence and Analytics: From Big Data to Big Impact’}

(2012) Business Intelligence Research, 1169 <http://hmchen.shidler.hawaii.edu/Chen_big_data_MISQ_2012.pdf> accessed 08 June 2016.

11_{Charles W. Lamb, Joe F. Hair, Carl McDaniel, MKTG 8 (8}th_{edn, Cengage Learning 2014) 167; BT concerns a “form of}

observation marketing research that combines a consumer’s online activity with psychographic and demographic profiles compiled in databases”.

12_{Bernard Marr, Big Data in Practice: How 45 Successful Companies Used Big Data Analytics to Deliver Extraordinary} Results (1st_{edn, Wiley 2016) (Hereon ‘Big Data in Practice’) 70-71.}

13_{Council Directive 96/45/EC of 24 October 1995 on the protection of individuals with regard to the processing of personal}

data and on the free movement of such data [1995] OJ L281 (Hereon ‘Directive’) Art. 2(a).

14_{Margrethe Vestager, ‘Keynote Address’ (Competition in a big data world, Munich, 17 January 2016) (Hereon 'Vestager')}

<https://ec.europa.eu/commission/2014-2019/vestager/announcements/competition-big-data-world_en> accessed 08 June 2016.

(11)

3

Consequently, without the income generated by this medium, said companies would have a reduced incentive to provide such services.

Evidently, consumer information is an invaluable asset for any business, particularly those of a commercial nature. Entities must (to an extent) know what a consumer wants in order for it to be offered (targeted) to them; a reality emphasized in the realm of online business, the success of which directly correlates to the acquisition and monetization of personal data. The comprehension and collection of behavioral nuances in the global consumer market is not only difficult,15 but logically presupposes an excessive dependence-upon, and consumption-of time. Consequently; access to Big Data is guarded by individual companies as if it were in itself a tangible asset.

1.3 The Position of Social Media Platforms in Relation to Data

Popular online platforms logically and empirically breed monopolistic tendencies, where direct and indirect network effects lead to increased efficiency, functionality and convenience for consumers in utilizing their services. In defining ‘social media platforms’, although an explicit definition is yet to be attributed thereto; an online platform is currently regarded as an “undertaking operating in two (or multi)-sided markets, which uses the internet to enable interactions between two or more distinct but interdependent groups of users so as to generate value for at least one of the groups”.16_{The problem with this}

definition is that its broad nature includes entities which one may argue are not similar, in which social media platforms may be classified within the same category as Newspapers.

Business models pertaining to the utilization of Big Data concern a four-step process which consists of collection and access, storage and aggregation, analysis and distribution, and the utilization of personal data sets.17 As such, those entities which have the relevant

infrastructures in place to undergo this process, already occupy a favorable position. The data acquired by social media platforms in return for services offered forms the foundation of

15_{Sandra J. Keiser, Myrna B. Garner, Beyond Design: The Synergy of Apparel Product Development (3}rd_{edn, Fairchild}

Books 2012) (Hereon ‘Beyond Design’) 87.

16_{European Commission, open consultation on ‘Regulatory environment for platforms, online intermediaries, data and cloud}

computing and the collaborative economy’ (Hereon ‘Consultation’) 5.

17_{OECS (2013), “Exploring the Economics of Personal Data: A Survey of Methodologies for Measuring Monetary Value”,}

OECD Digital Economy Papers, No. 220, OECD Publishing, 10 <http://dx.doi.org/10.1787/5k486qtxldmq-en> accessed 08 June 2016.

(12)

4

aggregated, generalizable, and prediction-worthy consumer data.18 In terms of scale, Facebook users are said to upload an estimated 2.5 million pieces of content per minute, which are subsequently dissected, analyzed and interpreted to provide clues for the benefit of advertisers.19

As platforms do not compete on a price-basis, they compete alternatively on other variables such as their user-base. The underlying elements thereof largely relate to the attention as well as loyalty of individuals who decide which platform to be active-on on the basis of various factors.20 The nature and position of social media platforms such as Facebook allow for the rapid collection of personal information, without the subject feeling as if they are being pressured or forced into surrendering it. Virtual profiles of individuals are compiled on the basis of categories such as gender, age, location, interests, education, relationship status and purchasing behavior.21 Inferences made therefrom allows for meaningful insights into individual preferences, the generalization of which is proven to provide a fruitful statistic to online platforms as well as advertising agencies in identifying those consumers most likely to engage with certain advertisements.

As such, Big Data in the right hands has the empirical consequence of providing a significant advantage to companies ‘active in (…) online advertising, online search, social networking services and software products’,22_{while the position held by social media platforms in}

relation to Big Data is perfect for its ‘covert’ collection.

1.4 Data-Collection Capabilities

The spectrum of data acquisition (processing)23 methods largely comprise three notable categories; ‘Volunteered’, ‘Observed’ and ‘Inferred’24 practices.

18_{Preliminary Opinion of the European Data Protection Supervisor, ‘Privacy and competitiveness in the age of big data: The}

interplay between data protection, competition law and consumer protection in the Digital Economy’ (European Data Protection Supervisor) March 2014, 10.

19_{Big Data in Practice Supra note 12, 72.} 20_{Infra note 46 and accompanying text.}

21_{‘Boost posts to reach more people’ <https://www.facebook.com/business/learn/facebook-page-boost-posts> accessed 08}

June 2016; see also ‘How to target Facebook adverts’ <https://www.facebook.com/business/a/online-sales/ad-targeting-details> (Hereon ‘Advertising Tool’) accessed 08 June 2016.

22_{Eleonora Ocello, Cristina Sjödin, Anatoly Subočs, ‘Competition merger brief: What’s up with Merger Control in the}

Digital Sector? Lesson from the Facebook/Whatsapp EU merger case’ 2015/I (Hereon ‘Merger Brief’) 6.

