Categories, Balancing, and Fake News: The Jurisprudence of the European Court of Human Rights

Freedom of expression has been often described as a necessary precondition for democracy and for the implementation of an effective system of human rights. A deliberative democracy cannot function if citizens are not granted the fundamental right to express their views and to criticize the government without being censored.1 The rule of law becomes an empty notion if legal orders do not protect the impartial, autonomous judgments of the judiciary.2

every participant can ask for and offer reasons. 4 While Amar relies on a structural argument with a 'historical twist', Alexy makes a more philosophical point. Notwithstanding this difference, the central claim raised by the two leading constitutional scholars is the same: the institutional act of enacting a constitution presupposes freedom of expression.
It would be tempting to conclude that freedom of expression is the most fundamental right, not from an axiological perspective (namely, the moral ranking of human and fundamental rights), but from a purely justificatory and conceptual standpoint. 5 Is it not the case that the European Court of Human Rights (hereinafter, ECtHR) emphasized that freedom of expression constitutes "one of the essential foundations" for the existence and implementation of any other right of the Convention and "the development of every man"? 6 The main goal of this essay is to explore the methods used by the ECtHR for balancing freedom of expression against competing rights, with an eye to the current debate on fake news. Section 2 offers a survey of the normative framework of Article 10 of the European Convention of Human Rights (ECHR). Section 3 explains how the Court's method for weighing freedom of expression against competing rights is much more structured than an ad hoc balancing because it is based on categorization and sliding-scales. Section 4 deals with the challenge of fake news. Taking cues from the case law and borrowing from recent developments in philosophy, a minimal definition of fake news will be proposed as a safeguard against censorship in disguise, "collateral" over-censorship, and judicial arbitrariness. Definitions that are too broad, vague, and ambiguous are open to abuse, and this could jeopardize freedom of expression. The present inquiry will also disentangle the relevant factors for balancing web-expression and isolate  several sub-categories of fake news. The final section contains a summary of the arguments and a denouement.

Freedom of Expression in the Balance
Defining the contours of freedom of expression is not an easy task. A first major issue is the risk of conflating freedom of expression with a general right to liberty or autonomy; 7 a second puzzle concerns judicial balancing and the idea that freedom of expression is not an absolute right. 8 The ECtHR, on the one hand, endorses a broad reading of the notion of expression, and, on the other, relies on balancing qua its methodology for introducing some exceptions outside the core of this human right. The Court's approach finds textual support in Art 10 of the ECHR, which reads as follows: 1. Everyone has the right to freedom of expression. This right shall include freedom to hold opinions and to receive and impart information and ideas without interference by public authority and regardless of frontiers. This Article shall not prevent States from requiring the licensing of broadcasting, television or cinema enterprises. 2. The exercise of these freedoms, since it carries with it duties and responsibilities, may be subject to such formalities, conditions, restrictions or penalties as are prescribed by law and are necessary in a democratic society, in the interests of national security, territorial integrity or public safety, for the prevention of disorder or crime, for the protection of health or morals, for the protection of the reputation or rights of others, for preventing the disclosure of information received in confidence, or for maintaining the authority and impartiality of the judiciary.
However, the Court's interpretation of Art 10 goes beyond its plain meaning: freedom of expression is not limited to the liberty of holding, sharing, receiving information and ideas through speech acts. 9 It also comprises (i) the freedom to produce, exhibit, and distribute artworks; ii) the liberty of realizing a public artistic performance; 10 iii) the access to relevant public information; 11 iv) the liberty of displaying pictures, icons, images, 12 and religious or political symbols; 13 v) preserving cultural heritage, 14  expression; 16 vi) the protection against self-incrimination. 17 The sphere of protection granted to the freedom of the press and political speech is even broader: it includes information and ideas that "offend, shock or disturb." 18 Political speech, too, is highly protected: for instance, obscene protests involving the sculpture of a large penis, held in front of the Protector's General Office, 19 or extremist political discourse are permitted under the ECHR legal regime. 20 Furthermore, if applied to the press, the restrictions established by the Second Paragraph of Art 10 must be "construed narrowly." 21 A peculiarity of the ECHR system is the protection of the form in which ideas and information are conveyed, also when the tone is polemical and even aggressive. 22 The Court allows the possibility of restricting expression by granting priority to a Contracting State's interests (national security, public safety, and protection of health and morals) as well as to other, conflicting, human rights such as the right to privacy (Art 8 ECHR). 23 In case of conflicts between rights (or between a right and a State's interest) the Court strikes a balance, deciding which right or interest takes priority over the other, and specifying under which conditions the relation of precedence holds.
