Governance Of Online Intermediaries: Observations From A Series Of National Case Studies

Authors: Urs Gasser & Wolfgang Schulz[1]

Editorial Note: Context, Character, and Purpose of the Case Study

This synthesis paper is part of a globally coordinated, independent academic research project by the Global Network of Interdisciplinary Internet & Society Research Centers (NoC). Facilitated by the Berkman Center for Internet & Society at Harvard University, the project is the first output of a larger policy-oriented research initiative that examines the rapidly changing landscape of online intermediary governance at the intersection of law, technology, norms, and markets. In concert with other research projects, it seeks to develop criteria, comparative methods, and a shared data repository, and to compile insights and lessons learned across diverse communities of knowledge aimed at informing and improving Internet policy-making globally.

The initial research output consists of a case study series exploring online intermediary liability frameworks and issues in Brazil, the European Union, India, South Korea, the United States, Thailand, Turkey, and Vietnam, and a synthesis paper that seeks to distill key observations and provide a high-level analysis of some of the structural elements that characterize varying governance frameworks, with a focus on intermediary liability regimes and their evolution.

The authors of this synthesis paper have participated in a multi-step process of in-person consultations and remote collaborations among a global team of researchers from the Network of Centers. Additionally, the paper is based on a set of broader questions regarding the role of online intermediaries in the digital age.[2]

The research effort is grounded in a diversity of global perspectives and collaborative research techniques, committed to objective and independent academic standards, and aspires to be useful, actionable, and timely for policymakers and stakeholders. More broadly, the Network of Centers seeks to contribute to a more generalized vision and longer-term strategy regarding the role of academic research, facilitation and convening, and education and communication in the Internet age. For additional information on the initiative, please contact Urs Gasser, Berkman Center for Internet & Society, at ugasser@cyber.law.harvard.edu

[1] Urs Gasser serves as Executive Director of the Berkman Center for Internet & Society at Harvard University and Professor of Practice at Harvard Law School. Wolfgang Schulz is Director of the Hans-Bredow Institute and the Humboldt Institute for Internet and Society, and Professor of Law at the University of Hamburg. The authors wish to thank Annie Pruitt for editorial assistance and the contributors to the Network of Center’s Online Intermediaries project for their collaboration, and the MacArthur Foundation for providing seed funding to the Berkman Center for this research initiative.
[2] The process is documented at: “Online Intermediaries: Functions, Values, and Governance Options”, The Global Network of Internet & Society Research Centers, 2014 https://drive.google.com/file/d/0B_ToTBKP5ITVWTl0UzV0U3B2RlU/view?usp=sharing.

Table of Contents

I. Introduction
II. Terminologies and Perspectives
A. Framework
B. Observations
III. Governance Structures and Models
A. Overview
B. Focus Areas
IV. Role of the Government
A. Functions
B. Branches
C. Enforcement
V. General Observations
A. Evolutionary Paths
B. Interplay Between Constitutional Rights and Intermediary Liability
VI. Conclusion
A. Summary
B. Future Considerations

I. Introduction

Online intermediaries in various forms – including search engines, social media, or app platforms – play a constitutive role in today’s digital environment. They have become a new type of powerful institution in the 21st century that shape the public networked sphere, and are subject to intense and often controversial policy debates. This paper focuses on one particular force shaping the emergence and future evolution of online intermediaries: the rapidly changing landscape of intermediary governance at the intersection of law, technology, norms, and markets. Building upon eight in-depth case studies and use cases, respectively, this paper seeks to distill key observations and provide a high-level analysis of some of the structural elements that characterize varying governance regimes, with a focus on intermediary liability regimes and their evolution.

Analyzing online intermediary governance issues from multiple perspectives, and in the context of different cultures and regulatory frameworks, immediately creates basic problems of semantic interoperability. Lacking a universally agreed-upon definition,[3] this synthesis paper and its’ underlying case studies are based on a broad and phenomenon-oriented notion of online intermediaries, as further described below. In methodological terms, the observations shared in this synthesis paper offer a selective reading and interpretation by the authors of the broader take-ways of a diverse set of case studies examining online intermediary governance frameworks and issues in Brazil, the European Union, India, South Korea, the United States, Thailand, Turkey, and Vietnam.[4] These case studies, in turn, have emerged in the context of an international research pilot by the Global Network of Internet & Society Research Centers (NoC), through a process of in-person consultations and remote collaborations among the researchers, and are based on a set of broader questions regarding the role of online intermediaries in the digital age.[5]

As a synthesis document, this paper is not aimed at providing a detailed or even comprehensive discussion of online intermediary governance, but it is rather intended to capture some of the insights and observations emerging from the analysis and comparative discussion of a limited – albeit diverse – sample of national regimes through an internationally coordinated academic research effort. The synthesis paper therefore does not cover all aspects of intermediary governance, but focuses on the issues that are examined in the case studies.

For a more detailed account of country-specific frameworks and their interaction with online intermediaries, as well as a deeper analysis of the issues highlighted in this paper, we refer to the set of case studies released in tandem with this synthesis. Together, these materials seek to complement important policy-oriented research efforts on online intermediaries by strengthening the evidence-base,[6] and contributing to our shared understanding of the various policy options available, including their impact on, and interplay with, online intermediaries.

[3] But see OECD. The Economic and Social Role of Online intermediaries (2010), 9. http://www.oecd.org/Internet/ieconomy/44949023.pdf: “’Online intermediaries’ bring together or facilitate transactions between third parties on the Internet. They give access to, host, transmit and index content, products and services originated by third parties on the Internet or provide Internet-based services to third parties.”
[4] See Appendices A – H for the full text of the case studies. Additionally, the case studies are available for comment at https://publixphere.net/i/noc/page/Online_Intermediaries_Research_Project_Case_Studies.
[5] The process is documented at: “Online Intermediaries: Functions, Values, and Governance Options”, The Global Network of Internet & Society Research Centers, 2014 https://drive.google.com/file/d/0B_ToTBKP5ITVWTl0UzV0U3B2RlU/view?usp=sharing.
[6] See in particular the efforts by the United Nationals Organization for Education, Science and Culture (UNESCO) (e.g. UNESCO, The Open Society Foundation, and the Internet Society, “The Freedom of Expression Online – The Role of Online intermediaries”. Executive Summary. Presented at IGF, Istanbul, September 5, 2014; the World Intellectual Property Organization (WIPO) (e.g. "Online intermediaries and Creative Content." World Intellectual Property Organization. http://www.wipo.int/copyright/en/Internet_intermediaries/); La Rue, Frank. "Report of the Special Rapporteur on the Promotion and Protection of the Right to Freedom of Opinion and Expression." 2013. http://www.ohchr.org/Documents/HRBodies/HRCouncil/RegularSession/Session23/A.HRC.23.40_EN.pdf; generally, the United Nations Department of Economic and Social Affairs (UNDESA), the Center for Democracy and Technology (CDT) (e.g. “Intermediary Liability | Center for Democracy & Technology.” https://cdt.org/issue/free-expression/intermediary-liability/.); the Association for progressive Communication (APC) (e.g. "Intermediary Liability." The Association for Progressive Communication. January 1, 2014. https://www.apc.org/en/irhr/intermediary-liability.), "Online intermediaries: Dilemma of Liability." Article 19. Accessed December 10, 2014. http://www.article19.org/resources.php/resource/37242/en/Internet-intermediaries:-dilemma-of-liability., “Intermediary Liability | Center for Democracy & Technology.” Accessed December 10, 2014. https://cdt.org/issue/free-expression/intermediary-liability/.; and “The Manila Principles On Intermediary Liability: Version 0.9,” Organization, December 1, 2014, https://docs.google.com/document/d/1kAkqgt3cRb65d8ik6vWYgpk6DYpP8ABA43ljgDiGOf8/edit?usp=sharing.

