• Contenu principal
  • Menu
OpenEdition Books
  • Accueil
  • Catalogue de 16370 livres
  • Éditeurs
  • Auteurs
  • Facebook
  • X
  • Partager
    • Facebook

    • X

    • Accueil
    • Catalogue de 16370 livres
    • Éditeurs
    • Auteurs
  • Ressources numériques en sciences humaines et sociales

    • OpenEdition
  • Nos plateformes

    • OpenEdition Books
    • OpenEdition Journals
    • Hypothèses
    • Calenda
  • Bibliothèques

    • OpenEdition Freemium
  • Suivez-nous

  • Lettre d’information
OpenEdition Search

Redirection vers OpenEdition Search.

À quel endroit ?
  • Presses des Mines
  • ›
  • i3
  • ›
  • Smart(er) Internet Regulation Through Co...
  • ›
  • Balancing fundamental rights
  • Presses des Mines
  • Presses des Mines
    Presses des Mines
    Informations sur la couverture
    Table des matières
    Liens vers le livre
    Informations sur la couverture
    Table des matières
    Formats de lecture

    Plan

    Plan détaillé Texte intégral Introduction What are fundamental rights? Freedom of expression Privacy Fundamental rights and proportionality Notes de bas de page

    Smart(er) Internet Regulation Through Cost-Benefit Analysis

    Ce livre est recensé par

    Précédent Suivant
    Table des matières

    Balancing fundamental rights

    Texte intégral Introduction What are fundamental rights? Characteristics of fundamental rights The cost of fundamental rights Economic vs. non-economic rights The expressive value of fundamental rights Freedom of expression General limitations to freedom of expression Is the internet like television? The nature of harms to freedom of expression Internet intermediary liability and free speech The Dennis formula and its limits Law and economics explanations for the high protection given to freedom of expression Freedom of expression and self-regulatory measures Privacy Privacy and data protection as fundamental rights Privacy rights in law and economics literature Behavioral economics and privacy Cost-benefit analysis applied to data protection How to measure costs and benefits in privacy Fundamental rights and proportionality The three-criteria test of the European Court of Human Rights Should a court give deference to lawmakers’ balancing? Identification of the conflicting rights and interests Balancing the relevant interests Absolute versus relative proportionality, cost-benefit analysis Proportionality and the “least injurious means” test Robert Alexy’s balancing test Nussbaum’s ethical filter Fundamental rights and the Hand formula Notes de bas de page

    Texte intégral

    Introduction

    1Many content policies protect fundamental rights. Laws prohibiting online copyright infringement protect the right to property, a fundamental right. Laws requiring search engines to delist certain old and irrelevant content under the “right to be forgotten” doctrine protect data privacy, a fundamental right. But those laws also restrict freedom of expression, which is also a fundamental right. Laws prohibiting content that incites racial hatred protect equality and non-discrimination, a fundamental right. When internet intermediaries take actions to enforce content policies, the actions will often favor one right at the expense of others. This requires balancing.

    2Courts in the United States and in Europe do this balancing routinely. In Europe, the balancing is known as the “proportionality test.” The purpose of this chapter is to look at the rights that are being balanced through the lens of law and economics, as well as to examine the balancing test itself. The proportionality test examined at the end of the chapter is a form of cost-benefit analysis, even though the costs and benefits associated with fundamental rights can rarely be quantified in monetary terms. This chapter will lay the groundwork for Chapter 6, where the balancing of fundamental rights will be inserted into the methodology used to evaluate proposed regulatory measures. I will propose in Chapter 6 several mechanisms for measuring impacts to fundamental rights, including a scoring mechanism inspired by Robert Alexy (2012), examined in this chapter.

    What are fundamental rights?

    Characteristics of fundamental rights

    3Most fundamental rights come from the philosophers of the enlightenment, who defined certain natural rights from which all human beings should benefit (Ségur, 2012). These principles first appeared in the French Universal Declaration of Human and Citizens’ Rights and the United States’ Bill of Rights, both adopted in 1789. After World War II, the United Nations adopted its Universal Declaration of Human Rights (December 10, 1948), and the Council of Europe adopted the European Convention on Human Rights (November 4, 1950). More recently, the European Union adopted its Charter of Fundamental Rights (December 7, 2000), which became binding on all European Member States under the Lisbon Treaty (2009).