23_{Direct Supra note 13, art. 2(b).}

24_{Meeting the Challenges of big data: A call for transparency, user control, data protection by design and accountability,}

(13)

5

Volunteered data typically concerns information over which the consumer exercises

individual control, as well as to which one may attach a ‘strong sense of ownership’. Data in this sense largely relates to photos, emails, online posts and so on. As the category suggests, this information is acquired on a voluntary-basis (upon acceptance of privacy policies, which are rarely read before being agreed to).25 The fact that the data of this category is voluntarily provided does not mean that this voluntary notion extends to the purpose for which such data might be utilized,26_{an assertion acknowledged in section 2.4.}

Observed data concerns that information which belongs to the undertaking, who by virtue of their own methods and largely unbeknownst to consumers has acquired it through the passive observation of consumer activity. Observed data may be acquired through location tracking, recording online activities,27 as well as documenting consumer reactions to online content and the overall effect that they have on consumers as individuals, as was the case in Facebook’s clandestine study conducted in 2013.28

Inferred data differs from the former two categories in the sense that it relates to a statistical aggregate derived from the two. ‘Inferred’ data can be said to be the most important for companies and the purpose of this paper, as it opens the door to ‘predictive capabilities’,29

and can largely be referred to as ‘metadata’- concerning data which ‘describes’ data. For this reason, inferred data can be considered a proprietary asset,30 the importance of which is indisputable.

1.4.1 Cookies

Cookies are regarded as the main method of data collection outside of its

(conventional) voluntary surrender. Cookies relate to an automated system of data collection to which a vast majority of the population is oblivious,31_{and regard the placing of small text}

‘Rethinking Personal Data: Strengthening Trust’, World Economic Forum (May 2012) (Hereon ‘Strengthening Trust’) 19, <http://www3.weforum.org/docs/WEF_IT_RethinkingPersonalData_Report_2012.pdf> accessed 08 June 2016.

25_{Aleecia M. McDonald, Lorrie Faith Cranor, ‘The Cost of Reading Privacy Policies’ (2008) Journal of Law and Policy for}

the Information Society, 6 <http://lorrie.cranor.org/pubs/readingPolicyCost-authorDraft.pdf> accessed 08 June 2016.

26_{Directive Supra note 13, art. 6(1)(b).} 27_{Strengthening Trust Supra note 24, 10.}

28_{Adam D. I. Kramer, Jamie E. Guillory, Jeffrey T. Hancock, ‘Experimental Evidence of mass-scale emotional contagion}

through social networks’, 3 <http://www.pnas.org/content/111/24/8788.full.pdf> accessed 08 June 2016.

29_{Challenges of big data Supra note 24, 7.} 30_{Strengthening Trust Supra note 13, 19.}

(14)

6

files on the personal computers of consumers by the websites that individuals have visited.32 These text files collect information on consumer preferences, including search history, articles read and anything else which can be used in building an independent consumer profile upon which marketing inferences are made.33

Cookies are largely categorized between ‘session cookies’ and ‘persistent cookies’, which in-turn can constitute either ‘third party cookies’ or not.34

1.4.1.1 Session cookies

‘Session cookies’- gather information from the time a webpage is opened until it is closed, to which no trace is left once the webpage is exited. Session cookies are referred-to in a generic sense as ‘user input cookies’ and are conventionally used to track user inputs when filling-in online information such as surveys, e-commerce inputs, and the items a user has shown interest-in by clicking thereon. Session cookies can also be used for security reasons such as online banking or webmail facilitation.35

1.4.1.2 Persistent cookies

‘Persistent cookies’ which although have the possibility of being programed to expire,36 are stored on the consumer’s hard drive between browser sessions, and -unlike session cookies which expire after the browser session-, can persist in collecting data for months and even years before expiring.37 Persistent cookies allow for the retention of a user’s information and preferences between sessions and can also be used to accommodate for targeted advertising.38 Due to their longer life-span, persistent cookies are considered more privacy-intrusive than session cookies.

32_{Beyond Design Supra note 15, 86.}

33_{Article 29 Data Protection Working Party, ‘Opinion 2/2010 on online behavioural advertising’ (22 June 2010) WP 171}

(Hereon ‘Behavioural Advertising’), 7 <http://ec.europa.eu/justice/policies/privacy/docs/wpdocs/2010/wp171_en.pdf> accessed 08 June 2016.

34_{Article 29 Data Protection Working Party, ‘Opinion 04/2012 on Cookie Consent Exemption’ (7 June 2012) WP 194}

(Hereon ‘Exemption’), 4 <http://ec.europa.eu/justice/data-protection/article-29/documentation/opinion-recommendation/files/2012/wp194_en.pdf> accessed 08 June 2016.

35_{Privacy and Electronic Communications Regulations, ‘Guidance on the rules on use of cookies and similar technologies’,}

(Hereon ‘Cookies’) 4-5 <https://ico.org.uk/media/for-organisations/documents/1545/cookies_guidance.pdf> accessed 06 June 2016.

36_{Exemption Supra note 34, 4.}

37_{Antony D. Miyazaki, ‘Online Pivacy and the Discolsure of Cookie Use: Effects on Consumer Trust and Anticipated}

Patronage’(2008) Vol. 27/1 Journal of Public Policy & Marketing, 20-21 <http://journals.ama.org/doi/abs/10.1509/jppm.27.1.19>

(15)

7

1.4.1.3 Third party cookies

Where a cookie is not of a third party nature, it will be that of a ‘first party’ one and vis-versa. The concepts of ‘first’ and ‘third’ refer to the websites or domains which have placed the cookie. While first party cookies pertain to those set by the websites which

individuals have visited and where the information recorded is acquired by the same website, third party cookies are those set by domains other than the website on which an individual has been active. As such, third party cookies are those administered by one website in order to retrieve data about an individual’s activities on another website.39

1.5 The Economic Characteristics of Data in 2-Sided Markets

In order to understand the utilizable nature of consumer data in the context of social media platforms, it is important to acknowledge the two-sided character of the market within which Facebook and entities like it operate.