II. Terminologies and Perspectives

A. Framework

Recently, there has been an exponential increase in the use of the ambiguous term “intermediaries” in policy debates, which corresponds with the emergence of a new category of actor in the digitally networked environment and suggests a structural – and not just incremental – change in the information ecosystem. Despite a number of important studies in this area,[7] the phenomenon is still a moving target and the term “intermediary” often serves as a fallback phrase in the absence of a clear-cut definition. In certain policy contexts and jurisdictions, the term is sometimes used as a rhetorical tool to indicate that a given service does not fall within the category of traditional media services and – consequently – is not encompassed by traditional media regulation. These issues of qualification and categorization under existing laws and policies are another reason why the meaning of the term should be carefully reflected upon.

Various disciplines conduct research on online intermediaries and are likely to frame this research differently, as an initial literature review in the context of this project suggests.[8] But different approaches to the phenomenon also exist within individual domains or disciplines. From a legal perspective, for instance, there are various angles from which to look at the phenomenon, which might lead to different definitions. In countries that have enacted a specific framework for media regulation, the main difference might be the service’s impact on public opinion making. In regards to liability, the focus might lie in the control over and the technical ability to take content down. From the perspective of freedom of speech, intermediaries perform a new type of activity and create a space for individuals to express themselves online, a function that can be examined in greater depth.

Economists are interested in the location of online intermediary services within the value chain. It is characteristic that intermediaries are positioned between content providers and customers. Furthermore, the added value that intermediaries create might help to frame this group of services. From a media studies perspective, the way that services are integrated into daily life and the meaning that we collectively create by using intermediaries – e.g. the creation of new types of public spheres – is important. Intermediaries are associated with specific social practices and these practices, in turn, reflexively construct the service provided. In a similar fashion, studies on technical artifacts as institutions can ask how entities like algorithms, on which many online intermediary services are based, can be seen as institutions.

In the context of this cross-jurisdictional and cross-disciplinary research effort, we do not attempt to come up with a uniform definition of online intermediaries. Rather, we take a phenomenological approach and use socially and economically significant real world services as guiding examples. As such, this research effort has focused on services that are: (a) “in between” content and users; (b) show structural relevance to public communication (i.e. are not merely private); and (c) are not traditional journalistic-editorial (“media”) services. We are aware that the various elements of this definition refer to complex concepts, however, the definition serves as a workable proxy for the purposes of the case studies.

Core examples of “online intermediaries” that surfaced within this framework and were examined in the context of the case studies include search engines, micro blogs, social media, and user generated content platforms, among others. The intention of this research effort is not to limit the study of this subject to those cases, but use them pars pro toto to distill the essential characteristics of online intermediaries.

[7] Most notably the work of the OECD, see in particular OECD. The Economic and Social Role of Online intermediaries. 2010. p. 9. http://www.oecd.org/Internet/ieconomy/44949023.pdf and the other efforts mentioned supra, note 4. See also, “Fostering Freedom Online – The Role of Online intermediaries,” United Nations Organization for Education, Science, and Culture (UNESCO). 19 Jan. 2015. http://unesdoc.unesco.org/images/0023/002311/231162e.pdf.
[8] Kulk, Stefan, Tijana Milosevic, and Melinda Sebastian. “Online Intermediaries: A Thematic Analysis of their Social Role and Functions,” The Global Network of Internet & Society Center, Working Paper, 2014. https://docs.google.com/document/d/1h1WmsijKVWgrky2GHhqqI1lytZejkwnMc35CZKeeUXU/edit?usp=sharing.

B. Observations

With this tentative framework in mind, the analysis of the case studies leads to a series of high-level observations regarding questions of definition, categorization, and typology. At a basic level, the legal frameworks we reviewed revealed a significant variety in the definitions of online intermediaries.[9] In some jurisdictions, platforms that might be seen as “edge cases” under the parameters outlined above are defined as intermediaries and have been examined in the respective country cases study. The IT Act of India, for instance, sets forth a very broad definition of intermediaries, including telecommunication carriers, Internet service providers, and other backbone services.[10] In the Turkish case study, to take a second example, e-commerce platforms have played a key role in conflicts over intermediary liability and thus serve as an important use case in this research effort.[11]

The type of intermediary that plays a key role in a given policy or legal debate is context-specific and depends on various factors, particularly the country or region’s political economy. In the European Union, search engines such as Google have largely dominated legal and policy conversations. The most visible manifestation of this situation can be seen in the recent CJEU ruling on the so-called “right to be forgotten” or, more precisely, the “right to be delisted”.[12] While the former could be construed to mean the right of an affected person to have certain information completely wiped from the Internet, the “right to be delisted” constitutes the right to have information deleted from the listings of search engines and web catalogues, thus merely erasing links to the actual content. Therefore the term “right to be delisted” will be used in this document when referring to the case. In the U.S., user-created content platforms have been the focus in many of the recent law and policy debates. Overall, several case studies indicate a potential shift of attention in law and policy-making towards heavily algorithm-based intermediaries.

While lawmakers around the world realize that intermediaries play a special role, they tend not to form strict categories and define such services as they used to define broadcasting. If there are specific rules, they often link to abstractly defined actions (such U.S. safe harbor rules and the Marco Civil, for example). The European E-Commerce Directive is a hybrid that attempts to define types of services based on abstract prototypes. The various prototypes in mind (“caching” providers, host providers, access providers) might be one reason for the challenges with this approach to categorizing intermediaries.

Finally, the case studies demonstrate how law and policymakers, regulators, and Courts in different parts of the world continue to struggle with the task of framing the specific functions that different types of intermediaries fulfill. In some instances, the functional approach is avoided altogether and replaced by more familiar questions of definition, as in the case of the recent CJEU ruling. Similarly, the possible effects of interventions are an area of concern and debate given the dynamic nature of the service ecosystem.