    4From an economic standpoint, fundamental rights can be approached based on their characteristics.

    5The first characteristic is that fundamental rights arise under a legal instrument, such as a constitution, that is difficult to change (Posner 2011). Fundamental rights generally arise from a constitution, or an international treaty having a status equivalent to a constitution. In the hierarchy of legal instruments, a constitution is at the top of the food chain. A constitution is more difficult to modify than a law. A law can be changed through consent of the government and a simple majority in the legislature. A constitution can be modified only with a supermajority vote of both houses of the legislature, plus other conditions that vary depending on the country. Because of these more stringent conditions, constitutions are difficult to modify, even by the ruling majority. Posner (2011) compares a constitution to a contract, which can only be modified with the consent of all the relevant parties. In other words, fundamental rights are legal rules that are very costly to modify, at least in a properly-functioning democracy.

    6The second characteristic of fundamental rights is that they are designed to constrain the actions of the respective branches of government, including the actions of the legislature itself. The constitution protects citizens against pitfalls of democracy, including abuses of the majority. Certain laws democratically adopted by the legislature could undermine the structure or ideals on which the nation is built. Constitutional rights are designed to prevent this from happening, or at least make it considerably more difficult. This aspect of fundamental rights can be compared to laws that restrain monopoly power, except that in the case of fundamental rights, the monopoly power is the power of the state. Posner (2011) calls the state the “most dangerous” monopoly. Many constitutional provisions protect individuals against actions of the state, including all its branches (government agencies, regulatory authorities, legislature or court). Monopoly power in the private sector can be constrained by antitrust laws voted by the legislature and government enforcement of these laws. Monopoly power of the state cannot be so constrained, because the persons who adopt and enforce laws are also the ones who benefit from the monopoly power. This creates a conflict of interest problem, which is why most fundamental rights protect citizens against actions of the state, as opposed to actions by other citizens. Citizens are protected against actions taken by other citizens through ordinary laws, adopted by majority vote of the legislature and enforced by the government and courts. Citizens are protected against actions taken by the state through constitutional rights. This is an oversimplification. In some cases, a fundamental right can be enforced against other individuals. For example, the right to privacy is a fundamental right recognized under the European Charter of Fundamental Rights. It requires that countries have legal mechanisms in place to allow individuals to assert their right to privacy against other individuals. The same holds true for the right to property: the state must have legal mechanisms in place to allow individuals to assert their property rights against the state and against other individuals. If a European Member State does not have those mechanisms in place, the state has violated a fundamental right. Nevertheless, it is fair to say that the primary function of fundamental rights is to protect individuals against actions of the state, including abuses of the then-governing political majority.

    7A third characteristic of fundamental rights is that they are generally framed as general principles, not as detailed rules. This characteristic flows from the first: if fundamental rights arise from legal instruments that are difficult to change, the rights must be drafted so that they can stand the test of time. Courts must apply rights to different circumstances; rights must survive social and technological change. Rights may also transcend geography. This is why rights are sometimes called “universal.” This characteristic of fundamental right is also shared by many “ordinary” laws, which are drafted as general principles. This characteristic is therefore not unique to fundamental rights.

    8The last characteristic of fundamental rights is that they generally have achieved historical and international legitimacy. A fundamental right is generally not found in just one country’s constitution, but in many countries’ constitutions and international treaties. A fundamental right generally flows from the core values originally identified in the United States Bill of Rights and the French Universal Declaration of Human Rights of 1789. Fundamental rights share the same family tree. A country could conceivably invent a new constitutional right, such as the right for all citizens to have broadband access. Being placed in the country’s constitution, the new right could arguably be considered a “fundamental right” in that country. However, in the absence of any international or historical legitimacy, the new right would probably be considered – outside the relevant country at least – as a legal rule lacking the status of a “fundamental” right. Once the new right is recognized in several constitutions and/or in an international treaty on fundamental rights, the new right may achieve the status of a “fundamental right” internationally.