In shaping the relations of conditional priority, the Court has always relied on an evolutionary approach, based on a form of "living constitutionalism." 24 The ECtHR does not interpret Art 10 according to its original meaning because the construction Art 10 is always the result of a "dynamic update" which takes into account societal values and beliefs, both from the internal perspective (the position of the participant) and from the external perspective (looking at national systems from a broader, European, standpoint). 25

More than a Simple Ad Hoc Balancing
The Court's method for weighing principles is much more sophisticated, structured, and responsive than a sheer ad hoc, freewheeling, balancing of competing interests. 26 In fact, the jurisprudence of the ECtHR combines two different approaches: the categorical approach, which establishes what counts as freedom of expression through stipulations, and the balancing approach, which proceeds through a case-sensitive assessment of the factual elements within previously isolated categories. This combination can be understood as an attempt in rational systematization: the weighing of competing rights and interests is performed through categories, 27 and the Court gradually develops clusters of rules establishing different levels of scrutiny and burdens of justification that the arguments in favor of the challenged law or measure must satisfy. These clusters of rules are "activated" by the type of expression at hand. For instance, only a light burden of justification is usually required for measures that interfere with commercial expression, 28 whereas a higher threshold must be met for justifying a gag-order imposed upon the press. 29 In this respect, the ECtHR follows a jurisprudential pattern which is quite close to the one endorsed by the U.S. Supreme Court. The ECtHR, too, has gradually developed a full array of categories of expression and balancing tests, which operate in virtually every corner of the ECHR system. 30 On many occasions, the ECtHR has relied on proportionality as a method for resolving conflicts between freedom of expression and competing rights. A disclaimer is in order here: the term 'proportionality' has two different meanings, proportionality as a balancing test and proportionality as a (meta-)principle that underpins every operation of balancing. In the first sense, proportionality denotes a three-pronged method for resolving a conflict between rights: the first phase checks the suitability of the measure; the second phase ascertains the necessity of the measure; the final step settles the relation of conditional precedence between the colliding rights or principles. 31 In the second sense, proportionality is understood as a broad principle which prescribes that any interference with a human right must be commensurate for the legitimate objective pursued by the legislator. While the proportionality test is used only for specific categories of speech, proportionality qua meta-principle orientates every single balancing operation of the ECtHR.
The Court has relied on the three-pronged proportionality test for resolving conflicts between freedom of expression and national security, 32 or between freedom of expression and liberty rights. 33 The proportionality test is frequently employed also for conflicts between Art 10 and Art 8: it is coupled with strict scrutiny for those classes of cases requiring the highest protection. For instance, in Observer and Guardian v. United Kingdom, the Court held that "the dangers inherent in prior restraints are such that they call for the most careful scrutiny on the part of the Court. This is especially so as far as the press is concerned, for news is a perishable commodity and to delay its publication, even for a short period, may well deprive it of all its value and interest." 34 The proportionality test is ubiquitous, 35 but the most common balancing method is an altogether different kind of three-pronged test which we might label "the necessity test." 36 The three-pronged necessity test finds textual support in Art 10(2), and requires the fulfillment of three conditions for a justified interference with the freedom of expression. 37 First, the interference must be prescribed by foreseeable, precise, and publicly accessible law; 38 second, the interference shall aim to protect significant interests (territorial integrity, public safety, prevention of disorder and crime, protection of health and reputation, preventing the disclosure of information received in confidence, and maintaining the authority or impartiality of the judiciary); third, the interference must be necessary for preserving the democratic values. Determining what counts as 'necessary'which is crucial for both the proportionality test and the three-pronged necessity test-is one of the most difficult tasks that the ECtHR has to fulfill. As explained in Observer and The Guardian v. the United Kingdom, the Court considers that a measure is necessary only if a "pressing social need" exists. 39 A third test, a form of rational basis "with bite", is confined to cases involving commercial expression. 40 It is the least protective test among the ones used for Art 10. The rational basis with bite considers the appropriate measures that suitably pursue the goal(s) of the legislature, according to a means/ends analysis. The test is yoked to a high standard of deference towards the Member States, which usually grants them a wide margin of appreciation. Under extenuating conditions, a national government may even impede the publication of the exact description of a publicly relevant fact. 41 However, contracting states are not entitled to act arbitrarily. 42 They are required to pursue one of the legitimate ends mentioned in the second paragraph of Art 10, which, in turn, the Court understands as indefeasible (i.e. no implicit exception is allowed). The metaphor of the "bite" is here used precisely for grasping the resistance to implicit exceptions, which strengthens the means/ends rationality requirement. The rational basis test, too, is coupled with the duty to rule through laws meaning that every interference with freedom of expression must be grounded in legal provisions. 43 The ECtHR has also designed more detailed tests for balancing freedom of expression in three recent landmark decisions: Axel Springer v. Germany, 44 von Hannover v. Germany, 45 and Perinçek v. Switzerland. 46 These are all examples of definitional balancing: the Honorable Judges have relied on categories of expressions, a set of procedural rules orienting the deliberative process, and specific burdens of justification designed for a specific class of cases.
The judgment of Axel Springer concerns the publication of two articles revealing that Mr. X, a well-known actor who played the role of Police Superintendent Y in a television series, had been convicted of unlawful possession of drugs. The German newspaper Bild published these articles under satirical titles, accompanied by three pictures of the TV star. In order to reach a decision, the ECtHR relied on a multi-pronged test based on the following criteria: i) the general interest of the debate; ii) how well known the person involved is; iii) the subject of the report; iv) the prior standard of conduct of the person concerned; v) the method for obtaining the information and its veracity; and (vi) the consequences of publications. The case of von Hannover introduces analogous criteria for the publication of photographs: i) the degree of notoriety of the person affected; ii) the subject of the picture; iii) the prior conduct of the person involved; and (iv) the consequence caused by the publication of the photos. Finally, Perinçek-a well-known case concerning the denial of the Armenian Genocide-counts as an attempt in systematization too. The Court has analytically isolated eight relevant factors for evaluating cases involving hate speech: i) the nature of the statements; ii) the geographical, historical, and temporal context; iii) the extent to which the statement affects the competing rights; iv) the existence of a consensus; v) the possibility that the interference is prescribed by the international obligation assumed by the State; vi) the method employed by the State to justify the conviction; vii) the severity of the interference; and (viii) the final weighing of freedom of expression against the right to private life.