[9] See also “The Manila Principles On Intermediary Liability: Version 0.9,” December 1, 2014. https://docs.google.com/document/d/1kAkqgt3cRb65d8ik6vWYgpk6DYpP8ABA43ljgDiGOf8/edit?usp=sharing. [10] Arun, Chinmayi, and Sarvjeet Singh. “Online Intermediaries Case Studies Series: Online Intermediaries in India”, The Global Network of Internet & Society Research Centers (2015), 8.
[11] Beceni, Yasin and Nilay Erdem. “Online Intermediaries Case Studies Series: Turkey (eBay Case)”, The Global Network of Internet & Society Research Centers (2015).
[12] Google Spain SL, Google Inc. v Agencia Española De Protección De Datos (AEPD), Mario Costeja González. European Court of Justice. 13 May 2014. Europa.eu. European Union, n.d. Web. 9 Dec. 2014. http://curia.europa.eu/juris/document/document_print.jsf?doclang=EN&text&pageIndex=0&part=1&mode=DOC&docid=152065&occ=first&dir&cid=437838.

III. Governance Structures and Models

A. Overview

The different systems for intermediary governance can be divided into two very broad groups. First, there are systems where intermediaries are explicitly addressed, and where there is a governance system especially designed to deal with intermediaries. Second, there are systems where general rules are applied to intermediaries.

The significance of this distinction becomes apparent when analyzing the situation in Brazil before and after the Marco Civil came into force. Before the enactment of the Marco Civil, decisions on the liability of online intermediaries were influenced by three completely different understandings applied by Brazilian Courts.[13] One understanding led to an exemption of the provider from any liability for third party behavior; a second interpretation enforced a stricter liability regime grounded in the concept of risk of the providers’ activity; and a third would trigger the liability of the provider to the existence of culpability on its part.[14] With the Marco Civil now in force, there is a special civil liability regime for intermediaries. However, it is too soon to foresee what rules the Courts will apply in specific cases based on the new law, and what types of intermediaries will be covered. However, though it is unclear how this will be applied in practice, at least there is now a coherent regulatory structure within Brazilian law.

The U.S. can serve as an example of a system with an explicit and far ranging special regulatory framework for intermediaries. Section 230 of the Communications Decency Act and Section 512 of the Digital Millennium Copyright Act[15] are pivotal pieces of regulation constituting a “safe harbor” for online intermediaries. The same is – at least in principle – true in the European Union, where the E-Commerce Directive[16] provides specific liability exemptions for online intermediaries. However, due to the limited scope of the exemption (injunctions are, for example, not covered by the regime) and – to some degree – uncertainty about how to apply the exemptions, European harmonization could not prevent the emergence of a rather fragmented system regarding intermediary liability.[17]

Studying the country cases presented here leads to the insight that different types of conflicts are predominant in different countries. While in Vietnam – and to some extent in Thailand and India – intermediaries mainly face takedown requests on grounds of state interest (in a broad sense, including the protection of the honor of the king in Thailand), in most of the other countries examined in this study it is user-user divergences that fuel the majority of conflicts. The first pattern may stem from a more interventionist approach of some governments; however, it may also be based on the differing strengths of personal rights and different cultures of the complainants in the countries examined. It is interesting to note that claims based on copyright are dealt with by a separate liability system in basically all of the countries covered by this study.

Another relevant distinction is whether a liability regime provides a safe harbor for intermediaries or not. Section 512 of the Digital Millennium Copyright Act (DMCA) in the U.S. is probably the most prominent example of a safe harbor clause. The basic rule here is that if there is no content curation, there is no liability. Thus, under the DMCA, the way that providers treat content is key. To some extent there is a safe harbor for intermediaries that do not exercise editorial control under the Indian IT Act.[18] Considering algorithmic journalism and similar developments, it will become rather difficult to draw such distinctions regarding curation in future.

The above-mentioned safe harbor clause in the E-Commerce Directive of the EU shows that not all harbors are equally protected against the strong wind of third-party claims. The directive defines different types of intermediary roles, and categorization into one of these roles typically leads to a provider being held not liable. Whether search engines fall into one of those categories and can consequently profit from safe harbor provisions is not entirely clear. However, while the explicit objective of the regulation was to create a safe harbor, especially to promote innovation in Europe, there are only limited cases in which the directive in fact provided such a “safe harbor.”[19]

Another significant characteristic of any intermediary governance system is whether it establishes a notice-and-take-down procedure or not, and whether this procedure is set up as a condition for obtaining optional safe harbor protection or as a mandatory test of liability (i.e., failure to respond to a lawful notice immediately triggers liability). There can be an explicit regime, like in India,[20] or a situation like in Europe, where the CJEU ruling on data protection and search engines[21] resulted in a de-facto notice-and-take-down-procedure without a clear legal basis. The CJEU claims that search engine providers are controlling data processing when a user searches for a name of an individual, and that they must consider requests to delist names from search results. The decision is based on the European Data Protection Directive and answers questions submitted by a Spanish Court. Since it enables any person to request that a link be removed from search results, the ruling has been associated with the “right to be forgotten.” This controversial judgment[22] requires search engine operators to establish a notice-and-take-down-procedure to comply with the European data protection framework.

[13] Lemos, Ronaldo, and Carlos Affonso Pereira De Souza. “Online Intermediaries Case Studies Series: Brazilian Courts and the Internet – Rulings Before and After the Marco Civil on Intermediary Liability”, The Global Network of Internet & Society Research Centers (2015), 2.
[14] Lemos, Ronaldo, and Carlos Affonso Pereira De Souza. “Online Intermediaries Case Studies Series: Brazilian Courts and the Internet – Rulings Before and After the Marco Civil on Intermediary Liability”, The Global Network of Internet & Society Research Centers (2015), 2.
[15] 47 U.S. Code § 230 - Protection for private blocking and screening of offensive material; http://www.gpo.gov/fdsys/granule/USCODE-2011-title47/USCODE-2011-title47-chap5-subchapII-partI-sec230/content-detail.html.
[16] European Parliament. Directive 2000/31/EC, “Directive on Electronic Commerce,” European Union. June 8, 2000, http://eur-lex.europa.eu/legal-content/EN/NOT/?uri=CELEX:32000L0031.
[17] Angelopoulos, Christina. Beyond the Safe Harbours: Harmonising Substantive Intermediary Liability for Copyright Infringement in Europe. SSRN Scholarly Paper. Rochester, NY: Social Science Research Network, November 28, 2013. http://papers.ssrn.com/abstract=2360997.
[18] Arun, Chinmayi, and Sarvjeet Singh. “Online Intermediaries Case Studies Series: Online Intermediaries in India”, The Global Network of Internet & Society Research Centers (2015), 8-9.
[19] For the effect of the safe harbor clause (and its limits) on a European level see in particular Case C- 324/09, L ‘Oréal v eBay, 12 July 2011, Case C-70/10, Scarlet v SABAM, 24 November 2011 and Case C- 360/10, SABAM v Netlog, 16 February 2012.
[20] Arun, Chinmayi, and Sarvjeet Singh. “Online Intermediaries Case Studies Series: Online Intermediaries in India”, The Global Network of Internet & Society Research Centers (2015), 21.
[21] Google Spain SL, Google Inc. v Agencia Española De Protección De Datos (AEPD), Mario Costeja González. European Court of Justice. 13 May 2014.
[22] Kuczerawy, Aleksandra, and Ausloos, Jef. “Online Intermediaries Case Studies Series: European Union and Google Spain”, The Global Network of Internet & Society Research Centers (2015), 19-20.