Online platforms adopt business models within which an inherent reliance is placed on personal data as a key input. Such data-driven models undeniably illustrate the efficiency of two-sided market regimes. The services offered in return for personal data between social media platforms and individuals on one side, directly impacts the utilization of data offered in return for monetary satisfaction between the same social media platforms and advertisement entities on the other side.40 The extensiveness of platform traffic in conjunction with the quality and quantity of data acquired is directly correlated to the value attributed to the platform by advertisement entities, and their desire to utilize the platform’s consumer data. See image 1.1.

The interrelation between the volume of consumers active on the first side of the platform is not the only variable to which an increased amount of advertisers on the second side has a correlation. Account must also be taken of the vertical and horizontal product/service differentiation in relation to price competition on the first side41 as compared to the price of

39_{Exemption Supra note 34, 4-5.}

40_{Maurice E. Stucke & Allen P. Grunes, ‘Debunking the Myths Over Big Data and Antitrust’ (2015) University of}

Tennessee Legal Studies Research Paper Series #276, 2. Debunking the 10 myths (saved) p.2

41_{Lapo Filistrucchi, Damien Geradin, Eric van Damme, Pauline Affeldt, ‘Market Definition in Two-Sided Markets: Theory}

(16)

8

the analogous product/service offered by competing platforms. As such, where a platform has more individual consumers on the first side of the market in comparison to competing

platforms, it can also be expected that it will have a relatively larger volume of advertisers on the second side when contrasted with the same competing platforms. As a determinate monetary value is yet to be attributed to data, it is difficult to establish boundaries of price competition. Instead, price-based competition can (in this instance) be substituted for quality-based competition (see section 3.2).

Data is evidently a key input in such a 2-sided market, and is the reason pursuant to which social media platforms offer the services that they do. Data is also –to a large extent- the overarching factor which advertisers will take into consideration in deciding upon which platforms to advertise (whether directly or indirectly).

1.1- Two-sided market effects, where an increase in consumers and subsequently data leads to an increase in advertisers

Difficulty arises in attributing a value to data where its value is inconsistent between both sides of the market. An entity must take into account the elasticity of demand of both its individual consumers to ‘purchase’ the service, as well as advertisers to place advertisements on the platform. Cross-side network effects are also be taken into consideration regarding the correlation between both sides of the market, as well as the extent to which prices on one side of the market influence the other side. For example, where the ‘price’ of a platforms’

‘services’ was to become cheaper- leading to a surge in demand on the consumer side, this would subsequently result in increased demand on the advertisement side.

(17)

9

An indication of the (direct or indirect) importance of data in the advertising market can be seen following Facebook’s IPO in 2012; after which in 2013, Facebook’s market share in the US mobile advertising market amounted to 18.6% valued at $3.3 billion. This can be

contrasted with a market share of over 25% valued at $6.8 billion in 2015, while Facebook’s market share is forecasted to reach just under 27% ($10 billion) in 2017,42_{primarily a product}

of advertising, the revenue from which forms the majority of Facebook’s overall revenue.43

1.6 Sub Conclusion

Characterized as large volumes of highly complex data requiring advanced technologies in its aggregation and utilization, Big Data accounts for all aspects of life produced in a continuous fashion into a statistical format. It fuels the provision of ‘free’ online services and -by virtue of a 2-sided market structure-, is monetized through targeted advertisement on account of its inferable nature based on behavioural assumptions. As such, it is protected. Social media platforms are uniquely positioned for its collection, and

consequently compete with each other in its acquisition. Its acquisition takes place on a voluntary and observed basis, pursuant to which ‘cookies’ play a predominant role.

42_{‘Trends in Global Advertising Industry: Winners and Losers -Part 2-’ (Forbes, 5 October 2015)}

<http://www.forbes.com/sites/greatspeculations/2015/10/05/trends-in-global-advertising-industry-winners-and-losers-part-2/#6dfad167638d> accessed 08 June 2016.

43_{Facebook Q4 2015 Results, 8}

(18)

10

Chapter 2: Portability

2.1 Switching Costs

‘Switching costs’ refer to the ‘price’ that consumers are compelled to pay when switching between platforms. A switching cost is brought about when an investment which is made by a consumer in their current service provider must be duplicated for the new service provider.44 This is an effect to the detriment of consumers where the service is associated with network effects, and is more harmful where consumers engage in ‘single-homing’ (using one platform) as opposed to ‘multi-homing’ (using multiple platforms).45 A key remedy to that of switching costs concerns the extent to which data is transferable (portable) between platforms. Consequently, switching costs and data portability are negatively correlated.

Switching costs lead to consumers being ‘locked-in’ -in an exploitative sense- to a particular standard, an externality which becomes graver the longer a consumer is active on a given platform. This ultimately disincentives consumers to switch to competing platforms, even where competing platforms offer consumers greater benefits in terms of reliability, functionality, modesty, entertainment and privacy.46 As such and in light of competing alternatives, the level of switching costs should be taken into consideration where “high switching costs may (…) facilitate foreclosure, as customers are less able to react to

foreclosure strategies (such as price increases, quality degradation or bundling)”.47_However,

a determinative extent to which switching costs constitute lock-in is yet to be identified and as such the degree required for one to lead to the other is largely subject to speculation.

2.2 Lock-in

Lock-in is the main consequence of an absence of data portability. Depending upon one’s characterization of what ‘social media’ encompasses, custom pertaining to portability

44_{The ‘cost’ concerns the excess efforts required in constructing a profile of similarity on the competing platform and the}

re-entering of information which one will have supplied the initial platform over the course of its use.