Coming full circle, a traditional form of strict scrutiny applies to the freedom of the press. As the Court emphasized in a leading case Jersild v. Denmark, 47 the "punishment of a journalist for assisting in the dissemination of statements made by another person in an interview would seriously hamper the contribution of the press to discussion of matters of public interest and should not be envisaged unless there are particularly strong reasons for doing so." 48 The press is not immune from liability, although this category "triggers" strict scrutiny, provided that the journalist is acting in good fate and her information is reliable. With an eye to potential conflicts with Art 8, the Court also clarifies that personal reputation prevails only when the attack to privacy is severe. 49 With regard to context-sensitivity, the jurisprudence of the ECtHR follows a sliding-scale approach, similar to the one that was advocated by Justice Thurgood Marshall of the U.S. Supreme Court. 50 The term "sliding-scale" denotes a balancing approach that associates a continuum of levels of scrutiny with different categories, which can be adjusted to the peculiarities of the case at hand. A category, in turn, is a bundle of interrelated properties and principles that define, by denoting a class of objects, entities, and actions. 51 On a sliding scale, the operative degree of scrutiny depends on a mix of factors: the nature of the classification in question, the constitutional and societal importance of the individual interest adversely affected, and the strength of the state interest asserted in support of the law under consideration. 52 In a nutshell, the ECtHR is not committed to a priori rankings between rights that consider freedom of expression as the "ace of trumps". The balancing tests endorsed by the Court, as well as the sliding-scale approach, are part and parcel of what we might call a "grand theory" for applying human rights.
As hinted above, the Court's approach involves a preliminary process of categorization. 53 The Court does not treat the ECHR as a system of absolute rights whose boundaries can be crafted with a precise, ostensive definition of which cases instantiate a particular right or an analytical definition of the semantic content of that particular right. Nevertheless, definitions map different categories of expression coupled with varying levels of protection and specific balancing tests. 54 The use of categories is not an attempt to purge balancing out of the ECHR system: definitions provide the scheme for allotting the appropriate burdens of justification and the appropriate tests. While judges and journalists are at the top of the scale, the categories of militaries and civil servants are placed right at the bottom. 55 Commercial expression, too, is with bite.
On many occasions, the Court has resorted on "content-based" categorization (namely, classifications based on the content). This type of categorization pursues two main goals: i) leaving certain forms of expression outside the protection of Art 10; ii) specifying sub-categories, in order to address them separately. The most striking example of the former strategy is Holocaust denial: a concept carefully situated outside the scope of Art 10. 56 An excellent example of the second strategy is offered by the four "maxi-categories" of speech: description of facts, value judgments, opinions, and insults. Value judgments, due to their intrinsic subjective nature, benefit extensive protection: 57 under the ECtHR's perspective, a value-judgment is nothing else than a positive or negative attitude towards a particular action, item, or state of affairs (i.e. being for/against). As for opinions and descriptions, the possibility of restrictive measures depends on the general 52. Jeffrey M Shaman, Constitutional Interpretation. Illusion and Reality (Greenwood, 2001) at 102. 53. On this specific point, the present inquiry disagrees with the analysis offered in Flauss, supra note 8 at 810. The opposition between the US tradition and the EU tradition is not as radical as it appears at first glance, and these similarities have a historical explanation. As Flauss himself notices: "The attachment to this ideological stance has grown stronger since the middle of the 1980s, notably with the presence inside the court of judges elected on account of "new European democracies." The judges, often having diplomas from American universities and sometimes even academic experience overseas, have promoted the legal precedents set by the United States Supreme Court relating to "freedom of expression" to a preferential source of inspiration" at 811. interest for the public debate: "[t]he extent of the State's recognized power of restriction structurally dependents on the contribution the speech or message makes to a general interest or public debate. The existence of a general interest debate leads ipso facto to a strengthening of European control. The same goes for political discourse." 58 Freedom of expression offers a reason for imposing waves of duties and responsibilities on those States that signed the ECHR. 59 However, categorization allows to isolate a "core" of the right, which includes the privilege to receive information and share ideas as well as the correlative duty of the State to facilitate the uptake of communication. The Member States shall not disrupt or restrict communicative exchanges without compelling reasons. 60 The right to have access to pertinent, public, information is also part of the core. 61 It should be recalled that not every kind of speech is protected under Art 10. Hate speech, 62 incitement to violence, 63 and Holocaust denial are speech acts; 64 however, they fall outside the protection established by Art 10. For instance, in Leroy v. France, 65 the Court considered that a satirical cartoon representing the attack to the World Trade Center coupled with the caption "We have all dreamt of it : : : Hamas did it!" was not worthy of protection, for it was merely a plea for terrorism. Holocaust denial, incitement to violence, and hate-speech cannot be the object of balancing, for they fall outside the scope of Art 10, even when cloaked as (pseudo)-artistic products or activities.