B. Focus Areas

Internet liability regimes can serve as a model for contextual regulation. Laws do not regulate the behavior of an operator or an intermediary directly by prohibiting or ordering a specific behavior.[23] Rather, the core of a liability regime is generally a set of conditions under which the operator will be held liable for third party content. Depending on mechanisms for enforcement and implementation, these regimes can lead to specific governance structures being formed within the operator of an intermediary, such as a notice-and-take-down-procedure to deal with user-user-conflicts on an online platform. Most countries examined in the case studies do not directly govern intermediaries but rather govern them indirectly via mechanisms of contextual regulation. While licensing regimes exist in some jurisdictions (Thailand can serve as an example),[24] it is noteworthy that they are reportedly used less for direct regulation in the traditional sense – for instance via licensing conditions – but rather as an enforcement mechanism.

The outcome of such indirect contextual regulation very much depends on the incentives and disincentives created by the system. For the functioning of an intermediary governance system, setting the right incentives is key. First and foremost this is true for the operator of the intermediary. The case of Korea[25] demonstrates that this is a difficult task in the complex environment defined by Internet regulation. The clause in the respective law in Korea[26] mentioning “temporary action” does create an incentive for the operator of an intermediary to remove content after having received notice, regardless of whether the content is legal or illegal. While this does not seem to be the intent of the law, it is rational for intermediary operators to act this way if they want to avoid liability. Looking at incentives created by the system, it is instructive to look at the users’ end as well. The notice and-take-down system in Turkey creates greater incentives for users who believe his or her rights have been infringed upon to file a lawsuit directly with the Courts, rather than use the notice-and-take-down-system.[27] Obviously there is the risk of undesirable secondary effects in this situation because it is hard for a lawmaker to anticipate the actions of intermediary operators, if the lawmaker even tries to do this – which is not always the case with liability rules. In India, the Supreme Court will hear a case in the spring of 2015 regarding the rules that the government enacted under the safe harbor clause, which have been criticized for creating incentives to remove all content – illegal or legal – in the event of a notice.[28]

This reflection about incentives already focused on the different actors in a governance system. Looking at the governance structure at large, which emerges from – or is at least influenced by – a liability regime, reveals the structure of rules and the roles of different actors in such a system. In many cases the structure does not seem to be the result of a regulatory strategy, but a result of the interplay of different actions. Take the CJEU ruling – the search engine case – as an example again.[29]

The CJEU decision can be seen as European institutions showing Google its limits. However, in doing so, the ruling moves the responsibility for deciding user-user-conflicts – i.e. conflicts between the person affected vs. the owner of the web page that contains the information about that person – to the search engine operator, since the operator is responsible under the data protection regulation for its search results. Any decision by Google can, of course, be subject to scrutiny by the data protection officer or the Courts, however, the initial decision remains with Google. Not only that, in absence of more detailed criteria – in the directive or in the Court’s reasoning – regarding how to balance the interest of the person that wants a link removed and the interest of the owner of the respective web page or the general public at large to have access to this web page, Google has to come up with rules regarding how to solve such conflicts and balance the rights involved. Unintentionally the CJEU has created a mandatory notice-and-take-down-procedure, the rules of which are governed by search engines.

This also tells the story about the role of another type of actor in the governance system: the Courts. While the CJEU ruling on search engines is a very special case, in many countries we can see the relevance of single Court’s decisions in shaping the given intermediary governance system. What we can learn from studying this aspect is that it puts a burden on the Courts to develop a coherent liability system in this complex environment (remember the three different ways to apply the general liability rules in Brazil). Furthermore, the Courts with their procedures and instruments to gain and process knowledge are not designed to anticipate the secondary effects of their judgments. The incentives created by judgments and the governance structure emerging from this cannot easily be anticipated by Courts.

It has already been mentioned that most of the intermediary governance systems contain notice-and-take-down-procedures – be it intentionally designed, as a de facto development, an optional procedure for obtaining a safe harbor, or a mandatory test of intermediary liability – as an essential part of their structure, and there is a great variety of such systems. The fact that notice-and-take-down has become a very “fashionable” way to treat user-user conflicts on intermediary platforms has been criticized from a normative standpoint, to the extent that the procedure becomes a mandatory test of intermediary liability. The report by Frank La Rue[30] clearly states that an operator of an intermediary should not be put in the position to decide whether to remove content or not; it should be up to an independent Court or another independent body within a government to judge the legality or illegality of making content available. A mandatory notice-and-takedown procedure is likely to violate La Rue’s recommendation.

Two aspects of a notice-and-take-down-procedure seem to be significant. The first is the design of the procedure; models range from having no procedural requirements at all to models with fine-tuned procedures set out in a bylaw, such as that established by the administration in India.[31] The other significant aspect is the treatment of the operator of the affected web site and the general interest in easy access to online information. Only very rarely is there a mandatory notice to the operator of the web page before a take-down happens,[32] or is a structured appeal procedure launched by the operator in question.

Depending on the nature of the claim, the procedure adopted by an intermediary to deal with notices can be an automated system like ContentID by YouTube, which allows YouTube to identify copyright protected content and have it removed, or a manual system, like the procedure which as been established by Google to deal with the over 160,000 requests as of end of October 2014 following the CJEU ruling. Alternatively, hybrid systems exist where a large amount of potentially protected content will be dealt with automatically, but the hard cases are treated manually.