45_{Autorité de la concurrence, Bundeskartellamt, ‘Competition Law and Data’ (10 May 2016) (Hereon ‘Cooperative’) 28}

<http://www.autoritedelaconcurrence.fr/doc/reportcompetitionlawanddatafinal.pdf> accessed 09 June 2016; “Consumers are said to multi-home when they use several providers to get the same kind of service”.

46_{Spencer Weber Waller, ‘Antitrust and Social Networking’ (2011) North Carolina Law Review (Hereon ‘Antitrust’) 20; the}

factors relate to; servers rarely crashing, the platform accessibility in navigating, excessiveness of advertisement, recreational enjoyment, data privacy, and protection considerations.

47_{COMP/M.4731 – Google/ DoubleClick C[2008] 927 final [2008], 297, 301; see also; Eastman Kodak Co. v Image}

(19)

11

largely concerns two categories; ‘portable’, and ‘non-portable’. The latter is of most

relevance in this context and is the area to which most controversy is overtly evident. It has been acknowledged in a reoccurring fashion that “Customers should not be locked-in to a particular company just because they once trusted them with their content”.48_Although

platforms such as Facebook do allow for users to obtain a copy of their data, there is no option to differentiate between the copied data, and as a result individuals are forced to download either their data in its entirety, or forego obtaining a copy. The format of the data which is downloaded is also problematic; as the issue may arise in which data-sets are not fully compatible to allow importation into other social networks.49 Third parties are also unable to mitigate this concern as Facebook does not permit third-parties to acquire consumer information directly.50 Consequentially, an individual is left with the choice of cumbersomely re-entering their data into an alternative social network (which is unlikely), or staying with the original platform. In conjunction with brand loyalty, poor substitutes and the fact that a consumer’s family and friends who are active on the same platform are also unwilling to switch (a result of network effects); individuals are essentially ‘locked-in’ to the initial platform and deterred from utilizing competing platforms. It is also peculiar that numerous instances have materialized in which entities such as Facebook have changed their privacy policies -provoking outrage-, yet the growth of such entities has gone largely unscathed,51 alluding to the extent that consumers are locked-in.

The concept of lock-in -the rationale of which was examined in the Facebook/Whatsapp merger- has the possibility of constituting an abuse. Here the premise was employed that an increase in the time and costs needed to switch between products or services correlated to a heightened sense of lock-in,52_{which consequentially overlaps exclusionary and exploitative}

theories of harm, a reality which is empirically evident in the case of Facebook, and an abuse to which portability may constitute a viable remedy.

For entities such as Facebook whose business models are heavily reliant on cross network effects between both sides of the market-, generating consumer lock-in as a result of

48_{Joaquín Almunia, (Privacy Platform event: Competition and Privacy in Markets of Data, Brussels, 26 November 2012)}

(Hereon ‘Almunia’) <http://europa.eu/rapid/press-release_SPEECH-12-860_en.htm> accessed 09 June 2016.

49_{Inge Graef, ‘Mandating portability and interoperability in online social networks: Regulatory and competition law issues}

in the European Union’ (2015) Vol.39 Telecommunications Policy 502 (Hereon ‘Mandating’) 509.

50_{‘Statement of Rights and Responsibilities’ <https://www.facebook.com/legal/terms> accessed 09 June 2016.} 51_{Antitrust Supra note 46, 20.}

(20)

12

switching costs is vital to such a model’s success. While the locking-in of consumers is not in itself a prohibited business model and is in effect the strategy adopted or aspired-to by the majority of contemporary businesses,53 where an undertaking is dominant; various

exclusionary as well as exploitative theories of harm are asserted to prospectively occur on both sides of the market. Not only are consumers exploited in the sense that they are forced to use a service of less quality and convenience, but as an indirect result, competing platforms are excluded from accommodating the needs of the same consumers to the quality and convenience that they would expect. Consequentially consumers are exploited where the price they pay for a service is larger than that which they would pay for the same service from a competing entity.

High switching costs therefore accommodate for entities to gain immediate profits where the reality of locked-in customers limits or eliminates long-term loss.54 As such, any artificial increase in switching costs (and consequently ‘lock-in’) may therefore constitute a

deterioration of competition, and result in an abuse. Although numerous remedies may provide the possibility of mitigating or dispelling the effects of lock-in, non-such are so fitting as that of portability.

2.3 Portability

In his capacity as European Competition Commissioner, Joaquine Almunia referred to data portability as going “to the heart of competition policy”.55_{Data portability and the rights}

attached thereto largely constitute two facets; that of the right to obtaining a copy of the data which an individual has provided the social media entity, and the right for said data to be transferred directly between entities (where technically feasible) without further input from the individual consumer.56 Data portability mitigates barriers to entry and their association with switching costs by allowing for consumers to transfer their data between service providers; combating lock-in where individuals would want to migrate but are unable to do-so on account of the volume and value of data that would be ‘left behind’.

53_{David G. Yosifon, ‘Consumer Lock-in and the Theory of the Firm’ (2012) 35 Seattle U.L. REV. 1429, 1460.} 54_{Kodak Supra note 47, 477-478.}

55_{Almunia Supra note 48.}

56_{Inge Graef, Jeroen Verschakelen,Peggy Valcke, ‘Putting the Right to Data Portability into a Competition Law Perspective’}

(21)

13

While mechanisms are already in place with regard to the right of copy, more emphasis as well as importance is devoted to the right of transfer. It is important to acknowledge that the current Data Protection Directive (95/46/EC) will be replaced by the General Data protection Regulation (GDPR) in May 2018, within which the concept of ‘portability’ has been

explicitly introduced. The concept of data portability can be seen illustrated in Article 20 of the draft proposal of the GDPR and regards the following;57

“The data subject shall have the right to receive the personal data concerning him or her, which he or she has provided to a controller, in a structured and commonly used and machine-readable format and have the right to transmit those data to another controller without hindrance from the controller to which the personal data have been provided (…)”

Portability essentially obligates service providers to help consumers in acquiring and (automatically) transferring data to new providers where conditions are sufficed. However, various complications arise, largely in terms of the classification of what data should be made portable (see section 2.3.2); a question which warrants consideration of the two most significant data sets,58 and the pertinence of their characterization.

