Approached to big data and disinformation strategies in Italy: Case study on the  Telecommunications Regulator (AGCOM)

The current regulatory framework

The existing national (i.e. Italian) and EU legal frameworks for electronic communications and media markets have origin in the analogue era, where regulators’ competences and powers were designed to govern completely different markets, business models and transactions. At EU level there has been a progressive adaptation to the digital world, by the recently revised EECC2 and AVMS3 Directives, however this review process has been slow, partial and approached by incremental and marginal changes to the old regulatory framework. This approach unavoidably implied problematic path dependency phenomena and the difficult trade-offs between:

• the risk of an unbalanced public intervention, based on an extensive enforcement of the old regulatory model, hampering innovation

• the risk of a lack of intervention, not addressing new market failure issues, not promoting a competitive level playing field among the different layers and/or nodes of the digital ecosystem and not properly protecting consumers’ digital rights.

Moreover, the traditional regulatory approach is based on market-specific rationales, whereas today’s markets are subject to a “new convergence”, following the first one between telco and media, which AGCOM, and other regulators, were designed to tackle. This new convergence is driven by digitalisation and the development of data economy: all different parts of the digital ecosystem are so closely interrelated and make it difficult to identify well-defined relevant markets, whereas the main players are (few) large global companies (e.g., Amazon, Facebook, Google, Apple, Microsoft) characterised by a high degree of vertical and horizontal integration in many segments of the ecosystem, working together with a myriad of small specialised companies.

Finally, current regulatory policies are designed and implemented at a national level (or at continental level), whereas the digital and data economy, the digital platforms and their end-users have global features and attitudes. This unavoidably implies a reduced regulatory capacity, misalignment of incentives and cross-country externalities among the different national public bodies involved. These mismatches are affecting the ability of national public bodies to effectively govern the new markets and new social interactions. This ineffectiveness is also due to an augmented asymmetry of information in the digital world: it is indeed very difficult for public bodies to “push” digital actors to reveal relevant information related both to 1) their internal technical functioning (they are almost “black boxes” from regulators’ perspectives); and 2) the global digital market(s) dynamics. Regulators very often do not have tools to fill this deep informational gap.

Digital platforms’ business models and economics

The main disruptive aspect of on-line platform economics concerns the establishment of a business model based on global digital transactions, where data is implicitly exchanged for cheap or free services in a multi-sided market context. The other relevant (and potentially disruptive) economic feature of digital platforms is that they can drastically reduce consumers’ transaction and search costs, allowing reaching cost-efficient outcomes similar to a perfect competition context. Yet, there is a huge difference. Within digital markets information at the base of such productive efficiency is not a public good, i.e. perfectly observable by everyone, but (de facto) internalised and exploited by (vertically and horizontally) integrated platforms. Indeed, the larger platforms are (getting closer to have all possible subscribers and embracing more and more traditional markets) the more efficient their algorithms and their services are. Larger and larger sets of customers’ data give platforms the ability to identify preferences and willingness to pay of each customer and thus perfectly discriminate. This also affects the users’ incentives to increasingly “trade” data in the implicit exchange for “free” services and to stay inert as consumers, therefore not sanctioning service-providers (and making data portability provisions useless).

What is positive under a productive-efficiency perspective is leading consumers to make their choices in smaller markets, finally tailored on each consumer This phenomenon, jointly with indirect and direct network externalities, make users locked-in an “information digital aftermarket” by gatekeepers which looks like “the market”, by selecting each user’s services, products and influencing their preferences. This outcome may not well be desirable from a competitive and social point of view, particularly when we consider the information system and informative contents that are able to influence personal and public opinion.

In fact, in the digital ecosystem there is a larger availability of informative contents on the supply side. However, the reduction of transaction and search costs also decreases time that consumers use to information seeking. Digital platforms thus solve the digital information overload and efficiently select relevant information by perfectly discriminating and matching consumers (current) preferences on informative contents. Nevertheless, this unavoidably undermines pluralism, because consumers receive their own world representation, whereas any informative content they might currently disagree with is not selected or offered.