[23] See Baldwin, Robert, and Martin Cave. Understanding Regulation: Theory, Strategy, and Practice. Oxford: Oxford University Press, 1999.
[24] Ramasoota, Pirongrong. “Online Intermediaries Case Studies Series: Online Intermediary Liability in Thailand”, The Global Network of Internet & Society Research Centers(2015), 3.
[25] Park, Kyungg-Sin. “Online Intermediaries Case Studies Series: Intermediary Liability – Not Just Backward but Going Back”, The Global Network of Internet & Society Research Centers (2015).
[26] Park, Kyung-Sin. “Online Intermediaries Case Studies Series: Intermediary Liability – Not Just Backward but Going Back”, The Global Network of Internet & Society Research Centers (2015), 6.
[27] Beceni, Yasin and Nilay Erdem. “Online Intermediaries Case Studies Series: Turkey (eBay Case)”, The Global Network of Internet & Society Research Centers (2015), 12.
[28] cf. Arun, Chinmayi, and Sarvjeet Singh. “Online Intermediaries Case Studies Series: Online Intermediaries in India”, The Global Network of Internet & Society Research Centers (2015).
[29] Google Spain SL, Google Inc. v Agencia Española De Protección De Datos (AEPD), Mario Costeja González. European Court of Justice. 13 May 2014. For more detail see: Kuczerawy, Aleksandra, and Ausloos, Jef. “Online Intermediaries Case Studies Series: European Union and Google Spain”, The Global Network of Internet & Society Research Centers.
[30] La Rue, Frank. "Report of the Special Rapporteur on the Promotion and Protection of the Right to Freedom of Opinion and Expression." United Nations Office of the High Commissioner of Human Rights. 2013. http://www.ohchr.org/Documents/HRBodies/HRCouncil/RegularSession/Session23/A.HRC.23.40_EN.pdf.
[31] Arun, Chinmayi, and Sarvjeet Singh. “Online Intermediaries Case Studies Series: Online Intermediaries in India”, The Global Network of Internet & Society Research Centers (2015), 5.
[32] U.S. 17 U.S.C. 512(g)(2)(A) (g) Replacement of Removed or Disabled Material and Limitation on Other Liability.—(1) No liability for taking down generally. — Subject to paragraph (2), a service provider shall not be liable to any person for any claim based on the service provider's good faith disabling of access to, or removal of, material or activity claimed to be infringing or based on facts or circumstances from which infringing activity is apparent, regardless of whether the material or activity is ultimately determined to be infringing. (2) Exception. — Paragraph (1) shall not apply with respect to material residing at the direction of a subscriber of the service provider on a system or network controlled or operated by or for the service provider that is removed, or to which access is disabled by the service provider, pursuant to a notice provided under subsection (c)(1)(C), unless the service provider — (A) takes reasonable steps promptly to notify the subscriber that it has removed or disabled access to the material; Holland, Adam, Chris Bavitz, Jeff Hermes, Andy Sellars, Ryan Budish, Michael Lambert, and Nick Decoster. “Online Intermediaries Case Studies Series: Intermediary Liability in the United States”, The Global Network of Internet & Society Research Centers (2015), 11-13.

IV. Role of the Government

The case studies reveal that governments – in addition to technological and market factors – are among the most important forces that shape the online intermediary landscape of a given country. The respective roles government can play are rather diverse and often overlapping, ranging from “governments as users” to “governments as regulators” of intermediaries. Focusing on the latter, the case studies demonstrate that, even within the role of the government as a regulator of online intermediaries, we can find important functional nuances in terms of different manifestations and interpretations of this role. Further, the case studies suggest that different institutions within the government might be involved in the respective online intermediaries governance regime, depending on the underlying regulatory model and strategy (see previous section). In some countries, government agencies are the key regulators; other governance regimes heavily rely on Courts. The analysis also points to structural similarities and differences among the case studies when it comes to the specific approach to compliance and enforcement, ranging from emphasis on technical means to licensing requirements. The following paragraphs highlight some of the key findings in each of these issue areas.

A. Functions

The case study series reveals that governments have varying motives for regulating online intermediaries. In broad terms of regulatory theory,[33] the primary reasons to intervene and regulate might have to do with externalities (e.g. compelling online intermediaries to bear the full costs of service rather than pass on to third parties), can be motivated by the desire to ensure certain levels of “essential” services (e.g. creation of and access to a diverse information ecosystem with multiple sources), or may be aimed at balancing unequal bargaining power (e.g. to protect vulnerable interests or populations, such as children), to name just a few examples. Viewed from a broader functional angle, however, the case studies suggest that the majority of governance models outlined above fall into three in practice overlapping but nonetheless analytically distinct categories: enabling, leveling, or constraining.

The most prominent example where the governance model serves largely the function of an enabler is the U.S. legal framework. As already mentioned above and described in detail in the respective country case study,[34] the U.S. framework is characterized by extensive safe harbors that dramatically limit the liability exposure of online intermediaries. The case study analysis and various other (including empirical) studies suggest that this particular governance arrangement has enabled the flourishing and growth of online intermediaries in the U.S. and, as a result, promoted the functions performed by online intermediaries.[35] While the historic motives for introducing these liability limitations were rather nuanced (in the case of the U.S. Communications Decency Act [CDA], for instance, the lawmaker wanted to enable content self-regulation by online intermediaries without exposing them to liability),[36] contemporary policy debates refer to this enabling function largely in relation to either economic benefits (e.g. incentives to innovate without fear of liability) or in the context of fundamental rights (e.g. elimination of chilling effects).[37]

Another function that online intermediary governance models (in general) and liability regimes (in particular) can perform is the role of a leveler. Traces of such a leveling function can be found in several countries with notice-and-takedown systems where the governance model is targeting online intermediaries as “the in between” to strike a balance between the interests of different parties, for instance between copyright owners and users in the realm of copyright. The CJEU’s right to be delisted decision might be seen as another manifestation of such an approach, aimed at leveling the playing field (“fair balance” in the words of the CJEU) between the legitimate interests of the Internet users potentially interested in having access to information and the data subject’s fundamental rights. As these two examples indicate, the leveling function of online intermediary governance models can either be implemented through a (generalized) rule such as a DMCA-style notice-and-takedown mechanism, or based on a standard that requires a case-by-case analysis, as in the case of the CJEU’s right to be delisted decision.

Third, governance models – especially in the form of liability regimes in the context of this study – typically perform a constraining function by ordering online intermediaries to take specific action or implement certain measures. Even leveling regimes often perform a constraining function, as in the case of notice-and-take-down regimes where online intermediaries have to meet certain obligations in order to benefit from safe harbor protection. But the case studies have also revealed situations where the constraining effects are more specific or targeted. In the case of Thailand, for instance, the law directly imposes content liability on online intermediaries to preserve the public order (lèse majesté)[38] or enable the control of the flow of information (through censorship and surveillance) under the coup-ruled government. Blocking statutes such as the Turkish Internet Law are highly visible and controversial examples where law serves predominantly a constraining function in the online intermediaries space.[39] The licensing regime in Vietnam imposes hard constraints under which online intermediaries have to operate, to give another example from the case study series.[40]

[33] See generally, e.g., Baldwin, Robert, and Martin Cave. Understanding Regulation: Theory, Strategy, and Practice. Oxford: Oxford University Press, 1999.
[34] Holland, Adam, Chris Bavitz, Jeff Hermes, Andy Sellars, Ryan Budish, Michael Lambert, and Nick Decoster. “Online Intermediaries Case Studies Series: Intermediary Liability in the United States”, The Global Network of Internet & Society Research Centers (2015).
[35] See, e.g., Bramble, Nicholas. "Safe Harbors and the National Information Infrastructure." Hastings Law Journal 64, no. 325 (2013). http://www.hastingslawjournal.org/wp-content/uploads/2014/04/Bramble-64.2.pdf.
[36] Cannon, Robert. "The Legislative History of Senator Exon's Communications Decency Act: Regulating Barbarians on the Information Superhighway." Federal Communications Law Journal 51 (1996). http://www.cybertelecom.org/cda/cannon2.htm.
[37] See, e.g., Bankston, Kevin, David Sohn, and Andrew McDiarmid “Shielding the Messengers: Protecting Platforms for Expression and Innovation.” Center For Democracy and Technolocy. December 2012.
https://www.cdt.org/files/pdfs/CDT-Intermediary-Liability-2012.pdf, But see Seltzer, Wendy. "Free Speech Unmoored in Copyright’s Safe Harbor: Chilling Effects of the DMCA on the First Amendment." Harvard Journal of Law & Technology 24 (2010): 171. http://jolt.law.harvard.edu/articles/pdf/v24/24HarvJLTech171.pdf.
[38] Ramasoota, Pirongrong. “Online Intermediaries Case Studies Series: Online Intermediary Liability in Thailand”, The Global Network of Internet & Society Research Centers (2015), 4.
[39] Beceni, Yasin and Nilay Erdem. “Online Intermediaries Case Studies Series: Turkey (eBay Case)”, The Global Network of Internet & Society Research Centers (2015).
[40] Nguyen, Thuy. “Online Intermediaries Case Studies Series: Roles and Liabilities of Online Intermediaries in Vietnam – Regulations in the Mixture of Hope and Fear”, The Global Network of Internet & Society Research Centers (2015).