Competition Commissioner Margrethe Vestager asserted in relation to the application of competition policy that we are “not talking about individuals’ personal data as such, but the huge collections of information that companies can use to understand their environment in a way they never could before”.59_{As commissioner Vestager implicitly makes reference to}

inferred data and excludes “individuals’ personal data”, a basis is formed for the determination of whether individual data (that to which the lack of portability has the resulting impact of consumer lock-in), is to be considered within the scope of competition law, or whether the commissioner solely refers to the invocation of competition law in relation to aggregated data. This is an important delineation as the intentional exclusion of un-aggregated data from the definition of portability also excludes individual consumers and an exploitative theory of harm. One may assert the contrary however, in that ‘portable data’

57_{Regulation (EU) 2016/679 of the European parliament and of the Council of 27 April 2016 on the protection of natural}

persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General Data protection Regulation) [2016] OJ L119 (Hereon ‘GDPR’), art 20(1).

58_{Volunteered and Inferred categories.} 59_{Vestager Supra note 14.}

(22)

14

concerns the personal data60 which individuals –through their inability to ‘transfer’- consider to be a significant contributory switching-cost. This data is comprised of an individuals’ preferences, photos, contacts lists, status updates and all other personal information which individuals would want to ‘take-with’ them when migrating to another platform. Such an assertion is postulated with the notion that it would be illogical for individuals to want to take their metadata with them to alternative platforms and subsequently be targeted by alternative marketing strategies.61_{Evidently an explicit and authoritative distinction of what constitutes}

‘personal data’ is characterized as being a difficult feat to accomplish.62

The relation between data protection and competition law was until recently a mere

theoretical possibility. Recent developments most notably through that of the German Federal Cartel Office (FCO) have witnessed this possibility become a reality in which the scope of competition policy (ex post) can now –to an extent- be seen to overlap with privacy concerns (ex ante) where privacy forms a competition-parameter (see section 3.2.2). Where privacy concerns do constitute competitive-parameters, the GDPR’s framework may be used under competition law to oblige portability.63 Such a situation relates to circumstances where “data serves as a main input of its [the platform’s] products or services”.64

Currently data portability can be seen as only extending to infringements which relate to privacy as a parameter of competition law, however this is not to say that it cannot extend to other parameters. While competition law and the GDPR both employ the common aim of consumer welfare and market integration, the two regimes differ fundamentally in terms of policy and application, an example of which concerns the fact that while the GDPR applies only to personal data; competition law applies to all data.

2.3.1 Competition Law and the GDPR

60_{Directive Supra note 13 and accompanying text.}

61_{Unless individuals see value in targeted advertisements pursuant to which Vestager commented that “generally, they make}

life easier, by finding things that are right for me, without me needing to search for them”.

62_{‘First brief results of the public consultation on the regulatory environment for platforms, online intermediaries, data and}

cloud computing and the collaborative economy’ (26 January 2016) <https://ec.europa.eu/digital-single-market/news/first-brief-results-public-consultation-regulatory-environment-platforms-online-intermediaries> accessed 09 June 2016.

63_{Cooperative Supra note 45, 23; “statutory requirements stemming from other bodies of law may be taken into account, if}

only as an element of context, when conducting a legal assessment under competition law”.

(23)

15

While the primary aim of data protection law and the GDPR is to provide users with more control over their personal data, the primary goal of competition policy relates to consumer and economic welfare. Concerns over data portability -with regard to the market for social media- can be said to be two-fold relating to (1) competition policy, and (2) data protection. Issues regarding data portability in the market for social media illustrates an overlap of the two regimes; however, in light of recent developments- most notably the FCO initiating proceedings against Facebook based on the suspected violation of data protection laws,65_the

overlap which previously allowed for a distinction between EU competition law and data protection has started to blur. Both Competition law and the GDPR fail to account for

peculiarities relating to switching costs and lock-in, yet where the GDPR applies in a general fashion, Competition policy is applicable in certain situations and applies to all data; not just that of a person nature.

Notwithstanding the fact that Facebook asserts that data transfers do not take place in relation to personally identifiable data,66 there is uncertainty as to where the boundaries in terms of the permissibility of portability and transfer reside.

2.3.2 Drawing the ‘Portability’ Line

A delineation must be made distinguishing the types of data considered within the scope of portability and those which fall outside of it. While the data directive and the GDPR relate solely to ‘personal data’, competition law accommodates for all data having the

possibility of being classified as ‘portable’, including the information that would not benefit consumers if transferred to a competing entity. Consequentially a ‘line’ should be drawn under competition law concerning the extent to which ‘personal data’ is to be considered ‘portable’. Numerous assertions position this line to include data of a voluntary nature which individuals consider the main product of switching-costs,67 excluding data of an ‘aggregate’ nature. This again breeds controversy, as data of a shared nature may not be objectively

65_{‘Bundeskartellamt initiates proceedings against Facebook on suspicion of having abused its market power by infringing}

data protection rules’ (Bundeskartellamt, 2 March 2016) (Hereon ‘Bundeskartellamt’)

<http://www.bundeskartellamt.de/SharedDocs/Meldung/EN/Pressemitteilungen/2016/02_03_2016_Facebook.html> accessed 09 June 2016.