This process has been observed as leading to pathological information phenomena such as confirmation bias, echo chambers and citizen polarisation (i.e. the tendency to acquire mainly information consistent with their ideological preferences) and to disinformation phenomena (such as fake news).

AGCOM’s vision on a new policy paradigm for big data

Building on its market inquiry about “Internet services and on-line advertising” (published in 2014),4 AGCOM has started a big data market inquiry (which is part of a joint assessment carried out with the competition and data protection authorities) and in September 2018 published its Interim Report.5

AGCOM strongly believes that, because of structural and lasting market failures in digital data markets, it is necessary to adopt an ex ante approach to the data regulation (and to possible regulation of related algorithms), while competition law enforcement and data protection are not sufficient. Moreover, since the digital ecosystem is very complex, a holistic/horizontal regulatory approach to its networks and services is necessary: an approach able to look at and address system interdependences, so neither risking to a negative impact on innovation nor to provide an incomplete or merely formal protection of consumers’ and citizens’ rights. In other words, it is necessary to build a horizontal regulatory framework for digital economy to protect and empower consumers and citizens in all digital ecosystem’s markets and segments, where consumers’ information and data are crucial.

Within data markets, regulators have a pervasive and structural information asymmetry vis-à-vis operators. As a first step of the new regulatory paradigm, therefore it is necessary to open up the “black box” and look at the complex and composite digital data transaction. Therefore, regulators need to be provided with relevant inspections powers in order to ascertain and assess, among others, timing and methods of data acquisition (data gathering and storage), functioning of the algorithms (algorithm accountability), methods of data conservation and analysis (data analytics), derived information, and deriving (primary and secondary) uses.

Second point is that big data dynamics, jointly with the Internet of Things, Artificial Intelligence and 5G, make it necessary to overcome the traditional distinction between different types of data (personal, non-personal, sensitive, etc.), whereas the new approach must refer simply to the data “per se”.

Third, regulators must be given mandate, competences and powers to address or minimise risk of new market failures (such as implicit transactions, incomplete markets, information asymmetries, hold-up and locked-in phenomena) and to disentangle the trade-off between platform efficiency and the informational aftermarkets they create. To do that it is necessary (yet not sufficient) to look at data as an economic good, so focusing on consumers’ incentives to engage in transactions and what are the economic effects of those transactions. In this way, it is possible to make the digital data transaction explicit and empower consumers to exercise their (existing) property rights and thus exerting a decentralised discipline on platforms’ market and contractual power. Indeed, a key policy issue in the data economy is the definition of entitlements on data property rights, coherent with data protection rules, in order to enable owners to explicitly exchange them in the data market or exclude any other party from accessing or using them – even from “internal use” as profitably done by vertically and horizontally-integrated multi-sided platforms.

This approach would also tackle the existing tension between objectives of data protection and pro-competitive rules, and it must be based on a strict co-ordination/integration of those rules and the implementing bodies. Protecting the digital data privacy but neglecting the implicit exchange of that data in an incomplete (and unregulated) market risk to jeopardise market dynamics, giving platforms incentives to vertically and horizontally integrate in order to internalise the most sides of the data market, and ultimately hamper the final user-consumer. As mentioned, it is crucial to look and regulate data as an economic good, and that is exactly the approach that GDPR, with data portability provisions, and other EU data policies are starting to build.