B. Branches

Looking at the role of governments as regulators, the case studies show that different branches of the government may serve as core pillars of a given online intermediary governance system. The series also demonstrates that the basic layout and different degrees of government involvement lead to key questions regarding incentives, legitimacy, accountability, and transparency. In addition to these fundamental issues, the case studies also hint towards a rather underexplored dimension of the governance problem: the role of knowledge when it comes to the regulation of online intermediaries, as such expertise – for instance with respect to the understanding of how different types of intermediaries technically work – might be distributed unequally across the different branches of the government that are involved in the respective governance models.

Most of the governance models studied in the context of this research project heavily rely on the Court system to put these models aimed regulating online intermediaries into practice. Until the recent enactment of the Marco Civil, Brazil was among the countries where online intermediary governance almost entirely resided in the realm of Courts. An alternative type of regime puts emphasis on government agencies when it comes to online intermediaries. With respect to non-copyright issues, Korea is an example where a government agency, in form of the Korean Communication Standards Commission, plays an important role within the intermediary governance framework.[41] An extreme version of a government agency-based governance regimes are countries with licensing requirements. In Vietnam, for instance, the providers of online social networking sites and general news websites have to obtain a license from the government before offering such services.[42]

Court-centric regimes are characteristic for democratic countries, while agency-focused intermediary governance frameworks are more prevalent in countries with limited rule of law. The U.S. governance system with its heavy reliance on Courts is at one end of the spectrum in the case study series, while Thailand with its tight control over online intermediaries through the National Council for Peace and Order marks the other.[43] Further, Court-based governance regimes play a particularly important role with respect to copyright issues, as even some countries with relatively strong government agency involvement in non-copyright issues refer to Courts in this area, as the case of Korea illustrates.[44]

But even in countries with largely Court-centric regimes lines might be blurring. While U.S. intermediary governance heavily relies on Courts, governmental agencies can play a prominent role at least when it comes enforcement, as the role of state government in the context of Section 230 CDA demonstrates.[45] Similarly, government agencies in the form of data protection authorities are important players in the EU when it comes to online intermediary governance.

[41] Park, Kyung-Sin. “Online Intermediaries Case Studies Series: Intermediary Liability – Not Just Backward but Going Back”, The Global Network of Internet & Society Research Centers (2015).
[42] Nguyen, Thuy. “Online Intermediaries Case Studies Series: Roles and Liabilities of Online Intermediaries in Vietnam – Regulations in the Mixture of Hope and Fear”, The Global Network of Internet & Society Research Centers (2015), 3.
[43] See Holland, Adam, Chris Bavitz, Jeff Hermes, Andy Sellars, Ryan Budish, Michael Lambert, and Nick Decoster. “Online Intermediaries Case Studies Series: Intermediary Liability in the United States”, The Global Network of Internet & Society Research Centers (2015); and Ramasoota, Pirongrong. “Online Intermediaries Case Studies Series: Online Intermediary Liability in Thailand”, The Global Network of Internet & Society Research Centers (2015).
[44] Park, Kyung-Sin. “Online Intermediaries Case Studies Series: Intermediary Liability – Not Just Backward but Going Back”, The Global Network of Internet & Society Research Centers (2015).
[45] Holland, Adam, Chris Bavitz, Jeff Hermes, Andy Sellars, Ryan Budish, Michael Lambert, and Nick Decoster. “Online Intermediaries Case Studies Series: Intermediary Liability in the United States”, The Global Network of Internet & Society Research Centers (2015), 6.

C. Enforcement

The previous sections already clearly illustrates that governments not only set the general – and at times specific – framework conditions under which online intermediaries operate, but are also instrumental when it comes to the implementation and enforcement of a given governance model. With respect to compliance and enforcement issues, a number of observations gained from the case study series are noteworthy.

At the most abstract level, the comparative analysis of different governance regimes indicates that the incentive structures created by the governments – whether by design or through mere practice – are key in understanding compliance with and enforcement of online intermediary governance frameworks. A key issue identified across the case studies is the question of whether a particular government creates a symmetric or asymmetric incentive structure for online intermediaries to take down content or leave it up in order to avoid liability. In the U.S., for instance, Section 230 CDA provides a symmetric incentive structure in the sense that Courts have been consistent about immunizing online intermediaries from liability as long as they did not author the content in question – whether they take it down, leave it up, or even restore content that was taken down.[46] In contrast, the governance models in India, Korea, and Thailand create asymmetric incentive structures, where intermediaries are incentivized to take down content in order to avoid liability, even if it results in over-compliance.[47]

A second observation related to asymmetric incentives and resulting compliance levels concerns local versus international online intermediaries. The case studies indicate that instances in which licensing requirements apply de facto only to local but not to international intermediaries lead to more compliance, or arguably even over-compliance, with government requests among these local intermediaries. The case study from Thailand is the most prominent example that highlights this asymmetry between local and international players.

Third, the case studies illustrate not only the different enforcement regimes and (e.g. ex post versus ex ante) strategies, including incentives and actors involved, but also indicate the range of enforcement techniques that can be utilized as part of the different governance models. The previous sections have already highlighted the role of licensing requirements as an enforcement tool, particularly in the cases of Turkey and Thailand.[48] Another interesting theme emerging from the case study analysis relates to the role of algorithms in enforcement. The phenomenon of computational compliance has become most visible in the context of the U.S. case study, where software plays a key role in dealing with large-scale problems of copyright infringement over user-created content platforms, specifically YouTube.[49] Algorithms not only play a role in “private ordering” a la YouTube, but also when it comes to government-imposed monitoring and filtering obligations, as the reports from Thailand, Turkey, and India demonstrate.[50]

Finally, and related to the previous issues, the case studies point out the importance of costs, in terms of both money or time, when it comes to compliance and enforcement. Again, the role of cost is multi-faceted and context-specific. For instance, the Turkish case study demonstrates that uncertainties surrounding the notice-and-take-down system and the fact that a criminal proceeding can be launched without costs leads to a preferred activation of the judicial system over private mechanisms.[51] The contrast between automated compliance and enforcement in response to copyright issues on YouTube, versus the human and labor-intense review of takedown requests that attempt to balance user interests under the CJEU’s right to be delisted, highlights yet another important dimension of the cost argument when it comes to online intermediary governance.