66_{‘Do advertisers have access to my personal information?’ (Help Centre)}

<https://www.facebook.com/help/146156215456790> accessed 09 June 2016.

67_{Perspective Supra note 56, 4; see also Mandating Supra note 49, 7; see also Inge Graef, ‘market Definition and Market}

Power in Data: The Case of Online Platforms’ (2015) Vol.38/4 Law and Economics Review (Hereon ‘Market Power’), 3 <http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2657732> accessed 09 June 2016.

(24)

16

considered property of the principle; and as such pictures of individuals taken by friends, messages about individuals and messages to which individuals have contributed may not be eligible for transfer. With respect to those individuals who are the objective owners,

consideration must be taken into account where they do not have a profile on the alternative platform to which the principle wishes to migrate, and how consent can extend to such circumstances.

In terms of virtual profiles, general consensus asserts a differentiation between two categories; ‘predictive’ and ‘explicit’ profiles. The former is established on the basis of observation and inference of individual and collective behavior over time, while the latter concerns profiles constructed around the personal data that individuals themselves have contributed,68 implying a separation between data which was previously characterized as ‘volunteered’ and that which is ‘observed’ and ‘inferred’ (see section 1.4), as well as providing a possible boundary pursuant to which the delineation of portability can be established. A complication however can be seen through the fact that social networks such as Facebook do not accommodate for the exportation of specific data sets and homogeneous formats. The current infantile state of policy pertaining to portability is reflected in the options, or lack of, that individuals are faced-with when wishing to differentiate between data-sets.

2.3.3 Portability and Data Protection

The Facebook/Whatsapp merger considered privacy-related concerns pertaining to data as ‘not fall[ing] within the scope of EU competition law’.69_{However, pursuant to recent}

events it would appear that is it not only a theoretical possibility for privacy concerns to constitute competitive parameters, but practically possible in circumstances where policy terms constitute unfair trading conditions with respect to consumers (see section 3.4.1.1). Where consumers are compelled to agree to terms and conditions which are difficult to comprehensively understand, the admissibility of such consent may be called into question where required by a dominant entity. As such -and as the door is opening for data protection and privacy concerns to be assessed under competition law-, the GDPR can be considered a

68_{Behavioural Advertising Supra note 33, 7.}

69_{European Commission Press Release, ‘Mergers: Commission approves acquisition of WhatsApp’ (Brussels, 3 October}

(25)

17

sufficient framework in order to facilitate portability through the application of competition policy. An obstacle in this respect however -with regard to evaluating whether dominant undertakings have abused their dominance with respect to privacy, thereby bringing about the need for portability-, is the need to establish the reality in which consumers base their

decision of ‘what platform to utilize’ on the privacy policies adopted by said platforms. Accordingly, where a privacy policy changes to the consumer’s detriment; data portability may constitute a viable solution by allowing for consumers to automatically migrate between platforms. A complication arises in utilizing the GDPR under competition law as it must be assumed that the GDPR’s principles have been adopted with competitiveness in-mind, and accurately reflect the threshold at which data collection and protection is competitive.

Notably, privacy is only one parameter to which portability may be applied, in which

numerous competition parameters outside of privacy concerns may bring about its invocation as a remedy.

2.3.4 Portability Outside of Data Protection

Portability regarding other competitive parameters which do not relate directly to data privacy are subject to extended hardships. This stems primarily from the fact that (in stark contrast with privacy and the GDPR), a normative framework of portability in terms of other competitive parameters is largely theoretical, and as such; largely inexistent. Consequentially, in the event that the answer of ‘portability’ is afforded to the question of ‘lock-in’ where the latter is procured out of circumstances non-related to privacy concerns, this would bring about the need for the formation of a completely new normative framework of portability, applicable to those other competitive parameters. This of course does not disqualify the possibility of ‘borrowing’ contextual as well as circumstantial analogies from the GDPR’s normative framework,70 but is to a greater extent unpredictable in terms of how it would be applied.

2.4 Consent

Consent holds a pivotal position in the acquisition of data and is a central problem faced by social networks. If used correctly, it becomes a tool permitting the data controlled

70_{Cooperative Supra note 45, 23.}

(26)

18

by the subject to be processed, but if used incorrectly- the subject’s consent constitutes an inappropriate basis for processing and only gives the illusion that the subject has control.71

There is difficulty in providing a meaningful provision of consent while allowing for individuals comprehend and effectively exercise control over their data where consent is a prerequisite for the use of services. This is illuminated by the fact that data may be used in ways unbeknownst to the individual, and for purposes to which consent was not

(comprehensibly) given.

The processing of personal data requires a legal basis. The most generally recognized elements thereof relate to ‘freely given specific and informed’ consent to data processing,72

where the use of the data does not extend past the legitimate purpose for which it was collected.73 In determining what is to be understood by these concepts; ‘freely given’ is said to pertain to an individual’s ability to exercise real choice, where the alternative would involve consequential coercion, intimidation or significant negative consequences where a failure to consent would arise.74 ‘Specific’ on the other hand relates to the fact that consent should be associated with the exact purpose of processing and not ‘blanket’ it’s every possible use,75 while ‘Informed’ is said to concern the fact that there must always be

information before the requirement of consent, in which an appreciation and understanding of what is being consented-to is realized in a clear manner.76 Consent is not always required for entities to process data collected from individuals, however where consent is required; its procurement must be genuine. The surrendering of consent relating to complex data

applications without an understanding of what is being agreed-to is not sufficient for its use in such a way.

Personal data may only be processed where “the data subject has unambiguously given his consent (…)”,77_{characterized as leaving ‘no doubt’ as to the intention of the data subject in}

delivering consent, wherein if a reasonable doubt exists as to an individual’s intention, so too

71_{Article 29 Data Protection Working Party, ‘Opinion 15/2011 on the definition of consent’ (13 July 2011) WP 187 (Hereon}

‘Consent’), 2 <http://ec.europa.eu/justice/policies/privacy/docs/wpdocs/2011/wp187_en.pdf> accessed 09 June 2016.