Fake News: A New Challenge
The ECtHR is now facing a new challenge: the dissemination of fake news and pervasive use of hate speech on social networks. 66  Shtekel v. Ukraine, the Internet can be both an instrument for increasing freedom of expression and a danger for the protection of the other human rights: "The electronic network, serving billions of users worldwide, is not and potentially will never be subject to the same regulations and control. The risk of harm posed by content and communications on the Internet to the exercise and enjoyment of human rights and freedoms, particularly the right to respect for private life, is certainly higher than that posed by the press. Therefore, the policies governing reproduction of material from the printed media and the Internet may differ. The latter undeniably have to be adjusted according to the technology's specific features in order to secure the protection and promotion of the rights and freedoms concerned." 67 Misinformation and propaganda are as old as time, but the impact that fake news had on central juridico-political events such as the U.S. Presidential elections and Brexit has raised the level of public awareness on this issue. 68 Most of our knowledge does not come from direct observation of facts because we tend to rely on reports and testimony. Therefore, the actual structure of the web represents a real risk: bad quality information overwhelms the global audience. Social networksthe "modern public squares" 69 -are often misunderstood as a reliable source of knowledge, even when they express unjustified or false beliefs. 70 The dissemination of fake news has two main reasons: the advertising revenue connected to the number of people accessing the information and the ideological and political strategy of disinformation. 71 The dangers of fake news are not limited to libel and hate speech: 72 information overload can disincentivize "quality journalism" and, paradoxically, this phenomenon might end up by frustrating the right of the citizens to be adequately informed. 73 By correctly informing the audience, news media ensure epistemic coverage to individuals. In a democratic republic, news media must provide a net of reliable beliefs to the audience. 74 In the next few years, the ECtHR will play a fundamental role in supervising the filtering and take-down practices prescribed by the Contracting States, hopefully preventing gross violations of Art 10 by striking down illegitimate measures. 75 To be sure, one might contend that fake news poses a pseudo-problem. The Internet is the new "marketplace of ideas" and, therefore, the Court should not impose restrictions on freedom of expression. However, this argument à la Holmes is affected by a major flaw. 76 The ECtHR acts in a legal order which is already regulated by the Contracting States, in order to check, whether the national entities fulfill their obligations. Furthermore, the media market of platform providers resembles an oligopoly more than a highly competitive framework, despite the measures aimed at weakening of the barriers to entry. 77 As the recent comparative inquiry of the Council of Europe shows, most European States have promulgated targeted regulations prescribing the removal from webpages and social networks of illegal content concerning child abuse, terrorism, hate crimes, intellectual property, and copyright. The primary trend in Europe is to impose a procedure of 'notice and take-down' whereby the contentious content shall be removed by the host after publication in response to allegations of illegality. However, most legal systems also require Internet hosts, social networks, and on-line service providers to act ex ante, by filtering in the phase between upload and publication. 78 The legal issues related to these practices are incredibly delicate. Sharing a content on the Internet is always the result of a cooperation between several actors: a) the agent who posts the content, b) the host who provides a domain-based platform (e.g. Facebook, Twitter, Word-Press), c) the audience, and (d) whoever facilitates or checks this process (e.g. ISP providers). 79 Web-expression is radically different from ordinary communicative exchanges. 80 The audience plays an active role in the uptake of the content, and in responding with likes and further comments. In principle, both the agent who writes a post or comment and the host are accountable, although, according to the European Commerce Directive and the Audio-visual Media Services Directive, web operators benefit of limited accountability. 81 Many hosts and intermediaries act as editors, making a profit out of what users share on their platforms, but, strictly speaking, they lack the legal status of a professional editor. In this case, the liabilities are not clearly and consistently established by the Member States. On that note, in the recent case of Rebechenko v. Russia, 82 the ECtHR proclaimed that bloggers, too, exercise the function of "public watchdogs." However, any enthusiasm for the analogies between journalists and bloggers as well as editors and social networks should be tempered. In Kate Klonick's words, "a central piece is missing in the comparison to an editorial desk: platforms do not actively solicit specific types of content, unlike how an editorial desk might solicit reporting or journalistic coverage. Instead, users use the site to post or share content independently." 83 Grave concerns about the risk of censorship in disguise and "collateral overcensorship" have already emerged in Europe, where the Member States have started to issue special regulations against fake news. 84 Germany has adopted the Network Enforcement Act in 2017 which requires social media and platform providers to swiftly remove fake news, hate-inciting content, and any "obviously illegal" content (an excessively vague and ambiguous term indeed), at the latest within twenty-four hours of being notified, imposing fines up to 50 million Euros. 85 In France, the so-called "law against the manipulation of information", which introduced three new articles (L. 112, L. 163-1 et L. 163-2) in the Electoral Code (Code Électoral), was challenged in the national courts. These legal provisions prohibit the dissemination of fake news about the political process during the three months preceding elections and prescribe sanctions for providers publishing false information that might influence the voter's preferences. It comes as no surprise that the Conseil Constitutionnel upheld the French network enforcement act, considering that the measure was not disproportionate since it was limited to the elections and concerned only information of public interest. 86 The European Union, too, has begun to wage war against fake news, by promoting the EU Code of Practice on Online Disinformation in 2018. The EU Code, on the one hand, proposes AI aided systems for preventing disinformation and for tracking and removing automatically illegal content from online platforms. On the other hand, the code introduces a number of measures-such as the external review of an independent third party-aimed at preventing potential side effects such as the inhibition of bona fide research and the possibility of political manipulation.