[46] Holland, Adam, Chris Bavitz, Jeff Hermes, Andy Sellars, Ryan Budish, Michael Lambert, and Nick Decoster. “Online Intermediaries Case Studies Series: Intermediary Liability in the United States”, The Global Network of Internet & Society Research Center (2015), 5-7.
[47] See Arun, Chinmayi, and Sarvjeet Singh. “Online Intermediaries Case Studies Series: Online Intermediaries in India”, The Global Network of Internet & Society Research Centers (2015); and Ramasoota, Pirongrong. “Online Intermediaries Case Studies Series: Online Intermediary Liability in Thailand”, The Global Network of Internet & Society Research Centers (2015).
[48] See Beceni, Yasin and Nilay Erdem. “Online Intermediaries Case Studies Series: Turkey (eBay Case)”, The Global Network of Internet & Society Research Centers(2015); and Ramasoota, Pirongrong. “Online Intermediaries Case Studies Series: Online Intermediary Liability in Thailand”, The Global Network of Internet & Society Research Centers (2015).
[49] Holland, Adam, Chris Bavitz, Jeff Hermes, Andy Sellars, Ryan Budish, Michael Lambert, and Nick Decoster. “Online Intermediaries Case Studies Series: Intermediary Liability in the United States”, The Global Network of Internet & Society Research Centers (2015), 31-34.
[50] See Ramasoota, Pirongrong. “Online Intermediaries Case Studies Series: Online Intermediary Liability in Thailand”, The Global Network of Internet & Society Research Centers (2015); and Beceni, Yasin and Nilay Erdem. “Online Intermediaries Case Studies Series: Turkey (eBay Case)”, The Global Network of Internet & Society Research Centers (2015); and Arun, Chinmayi, and Sarvjeet Singh. “Online Intermediaries Case Studies Series: Online Intermediaries in India”, The Global Network of Internet & Society Research Centers (2015).

V. General Observations

A. Evolutionary Paths

The analysis so far has focused on the governance structure and was therefore based on more of a static view. Another perspective from which intermediary governance can be analyzed is the process of development. A first and rather obvious observation is that the political discourse in the countries covered by this study recognized the relevance of intermediary governance at different points of time. So it may be fair to say that the systems are not equally mature. The U.S. appears to have been fast in addressing the issue, and as a result the system has been in force for several years and proven to be relatively stable. Other countries are still in the process of designing a system.

A less obvious, but also significant aspect seems to be the cultural context. Protecting the honor of the king in Thailand, for example, is deeply rooted in Thai society and has to be guaranteed against defamation online and offline. Consequently the role of all actors, including operators of intermediaries, is addressed. Countries with an aspiration to govern society more strictly than a western democracy face the dilemma of finding a way to govern the Internet – including intermediaries – without tampering with innovation and the economic potential of the Internet. Vietnam[52] can serve as an example of a country grappling with such a balance. The U.S., additionally, bases their regulation on a shared understanding on the importance of freedom of speech; thus the cultural context again is key.

What we can see in the development of intermediary governance, as well as in other sectors of regulation, is that sometimes single events change the development path. A significant event has been the CJEU ruling on search engines, which has fuelled the debate on the responsibility of operators of intermediaries in Europe and beyond. This can even affect the construction of the relationship of whole bodies of law, like the right to private life on the one hand and data protection on the other, which has come into the spotlight as a result of the CJEU ruling. Furthermore, general political developments in a country, like the Coup in Thailand, can affect the regulations of intermediaries and lead to restrictions on free speech.

Another driver of change might be developments in the international arena. While not necessarily visible in the present case studies, standards for intermediary liability in particular might be the subject of agreements between countries in the context of bi- or multilateral trade agreements. The negotiations on the Trans-Pacific Partnership Agreement serve as a recent case in point.[53] Other important impulses at the international level might come from Human Rights frameworks (of particular importance in this context is the Declaration of Human Rights and the International Covenant on Civil and Political Rights), and also global multi-stakeholder efforts such as NETmundial.[54]

[51] Beceni, Yasin and Nilay Erdem. “Online Intermediaries Case Studies Series: Turkey (eBay Case)”, The Global Network of Internet & Society Research Centers (2015), 13.
[52] Nguyen, Thuy. “Online Intermediaries Case Studies Series: Roles and Liabilities of Online Intermediaries in Vietnam – Regulations in the Mixture of Hope and Fear”, The Global Network of Internet & Society Research Centers (2015).
[53] On the role of intermediary liability in trade, see, e.g., “Harmonizing Intermediary Immunity for Modern Trade Policy,” The Internet Association, May 5, 2014. http://Internetassociation.org/wp-content/uploads/2014/05/May-2014-Section230.pdf.
[54] See “The Manila Principles On Intermediary Liability: Version 0.9,” Organization,, December 1, 2014, https://docs.google.com/document/d/1kAkqgt3cRb65d8ik6vWYgpk6DYpP8ABA43ljgDiGOf8/edit?usp=sharing.

B. Interplay Between Constitutional Rights and Intermediary Liability

The CJEU decision highlights another relevant aspect of intermediary governance systems, which is the relationship between intermediaries’ liability and constitutional rights. The ruling has been criticized for not sufficiently taking freedom of speech and freedom of information into consideration.[55] It is not unlikely that cases triggered by the decision of the European Court will lead to lawsuits on which national constitutional Courts, as well as the European Court on Human Rights, will have to decide. At the same time, the liability regime in India has been challenged due to constitutional reasons, and the same is true for Korea. At least two aspects are noteworthy when it comes to the constitutional rights aspect of intermediary liability.

The first is that liability systems cannot trust the publisher of a web page to stand up for his or her right to freedom of speech if he/she is not informed about the take-down, if it is costly to respond, or he/she is not interested in pursuing the matter. Furthermore, the general interest in easy access to information on the Internet is not protected under freedom of information clauses because it is framed as a subjective not objective right. However, some Courts have emphasized the role of the Internet in this respect.[56] In terms of actors, there is an imbalance when there is a situation where there is a person highly interested in getting the content removed on one side and a potentially uncommitted person on the other – if any.

Secondly, the role of the operator of an intermediary is under consideration. On one hand, the operator might be enjoying freedom of speech privileges itself[57] – but the conditions under which this is the case are not easy to construe. On the other hand, the operator might be a powerful entity that decides the accessibility to a piece of information should be bound to respect freedom of speech vis-à-vis the users as well. The debate about the implication of these constitutional issues has just started.

[55] “Google Starts Removing Search Results Under Europe’s ‘Right to Be Forgotten.’” WSJ. http://www.wsj.com/articles/google-starts-removing-search-results-under-europes-right-to-be-forgotten-1403774023.
[56] E.g. the ECtHR: “In the light of its accessibility and its capacity to store and communicate vast amounts of information, the Internet plays an important role in enhancing the public’s access to news and facilitating the sharing and dissemination of information generally (accessible) (Times Newspapers Ltd v. the United Kingdom (nos. 1 and 2), nos. 3002/03 and 23676/03, § 27, ECHR 2009, and Ashby Donald and Others v. France, no. 36769/08, § 34, 10 January 2013).
[57] ”Volokh, Eugene. “First Amendment Protection for Search Engine Search Results.” The Volokh Conspiracy, May 9, 2012. http://volokh.com/wp-content/uploads/2012/05/SearchEngineFirstAmendment.pdf.