72_{Directive Supra note 13, art. 2(h); see also GDPR Supra note 57, preamble. 43.} 73_{Directive Supra note 13, art. 6(b).}

74_{Consent Supra note 71, 12.} 75_{Ibid, 17.}

76_{Ibid, 19.}

77_{Directive Supra note 13, art. 7(a); ‘unambiguously’ has been removed from the GDPR’s analogous Article 6(1)(a), and}

(27)

19

exists an ambiguity.78 Alongside this, consent must also be ‘explicit’,79 concerning an active response, either orally or in writing.

An individual’s (partial) refusal to consent to cookie notices should not hinder their webpage use. 80_{In terms of cookie consent, four main requirements in relation to consumers must be}

adhered-to regarding ‘specific information, prior consent, indication of wishes expressed by the user’s active behavior and the ability to choose freely’.81_{Platforms such as Facebook are}

evidently required to balance various externalities in terms of what is fair and legal in complying with contemporary practice pertaining to Big Data analytics. This ‘balancing’ includes elements such as whether the “data initially used in one context can be considered adequate, relevant, and proportionate to be reused in another context”, and “whether in the absence of obtaining consent from individuals, an organization can rely on its legitimate interest to process any data”.82

A question posited in the context of consent as a whole concerns that of ‘whether consent to the harvesting of individual data is genuinely procured where its acceptance is a prerequisite to utilizing the platform of a dominant undertaking’. Where it is not genuinely procured, this may constitute an unfair trading condition and an abuse of competition law where an

undertaking has a dominant position and consumers are locked-in by virtue of switching costs.

A dilemma exists in the supplementation of this framework to Competition law where the latter does not benefit from an analogous framework if its own, however this is not to say that sector-specific regulation and competition law cannot work in a complementary fashion.

2.5 Sub Conclusion

Lock-in regards the fact that individuals are unable to easily migrate from one platform to another, by virtue of the switching costs one would need to account-for in doing

78_{Consent Supra note 71, 21.}

79_{Directive Supra note 13, art. 8(2)(a).}

80_{Article 29 Data Protection Working Party, ‘Working Document 02/2013 providing guidance on obtaining consent for}

cookies’ (2 October 2013) WP 208 (Hereon ‘Cookie Consent’), 5 <http://ec.europa.eu/justice/data-protection/article-29/documentation/opinion-recommendation/files/2013/wp208_en.pdf> accessed 09 June 2016.

81_{Ibid, 3.}

(28)

20

so, which indirectly leads to the foreclosure of competing platforms. Portability is illustrated in the GDPR and provides a remedy for lock-in through the obligation for platforms to automatically transfer personal consumer data to the consumer’s rival platform(s) of choice. The delineation of the scope of portability is disputed, however, its possible application to (future) data protection is apparent. Whether its application extends to other qualitative factors is theoretically possible, yet this remains to be seen. Consent is pivotal to lock-in and remedial portability. Consent must be (a.o.) informed and unambiguous, however to the extent that consumers are unable to understand the terms and conditions to which they consent, the consent given might not be genuine where it is procured as a prerequisite to utilizing the platform of a dominant entity.

(29)

21

Chapter 3: Competition Law

In the determination of whether competition law is applicable to entities and their conduct, it should be reiterated that a distinction exists between two general categories of competition law. The first concerns agreements between undertakings and associations

thereof- vertically and/or horizontally situated on a relevant market constituting a competition matter governed by Article 101 TFEU, while the second pertains to situations of unilateral conduct where undertakings control a dominant position on a relevant market, and where the subsequent abuse of said dominance constitutes a competition-law matter under Article 102 TFEU.

Article 102 TFEU concerns;83

“Any abuse by one or more undertakings of a dominant position within the internal market or in a substantial part of it shall be prohibited as incompatible with the internal market in so far as it may affect trade between Member States

Such abuse may, in particular, consist in:

a) Directly or indirectly imposing unfair purchase or selling prices or other unfair trading conditions;

b) Limiting production, markets or technical development t the prejudice of consumers; c) Applying dissimilar conditions to equivalent transactions with other trading parties, thereby

placing them at a competitive disadvantage;

d) Making the conclusion of contracts subject to acceptance by the other parties of

supplementary obligations which, by their nature or according to commercial usage, have no connection with the subject of such contracts.”

The particularities surrounding big data and the example of Facebook in relation to it pertains to an Article 102 matter and to which the underlying preconditions of ‘undertaking(s)’

‘dominant’ on a ‘relevant market’ and the ‘abuse’ therein will forthwith be discussed. The notion of an effect on interstate trade will not be considered as its hypothetical existence is overtly apparent.

3.1 The Notion of an ‘Undertaking’

The concept of an ‘undertaking’ is not defined in the Treaties, its rationale however has been clarified in the Court’s case law to regard;

“Every entity engaged in an economic activity, regardless of the legal status of the entity and the way in which it is financed”.84

83_{Consolidated version of the Treaty on the Functioning of the European Union [2012] OJ C 326 (Hereon ‘TFEU’) art. 102.} 84_{Case C-41/90 Höfner and Elser v Macrotron [1991] ECR I-1979, 21.}

(30)

22

The term ‘economic activity’ has since been clarified to constitute “offering goods and services on a given market”.85_{Needless to say, social media entities such as Facebook}

suffice this aforementioned definition to be considered ‘undertakings’ within the meaning of Article 102 TFEU

Pursuant to establishing that Facebook suffices the requirements to be considered an

undertaking, the determination of a relevant market is necessary as its characterization forms the prerequisite to proving dominance and the subsequent abuse thereof.