On 26th April 2018, the European Commission adopted a Communication on "tackling online disinformation" 87 which proposes guidelines for preventing fake news and disinformation through an intricated system of strategic measures: a) the use of new technologies, such as blockchains and AI systems, for tracking the origin of the information; b) the introduction of an independent European network for checking the quality of the information; c) several measures aimed at fostering education and "media literacy"; 88 d) cyber-security safeguards for the elections; and (e) a plan of instructing journalists and public media on how they should act in a digital environment.
There is an actual risk that, under specific interpretations, the abovementioned legislative interventions might offer a pretext to national governments and EU institutions for over-censorship and ideological manipulation. For instance, improving the "media literacy" of journalists, on the paper, might sound like a splendid idea, but it could quickly degenerate into forms of libertarian paternalism aimed at brainwashing the fourth estate. True enough, the risk that fake news might harm other fundamental liberties, such as privacy, is considerable, especially considering the powerful effect of multi-media contents, the degree of anonymity of the users, 89 and the persistence of the information on the web. 90 However, the battle against fake news could result in disproportional censorship and political interference if the notion of fake news is defined by legislators in vague and broad terms such as 'unreliable information that can represent a threat to life, public order, public safety, health, and vital elements of social and transportation structure', as the Russian law notably does. 91 If the definition of "fake news" is too broad, vague, and ambiguous then, in principle, any content that does not meet with the governmental approval could be legitimately removed, and the system of sanctions would become unpredictable. It behooves the ECtHR, as an independent judiciary, to have the last words on these fundamental issues concerning human rights; however, we might wonder whether the traditional approach based on categorization and sliding-scales is still suitable for dealing with web-expression and fake news. Have we already reached such radical change in our ways of expressing ourselves, as "digital speakers" that requires the abandonment of the more traditional categories for balancing Art. 10? Is it still possible to rely on "fixed" stipulations and categories? Wouldn't it be better to decide intuitively, case-by-case, without the ambition of drawing general lines? Should "fake news" be considered as an autonomous category of expression, and at what level of the scale should it be placed? The current analysis will propose arguments in favor of the traditional categorization-based approach. I will suggest that a minimal definition of "fake news" acts as a shield against excesses of the censorship machine. Furthermore, several subspecies of fake-news and several standards for balancing web-expression will be disentangled.
There are good reasons to think that the Court might still rely on categorization, for this seems to be the solution preferred also by EU Member States. National regulations tend to split web-expression into various sub-categories (e.g. child pornography, illegal online gambling, intellectual property, and personal data protection) 92 ascribing different levels of legal protection to each of them. Hence, the ECtHR might decide to associate systematically every sub-category of "web-expression" and "fake news" with specific balancing tests. The most demanding tests and, accordingly, the highest level of protection (strict scrutiny) might be granted to online professional newspapers, whereas a lower level of protection and more lenient tests (e.g. intermediate scrutiny) might be associated to purely commercial speech. Then, with systematic effort, every category of "web-expression" might be placed at different levels of the scale. There are at least three recent decisions of the ECtHR that move in this direction.
In Pirate Bay, the Court upheld a Swedish ban on specific peer-to-peer downloading systems-viz sharing "torrent" files-that violated copyright protection according to the intermediate scrutiny. The ban was deemed consistent with the legitimate aim pursued by the legislator. 93 The intermediate scrutiny was "activated" by the specific category of 'informational piracy,' which, under its manifestly illegal nature, precludes heightened levels of scrutiny. Furthermore, the presence of an economic interest in spreading digital contents played a fundamental role in "tuning" the Court's assessment toward the lowest levels of the scale.
The ECtHR has resorted on the traditional approach (namely categorical balancing) also in the recent case of Cengiz and Others v. Turkey. 94 By applying a rational basis with bite test, the Court expressly held that the Turkish government's restrictions on YouTube broadcasting were aimed at affecting the citizens' right to receive information. First, the restrictive measure was not grounded on an enacted statute. What is more, the means/end rationality requirement was afflicted with over-inclusiveness. 95 National authorities are not allowed to block an entire website if only the content of a few hosted pages is illegal therefore, the restrictive measure was too broad compared to its underlying reason. The Court also considered that, strictly speaking, the case at hand was not an instance of purely commercial speech. In that specific context, YouTube was acting as a public intermediary and the disputed content concerned the political figure of General Atatürk, the first President of Turkey (1923-38). Therefore, a heightened burden of justification was appropriate.