AGCOM’s approach to on-line pluralism (“pluralism 2.0”), fake news and disinformation

Digital platforms today play an increasingly pivotal role in the information system, which has been structurally changed by big data. This is due mainly to the capacity of digital platforms to gather personal information and extract value from data by means of accurate profiling (“data analytics”), which makes these actors the world leaders in the (on-line) advertising sector – a resource that is still the main source of funding for online and off-line information. Moreover, search engines and social networks play an increasingly crucial role as 1) direct information tools, because a large and growing portion of people use social networks to keep up with the news; and 2) distribution channel for online news, provided that they operate as gatekeepers for access and distribution of online contents. Big data is crucial also under a pluralism point of view, as it is at basis of platforms algorithms functions, and allow personalisation of information generation and distribution, and thus disinformation, polarisation and other pathological information phenomena. Within social networks, automatic customisation systems (operating by means of algorithms and big data) and the sharing user-generated informative contents support the viral widespread of polarising contents and the proliferation of fake news.

Online information distortions represent a complex and multifaceted phenomenon, varying in terms of actors involved, underlying motivations, communication techniques, tools and technologies used, and resources invested. To understand this phenomenon, AGCOM has conducted market research on “News consumption” (published in February 2018)6 and is carrying out a market inquiry on “On-line disinformation strategies and the fake content supply chain” (interim report published in November 2018).7

The main outcomes are the following. First, online disinformation distortions operate on both the supply and the demand side of the information system. Secondly, the fake-news production, distribution and dissemination processes hinge on a full-fledged supply chain of fake contents, which may be structured according to promoters’ motivations and the kind of audience targeted by the disinformation strategy. Thirdly, it is possible to detect on-line disinformation strategies by exploring each specific case of online disinformation strategy and its supply chain, and AGCOM is doing that by its observatory on online disinformation (see below). Those strategies are characterised by a certain standard number and types of disinformation actions, including an array of publishing and re-publishing fake contents.

Against this variety of online disinformation problems, AGCOM believes that both self-regulatory and (co) regulatory measures should be adopted. With this aim, on November 16, 2017, with its resolution n.423/17/CONS, AGCOM launched a co-regulation “technical committee (‘tavolo tecnico’)” on pluralism and fair information on digital platforms”,8 comprising AGCOM and most of the market stakeholders (e.g. social networks –Facebook–, search engines –Google–, press editors and audio-visual service providers). AGCOM’s technical committee aims to create consensus, identify and promote self- and co-regulatory solutions to online disinformation problems. The committee is co-ordinated by AGCOM’s directors of the Economic Analysis and AVMS Directorates.

The “technical committee” comprises five working groups, dealing with:

• methodologies for classifying and detecting online disinformation phenomena;

• definition of monitoring systems for economic advertising flows aimed at financing fake content;

• fact-checking: organisation, techniques, tools and effects;

• media and digital literacy;

• design and implementation of information campaigns about disinformation aimed at consumers.

One of most important achievements of the technical committee was the approval of the “Guidelines for equal access to online platforms during the election campaign for the political elections”, which identified a few goals in view of the last national political elections, that the stakeholders involved committed to comply with:

• Equity of access to all political subjects to the information and political communication tools offered by digital platforms

• Transparency of political advertisements

• Ban of illegal content (i.d. defamatory content against candidates or the circulation of opinion polls in the last15 days before the elections)

• Improvement of the fact-checking activities

In 2019, a new set of Guidelines was approved, and it contained a more detailed set of commitments, drafted on the basis of the commitments established by the EU Code of Practice against disinformation.

The pluralism committee, and the related co-regulation approach, is another way to open up the “black box”, by means of analysis and surveys based on information provided by online platforms. With this approach, AGCOM’s is looking at the techniques of newsfeed and algorithmic selection, editorial framing and organisation, stakeholders’ transparency policy for users.

AGCOM’s pluralism committee is working to establish operative solutions, among which there are:

• Constant monitoring of online disinformation and hate speech phenomena, by establishing the observatory on online disinformation: the observatory first report was published in March 2019.9

• Establishing a co-ordination platform for autonomous fact-checking activities and setting journalistic standards in terms of transparency, ethics and quality of the editorial structure.

• Actions for media literacy and against online hate speech phenomena: a draft regulation on hate speech was published in 2019.10

• Definition and implementation of direct access procedures – by a third independent party – to profiled data, in order to assess the impact of self and co-regulatory measures.