3.2 The Relevant Market Dilemma

A relevant market is yet to be defined in terms of social media and personal data, however this does not mean that it is not possible to do so. For data to be analyzed as

intended, the same framework and rules as applied to other assets must also be applied to it.86 In the determination of a relevant market with regard to social media and big data, one should take into account the 2-sided nature of the market as illustrated in section 1.5. Although the Platform-Advertisement side of the market will be acknowledged in the interest of

comprehensiveness, more emphasis will be placed on the Platform-Consumer side in accordance with the research question.

3.2.1 Characterizing a Relevant Market

Predominantly dealt with from an economic perspective, there is no need for the Court to establish a determinate relevant market test to be systematically and consistently utilized in establishing competition concerns.87 The relevant market is conventionally split between geographical and product markets, wherein both must be determined in establishing the jurisdictional application of competition law.88 The product market ‘comprises all those products and/ or services which are regarded as interchangeable or substitutable by the consumer, by reason of the product’s characteristics, their prices and their intended use’.89

85_{Case C-180/98 Pavlov and Others v Stichting Pensioenfonds Medische Specialisten [2000] ECR I-6451, 75.} 86_{Strengthening Trust Supra note 24, 5-7.}

87_{Case T-427/08, Confédération européenne des associations d'horlogers-réparateurs (CEAHR) v Commission [2010] ECR}

II-5865, 66.

88_{Commission Notice on the definition of relevant market for the purpose of Community competition law (97/C 372/03)}

[1997] OJ C372 (Hereon ‘Relevant Market’), 2.

(31)

23

The geographical market relates to ‘the area in which the undertakings concerned are involved in the supply and demand of products or services, in which the conditions of competition are sufficiently homogeneous and which can be distinguished from neighboring areas because the conditions of competition are appreciably different in those area[s]’.90_The

geographical market can be said to include the community territory,91_{by virtue of the}

noteworthy presence entities such as Facebook hold in relation to online social media.

In the characterization of a relevant market, entities face 3 main competitive constraints which can be systematically identified; ‘demand-side substitutability’, ‘supply-side substitutability’ and ‘potential competition’.92 Supply-side substitutability is arguably not applicable in the context of an abuse of a dominant position in the social media market. Demand-side substitution is generally considered the most important constraint from an economic point of view and is determinable through a SSNIP-test assessment. The rationale behind the SSNIP test asserts that “a firm or a group of firms cannot have a significant impact on the prevailing conditions of sale, such as prices, if its customers are in a position to switch easily to available substitute[s]”.93

3.2.2 Platform-Consumer side

3.2.2.1 The SSNIP test

Demand-side substitution relates to the identification of products considered to be alternatives. The hypothetical monopoly (SNNIP) test concerning a small but signification non-transitory increase in price of 5-10% of the candidate products under review is regarded as “one way of making this determination”,94_{and is the method most widely utilized by the}

Court.95 This method of market delineation -although not explicitly adopted- is mentioned in numerous past proceedings.96 Where such a hypothetical price increase subsequently results in increased profits, this will illustrate the identification of a relevant market, as well as provide an insight into the entity’s position thereon. Where profits are not accrued, more products must be added to the candidate market. The SSNIP test essentially illustrates the

90_{Ibid, 8.}

91_{Antitrust Supra note 46, 9.}

92_{Relevant Market Supra note 88, 12-13.} 93_{Ibid, 13.}

94_{Ibid, 15-17.} 95_{Ibid, 38-43.}

(32)

24

price elasticity of demand and attempts to answer the question of what ‘consumers consider to be reasonably effective substitutes’.97

As the SSNIP test is ‘price based’, a complication is brought about in its application where the ‘product’ being defined does not have an established monetary value. Logically the SSNIP test must be modified accordingly, not least because it is being applied to a 2-sided market where it cannot capture the consequential effects of its use on the second side, but also to explore different values in accommodating for zero-price products and services. The SSNIP test conventionally suffers from inherent limitations such as that of the cellophane fallacy, pursuant to which; if the product or service being examined is already ‘priced’ at a monopoly level, its hypothetical ‘price’ increase may result in reduced profit as consumers refuse to purchase the product or service, resulting in the need to add products to the candidate market; wrongfully illustrating the relevant product market.98

In the market for social media, data utilization is not the ‘medium’ upon which the Platform-Consumer relationship is based but rather utilization of the platform’s services. Consequently the market for social media platforms should be identified. The rationale adopted pertains to the assertion that the market for data in this context may be determined by virtue of the market for social media, a prospect arguably supported by Commissioner Vestager in her statement that “it might not be easy to build a strong market position using data that quickly goes out of date”,99_{asserting the possibility for one to build a position on a (‘separate’)}

market through the use of data. Accordingly, the data market may be identifiable through the delineation of an associated market. This requires an assessment of the extent of data’s importance on the associated market in order to combat shortcomings in the form of the inclusion of other online markets into the contextual market for data.

3.2.2.2 Application of the SSNIP test

A modification of the SSNIP test’s price-based analysis in relation to demand-side substitutability -without disregarding the rationale behind a SSNIP-style analysis-, is achievable through shifting from price to a quality analysis, with the underlying notion that consumers -in their purchasing behaviour- will take into account the quality of the product in

97_{Relevant Market Supra note 88, 7.}

98_{Richard Whish, David Bailey, Competition law (8}th_{edn, Oxford University Press 2015) 32.} 99_{Vestager Supra note 14, 4.}

THE ILLEGALITY OF DATA LOCK-IN: AN ANALYSES OF WHETHER EUROPEAN COMPETITION LAW OBLIGES UNDERTAKINGS WHO ARE DOMINANT IN SOCIAL MEDIA TO ENGAGE IN DATA PORTABILITY

Abstract

Contents

Abbreviations