Finally, in the case of Magyar Jeti Zrt v. Hungary 96 the ECtHR considered that the dissemination of defamatory content cannot be equated to the mere posting of a hyperlink that opens a defamatory content; in cases involving hyperlinks, any restrictive measure shall satisfy the three-pronged necessity test. 97 Furthermore, the individual assessment on a hyperlink case must ponder eight relevant factors: "i) did the journalist endorse the impugned content; ii) did the journalist repeat the impugned content (without endorsing it); iii) did the journalist merely include a hyperlink to the impugned content (without endorsing or repeating it); iv) did the journalist know or could the journalist reasonably have known that the impugned content was defamatory or otherwise unlawful; v) did the journalist act in good faith, respect the ethics of journalism and perform the due diligence expected in responsible journalism?" 98 Now, categorial balancing would be a promising path for dealing with the expression on the web and fake news only if integrated with a comprehensive theoretical development that identifies new sub-categories of web-expression and a minimal definition of the concept "fake news", a philosophical puzzle that the ECtHR is called to resolve.
It is quite challenging to stipulate a foundational, precise, and neutral working definition of the compound term "fake news". 99 However, such a definition is necessary for both preventing the risk of censorship in disguise, and for a reasoned elaboration of the ECHR principles. Without a neutral definition of fake news, any balancing operation on this matter would be like drawing lots, 100 and the crusade against the spread of bad quality information on the web might gradually lead to abuses. "Fake news" cannot become a "catch-all term" 101 which empowers biased censors to act as the ultimate arbiters of truth. 102 As plausibly suggested by the philosophers Romy Jaster and David Lanius, the best solution for a stipulative approach is to start with a minimal two-fold notion of "news" as "newly received or noteworthy information, especially about recent events" and, in a more specific usage, as a "broadcast or published report of news." 103 Then, it could be added that the qualification "fake" means that the news is alternatively: a) false; b) true but highly misleading; c) made without any intention of seeking the truth. The presence of the third element c) is fundamental for it captures the idea that fake news is "misleading by design". 104 A piece of fake news is intended to have a broad, increasing circulation, deceive a targeted audience, and to exploit prejudice and biases of the audience (e.g. biases about some social issues such as the impact of immigration) for reaching lucrative or political goals (e.g. influencing an election). 105 This definition has the virtue of separating fake news from editorial errors gone viral. Fake news comprises either outright false reports (e.g. the news that Obama "will refuse to leave his office if Trump is elected" shared by 98. Magyar, supra note 96 at para 77. 99. Nabiha Syed, "Real Talk About Fake News: Towards a Better Theory for Platform Governance" (2017) The Yale LJ 337; Claire Wardle, "Fake News. It's Complicated" (2017), working Paper available at: medium.com [perma.cc/FFE3-WKVQ]. 100. See Gelfert, supra note 74 at 85: "if one holds that the term refers to a distinct and identifiable phenomenon, it is all the more important to attempt to come to conceptual grips with it and, if necessary, put forward a definition that can then be refined in the course of future scholarly debate." 101. McGonagle, supra note 84 at 203-09. 102. Tackling online disinformation, supra note 67 at para 3.1.2 alludes at the opportunity of reliable "independent fact-checkers", whose task is to analyze the credibility of the content, as well as the correctness of the creation and dissemination process. Burrardstreetjournal.com on Sept 7, 2016) 106 or misleading information (for instance, the report that, after Hurricane Dorian, bricks of cocaine washed up on Florida's beaches was misleading because only one single brick was found). 107 In both cases, the "fake" report is aimed at deceiving the audience with an utter disregard of the truth. 108 In this respect, fake news is radically different from journalistic errors gone viral. Viral journalistic errors, although false, concern issues of general interest and have been obtained through bona fide accurate research. 109 The journalist genuinely believes in the truth of what she says (i.e. the speaker is truthful, although the content is false), and her conclusion is supported by evidence. 110 Genuine journalistic errors gone viral deserve special protection, for excessive constraints imposed on professional journalists, acting bona fide, might frustrate their fundamental role as watchdogs of the democracy, 111 as well as their ability to debunk fake news. The Stoll v. Switzerland 112 case is an indication that the ECtHR is perfectly aware of this risk.
According to the minimal definition, not every piece of fake news is straightforwardly false and, conversely, not every false report is "fake news". Sensational and purely fictional stories, albeit false, do not count as fake news if they are presented as unreal contents, shared as a pure source of entertainment. 113 Websites that deliberately and unabashedly offer purely fictional stories are not spreading, strictly speaking, fake news. With an eye to the responsibility of the platform providers and social networks, the flagging strategies employed deserve close attention. These intermediaries might flag inaccurate content either as "disputed" or as "false," and the choice between these two options has a substantial impact on the audience. 114 In this connection, satire deserves close attention. Since the times of the Greek comic playwriter Aristophanes, satirical expression plays a fundamental role in encouraging the political debate. This historical evidence shows that purely fictional political content might become harmful only when it is misleadingly used to fool the audience. The fictional character of the content shall be clearly stated in the text. For instance, this type of warning was intentionally omitted when the webpage wtoenews5.com reported the fake news that Pope Benedict supported Donald Trump for the presidential elections. 115 By contrast, the news that Rep. Ilhan Omar walked out of a 9/11 Memorial Service, published by Bustatroll.org, was unequivocally portrayed as a piece of satire by the editors. 116 Ultimately, under the proposed account, genuine satire falls outside the scope of the term "fake news".