VI. Conclusion

A. Summary

A review of online intermediary governance frameworks and issues in Brazil, the European Union, India, South Korea, the United States, Thailand, Turkey, and Vietnam creates a picture full of nuance, whether looking at the genesis of intermediary frameworks, the reasons for intervention, or the specifics of the respective governance models, including strategies, institutions, modalities, and the effects of regulation, among other dimensions. The country case studies both highlight and illustrate the importance of cultural and political context, which is not only reflected in the respective legal norms aimed at regulating intermediaries, but also expressed through different views and perceptions regarding the social function of intermediaries. In some sense, the case studies and the way in which the authors tell the story themselves mirror the same context and diversity. Similarly, the importance of the socio-economic context has become clearly visible. Many of the features of various intermediary governance models can hardly be understood without considering their economic context, in conjunction with demographic characteristics and shifts.

Despite context-sensitivity, certain categories, clusters, and patterns can be distilled from the various case studies and analyzed. As suggested in this synthesis document, online intermediary frameworks can be grouped and mapped based on a number of core criteria and dimensions. Specifically, and from a conceptual angle, the synthesis shows that there are three basic groups of countries, i.e. countries that lack a specific intermediary governance framework, countries with existing and differentiated specific frameworks, and countries with emerging frameworks. The discussion also reveals patterns with respect to the key drivers and motivations for specific regulations or governance, including “bad headlines”, but also forces to be analyzed through the political economic methods. The analysis of the case studies further suggests that the governance models regulating online intermediaries are typically a case of context regulation, particularly when coming in the form of liability regimes. Against this backdrop, the analysis highlights the key role of incentives among the different actors that shape the intermediary landscape, and the interaction among them, when we seek to understand and evaluate the performance of alternative governance models or approaches.

In addition, the case studies have revealed a series of crosscutting and highly dynamic issue-specific challenges, including the problem of definition (what is an online intermediary?), the question of the different types of intermediaries, the design of notice-and-takedown systems, and the cost of compliance and enforcement, among other things. Zooming in on the role of governments, this case study analysis suggests three basic functions that governments can serve, i.e. an enabling, leveling, or constraining. With a view to the basic institutional set-up of the different governance regimes, the surveyed countries either follow a Court-based system or heavily rely on government agencies in the context of the different regulatory strategies and techniques – with lines between the two models often blurring, depending on the issues at stake. The question of incentives also plays a decisive role when it comes to the analysis of compliance and enforcement issues, including the problem of over-compliance in the case of asymmetric regulation.

B. Future Considerations

Both with respect to the conceptual and issue-specific analysis, the mapping exercise summarized in this paper is initially mostly of descriptive value and does not immediately lead to firm normative conclusions or “best practices”. That said, a more robust description of the core elements of online intermediary governance frameworks and the various forces at play can lead not only to a deeper phenomenological understanding, but also highlight some of the key considerations and issues to be taken into account when designing, implementing, or reforming governance models for online intermediaries. Such a descriptive map can and must be enriched over time by a growing body of anecdotal, and in some instances even empirical, evidence regarding the performance of varying governance models and their impact on the digital economy and society at large.[58] In that spirit, the synthesis paper and the underlying case studies seek to contribute to a stronger evidence-base that might inform debates about “best practices” regarding online intermediary governance systems by documenting some of the key feature of such regimes.[59]

With these caveats in mind, we would like to highlight the following points from the case study analysis for consideration and further deliberation in the debates about the present and future governance of online intermediaries:

1. Understand the function and economics of intermediaries. Online intermediaries are a relatively recent phenomenon, and both a driver and mirror of structural changes in the information ecosystem. Functionally, online intermediaries challenge traditional notions of what qualifies as “intermediary”: though online intermediaries are still not the source of content creation, they are increasingly involved in its dissemination, combination, etc. Consequently, much emphasis in legal and policy debates is currently on definitions and categorizations of intermediaries vis-à-vis existing laws and other norms. In addition to these definitional questions, the analysis highlights the importance of a deeper functional understanding of the roles of online intermediaries when seeking adequate regulatory frameworks. The same applies with regard to the economics of intermediaries, given the presence of strong network effects and two sided markets.

2. Emphasize the normative dimension of intermediary regulation. Recently, the interplay between intermediary liability and the digital economy has gained significant attention across jurisdictions. Even architects of systems with rather broad safe harbor regimes seem to be primarily focused on the economic benefit of lean intermediary regulation. While economic arguments are of course important in policy debates, one should equally emphasize the normative dimensions, especially the impact of different governance regimes on Human Rights. That the interest in access to information has no natural “guardian” marks a structural problem in that respect.

3. Analyze and evaluate the full range of regulatory mechanisms. The case studies show that intermediaries are regulated by different mechanisms, directly and indirectly, ex ante and ex post, through “hard” as well as “soft” obligations. Different actors follow different approaches, have different types of resources at their disposal, and show different levels of expertise. In order to analyze, assess, and improve the state of regulation and its effects, it is key to take a holistic view and consider all of these elements as well as their interplay (or lack thereof). A governance perspective is a helpful lens for such an analysis.

4. Consider the full costs of intermediary regulation. Given the complexity of the digital ecosystem, it is tempting for governments to target intermediaries. At the surface, interventions at the gateways of Internet communication seem to reduce the costs of regulation. The case studies suggest, however, that such a “window” comes with the risk of over-regulation, with a negative impact on users’ fundamental rights, as well as on innovation and the digital economy. Research also suggests the importance of taking into account less visible costs of interventions, such as the risk of empowering already powerful intermediaries by forcing them to make content related choices.

5. Strengthen mechanisms of mutual learning. Despite all the nuances, the case studies also reveal commonalities and patterns among different governance regimes. In particular, the study highlights similar challenges among countries with notice-and-takedown systems, with problems like defining the requirements for notices, whether and how to inform the owner of the effected content, regulatory oversight, etc. At least with respect to public policy-makers, the analysis suggests a great potential for transnational learning, complementing the increased sophistication of the operators of intermediaries, who tend to take a global perspective when designing their internal governance regimes.

[58] See, e.g., “Closing the Gap: Indian Online Intermediaries and a Liability System Not Yet Fit for Purpose.” Accessed December 10, 2014. p. 31-35 https://globalnetworkinitiative.org/content/closing-gap-indian-online-intermediaries-and-liability-system-not-yet-fit-purpose.
[59] In this sense also see “The Manila Principles On Intermediary Liability: Version 0.9,” December 1, 2014. https://docs.google.com/document/d/1kAkqgt3cRb65d8ik6vWYgpk6DYpP8ABA43ljgDiGOf8/edit?usp=sharing.