The minimal definition of "fake news" includes misleading statements which, while being literally true, can be associated with implicit false content and, therefore, misrepresent some facts (what we might dub "distorted news"). 117 Misleading statements are spurious, although not, strictly speaking, counterfeit. An instance of a misleading statement is the use of hyper-partisan interpretations of facts or the malicious selection of facts. These kinds of statements might affect significantly the right to be correctly informed.
Lastly, according to the proposed definition, the notion of fake news includes those statements made without any concern for the truth, where the content is made up by the speaker only to shape the beliefs of her audience. Let us imagine, for instance, a journalist that, in an acclaimed online newspaper, describes NGO's vessels involved in the operation of rescuing immigrants at the high seas as "seataxis." This misleading description would be fabricated without any intention to seek the truth, for the sole purpose of eliciting a negative bias towards NGO's rescue operations and, plausibly, also towards immigration in general.
To close the square, we can say that fake news, as a broad category, includes lies (i.e. false statements made with the intention of deceiving), true but (intentionally) misleading statements (recall the "sea-taxis"), and statements made without any concern for the truth. The glue that keeps these notions together is the lack of truthfulness of the speaker.
This minimal definition will pay good dividends in several ways. On the one hand, it "saves" earnest and accidental mistakes but, on the other, it does not restrict the notion of fake news to "completely fabricated stories" and "contents that are 100% false", 118 a stipulation that seems too narrow. Furthermore, it maps widespread intuitions about fake news, such as the presence of malicious intent, the lack of accuracy, the absence of factual basis, and the idea that a piece of fake news is purported as true. In this way, the minimal definition rules out several categories that are not harmful, such as satire, (subjective) value-judgments, and urban legends. In fact, value-judgments and personal opinions-which, according to the jurisprudence of ECtHR are protected by Art 10-fall outside the minimal definition of "fake news" advocated by the current analysis. Why should we consider the subjective opinions or value-judgments of a blogger as fake news, if they are not purported as a truth (or truthful) description of facts? Finally, the minimal definition here proposed is neutral since it is not charged with moral and political notions. Now, a minimal definition of "fake news" might be a starting point for preventing over-censorship, but it is not sufficient. As mentioned above, the categorial balancing and the sliding-scales approach require a more fine-grained characterization through precise sub-categories that provide a "scalar framework" for the Court's balancing operations. 119 The peculiarities of the Internet as a virtual space require an altogether new sub-categorization. A perspicuous analysis of fake news should be informed by the mediating digital and cultural infrastructure that facilitates the online circulation. 120 First, it can be useful to distinguish different types of websites, based on their primary contents-arts, business, health, recreation, and politics, to name just a few-because it seems reasonable that the standards for assessing bad quality information and its potential harmful effects vary according to this parameter. 121 The structure of the multi-media service shall also be taken into account. On this point, the Visual Media Service Directive proposes a legal regime that distinguishes between "linear services"-namely, scheduled audio-visual services (e.g. BBC) or on-demand services (e.g. Netflix)-and "user-generated social media" (e.g. Twitter). 122 Only the former is required to assume editorial responsibility. The ECtHR might wish to endorse these classifications. More broadly, considerations about the corporate identity (namely, the values endorsed by the company) may figure in the Court's reasoning. Second, balancing operations should look attentively at the importance of the website and separate highly ranked websites from small websites. 123 Generally speaking, a piece of fake news spread by a major website has a stronger impact than a piece of fake news divulgated by a small webpage, although the end-to-end infrastructure of the Internet and the audience tendency of submitting to purported authorities might defeat this generalization under specific circumstances. More broadly, the impact factor of a piece of fake news (which depends on the aggregated quantity of likes, shares, and comments) shall be measured. News coming from top-ranked journals (The Sunday Times, BBC, Daily Mirror, Le Monde) and fact-checking websites (such as "FakeCheck.org", "Snopes.com", and "mercola.com") shall be treated separately, both for their positive debunking capacity and for the detrimental potential drawbacks of their possible mistakes. Top-ranked journals, in particular, are often considered as authoritative sources by the audience and, therefore, granted higher credibility.
Finally, the peer-to-peer distribution and, in general, the reaction of the audience should be assessed through consequential reasoning. If we consider the peer-to-peer distribution, it is hard to see how false information, tweets, and comments shared by troll farms, troll brigades, and bots might count as protected expression. 124 "Troll farms" are organized teams of users paid to spread fake news and inflammatory, offensive, provocative, or senseless content on social media and platforms, in order to create conflict and disruption. "Troll brigades", by contrast, are State-sponsored anonymous Internet political commentators that rely on sock-puppets, large-scale orchestrated disinformation, propaganda campaigns, and hate speech, to exercise influence on the political and social life of other states. 125 "Bots" are automated advertising engines that commonly spread spam, sensational content, and inflammatory speech (e.g. the news that the Pope supports Trump). As the Court clarified in Cengiz and Others v. Turkey, Art 10 ECHR protects the user-generated expressive activity, 126 and not bot-or troll-generated expressed activity.
To be sure, the ECtHR has grappled with web-expression in a few other recent cases, including the landmark decisions Delfi v. Estonia and MTE Index v. Hungary. 127 However, the Court's reasoning mainly addresses hate speech and defamatory contents, without meeting the challenge of fake news head-on. In Delfi, the Grand Chamber held with a striking majority (15-2) that companies are liable for the late removal of unlawful, offensive, vulgar, and violent content from their website. The Court emphasized the commercial nature of the portal and, correspondingly, the economic interest of the company in the posting of comments. Delfi was not acting merely as a disinterested service provider, for it allowed false, offensive, comments that were not protected by freedom of expression in order to gain economic benefits. In this way, the Court substantially equated the company's liability to the liability of a journalistic publisher, in consideration of the fact that the company was operating de facto as a discloser of information. The ECtHR applied the same rationale MTE Index v. Hungary, which, in this case, led to the opposite solution because the relevant facts were radically different. MTE was a non-profit service provider therefore, it was not economically interested in the number of comments. Furthermore, the negative comments that were published on the MTE's platform did not amount to hate speech. 128 These are important and most welcome achievements. The presence of an economic interest is a marker for 'commercial expression', which attracts low levels of protection and less demanding tests. However, this solution cannot be hailed as a turning point in ECtHR jurisprudence. As adumbrated above, in future analogous cases the Court might wish to consider a more extensive set of variables directly related to 'fake news' in order to move beyond the decisional strategy based solely on the dichotomy commercial/noncommercial and the possible presence of hate-speech. 129 This seems quite trivial if we consider that the vast majority of on-line service providers are economically motivated entities. Furthermore, the social costs of bad quality information (less accurate and mistaken beliefs, skepticism toward news producers, and unreliable information regarding the political process) should be balanced against: i) the interest of the news consumers in enjoying the reading of fake news, as a sheer form of pragmatic markers yield implicit contents. 137 Eventually, politically biased algorithms can "disproportionally favor people with power over individual users" and shrink the right to equal access and participation. 138 For all these reasons, the structure of the filtering system shall also figure among the factors for balancing fake news and, more broadly, web-expression.
If the notion of intention plays a fundamental role in defining "fake news", as I have suggested, the ECtHR might enter deep water. the possibility of relying extensively, for the first time, on "intention-based tests", namely, complex judicial standards for evaluating the intentions undergirding expressed content in order to ascertain the presence of actual malice. What is more, the Court might also rely on an "epistemic entitlement test", which checks whether the content of the expression is epistemically justified in terms of transparency of the information, trustworthiness of the source, and soundness of the inferential patterns that led to its creation. This solution would really amount to a radical change in the ECtHR jurisprudence, for human rights courts generally hesitate to focus on the motives of the speakers, because motives are difficult to prove. 139 Now, one question remains open: could we trace back the notion of fake news to the more general, old, and broader category, of false news? False news, propaganda, and other forms of weaponization of information are not at all new. One might recall Octavian's propaganda campaign against Antony based on the dissemination of slogans imprinted on coins (1st Century BC), or the War of the Worlds radio drama (1938), narrated by Orson Wells, which tricked the listeners into believing that a Martian invasion was actually taking place. 140 The notion of false news is not unprecedented in the jurisprudence of the ECtHR. For instance, in her insightful essay, Irina Katsirea mentions Raëlien Suisse among the prominent cases of non-digital dissemination of fake news. In that case, the ECtHR was dealing with a poster campaign aimed at convincing people about the existence of Aliens. 141 However, there are several properties that enable distinguishing between fake news and false news tout court: a) the fact that "new technologies can be used, notably through social media, to disseminate disinformation on a scale and with speed and precision of targeting that is unprecedented which create personalized information spheres and powerful echo chambers for disinformation campaigns"; 142 b) the radical change of the European social context, which is now facing a massive crisis coming from "economic insecurity, rising extremism, and cultural shifts generate anxiety," and a general condition of "societal tensions, polarization, and distrust;" 143 c) the data that signals how traditional tools for fact-checking are not effective with web-expression; 144 and (d) the fundamental role that digital intermediaries (Facebook, Google, Twitter) have come to play in influencing the news media, by offering platforms where anyone can act as a publisher, and readers themselves-frequently "tricked" by clickbait titles-are invited to add comments and links to further contents. 145

Taking Stock
This essay has explored the tests for balancing used by the ECtHR in matters of freedom of expression and proposed new categories for dealing with fake-news. It was suggested that the Court has usually relied on categorization and slidingscales. The peculiarities of web-expression and the dangers connected to the enormous amount of bad quality information pose a new challenge. However, the legislative measures endorsed by the Member States and European institutions to counter "fake news" and disinformation on the Internet, while increasing "media literacy," might eventually yield illiberal consequences.
With an eye to the future, this essay argued that a minimal definition of "fake news", which incorporates the notion of intention, is fundamental for avoiding censorship in disguise and providing a conceptual framework for "constrained" balancing operations. Furthermore, the theoretical analysis has isolated a set of parameters that might be considered for weighing Art 10 in matters of web-expression. Finally, this inquiry has briefly addressed the opportunity of distinguishing fake news from traditional forms of false news since propaganda and misinformation are not new. However, new media, video-sharing, and bloggers are more powerful and effective means for the transmission of ideological contents.