    The cost of fundamental rights

    9Economists focus on the costs and benefits of fundamental rights. Fundamental rights are sometimes classified as “negative” or “positive” rights (de Vries, 2013). A negative right is the right to be free from action of the state, such as confiscation by the state of property without compensation. A positive right is the right to receive some benefit from the state, such as the right to receive health care. Holmes and Sunstein (1999) argue that all rights – including so-called negative rights – require expenditures by the state. No right, including a negative right, exists without an effective enforcement mechanism, which costs money. A right to private property does not exist if there is no court system to recognize the right, and no police force available to enforce the courts’ decisions. Enforcement mechanisms require state expenditures. Thus every right, whether positive or negative, can be seen as a competing claim against limited government resources. The allocation of scarce tax resources is generally determined by the legislature, by simple majority vote. A fundamental right may therefore exist in a country’s constitution but become ineffective if the legislature fails to devote sufficient resources to the right’s enforcement.

    10For example, a country may include data protection as a fundamental right in its constitution. But if the legislature of the country does not grant sufficient resources to the data protection authority, the fundamental right may have little existence in practice.

    11The absence of corruption is also a precondition for effective rights to exist. A fundamental right that exists on paper may dry up in practice if the financing for its enforcement is insufficient, or if the financing comes in part from graft. Fundamental rights therefore depend on taxes and on an efficient distribution of tax resources to non-corrupt courts and police authorities.

    Economic vs. non-economic rights

    12Fundamental rights are sometimes classified as “economic” and “non-economic” rights (de Vries, 2013). An example of an economic right is the right to conduct a business, or the right to protection of property. An example of a non-economic right is the right not to be tortured, or the right to freedom of expression. The distinction between economic and non-economic rights does not necessarily render one of the rights inferior to others, but will affect the balancing tests applied by courts. Some rights, such as the right not to be tortured, are absolute and can suffer no exceptions. Most rights, however, including non-economic rights such as freedom of expression and privacy, can be restricted based on a balancing test.

    13Finally, certain international instruments contain two kinds of fundamental rights: rights that are directly binding, and aspirational principles that are not directly binding (De Vries, 2013). Article 25 of the European Charter on Fundamental Rights provides an example of an aspirational principle: “The Union recognises and respects the rights of the elderly to lead a life of dignity and independence and to participate in social and cultural life.” Being aspirational, this right could not be directly enforced by an individual against the state.

    The expressive value of fundamental rights

    14Fundamental rights often have a symbolic function intended to influence social norms (Sunstein 1996). At one level, the statement “everyone is equal before the law”1 signifies that the state may not make different laws or punishments for different people. This literal reading of the statement constrains government behavior by limiting the kind of law that the legislature can adopt. This function aims to limit abuse of government monopoly, which is the primary objective of fundamental rights.

    15On another level, the statement that “everyone is equal” has a larger symbolic meaning intended to send a message to society that discrimination is wrong. It attempts to push social norms in the right direction. In this sense, fundamental rights can have a strong moral component, or as Sunstein (1996) puts it, an “expressive function.” The moral component can, but does not necessarily, mean that the relevant right is absolute and cannot be balanced. Some rights with a high moral component, such as the right to life2, are absolute. But other rights, such as freedom of expression and privacy, carry a strong moral component, but can nevertheless be balanced.

    16The expressive or moral component of fundamental rights also leads to debates as to whether fundamental rights can be integrated into a broader welfare analysis in which the objective is to maximize social welfare. Nussbaum (2000) argues that the function of rights is to preserve values that are different from the maximization of wealth typically reflected in a welfare analysis, and that violations of constitutional rights cannot be traded for benefits. Kaplan and Shavell (2006) argue on the other hand that a welfare-based approach is perfectly capable of capturing values other than maximization of wealth in a monetary sense. Welfare economics seeks to make everyone better off, and making people better off includes various indicia of happiness, including the existence and enforcement of individual rights. While these factors for happiness are difficult to quantify in a monetary sense, there is nothing inherently incompatible between an approach that seeks to maximize collective wellbeing and an approach that recognizes individual rights, including their moral or expressive component.

    17In an article examining enforcement of national content policies on the internet, Schultz (2008) refers to the normative role of laws, and the important social role played by court or regulatory decisions that declare certain forms of content unacceptable. The actual efficacy of the decision in preventing access to content may, in some cases, be secondary:

    “Brutally simplified, it is one of law’s functions to say what, according to the law governing and tying together a nation, is right or wrong; without necessarily punishing as a consequence.” (Schultz, 2008, p.822)

    18Focusing on the symbolic importance of regulatory decisions, particularly those upholding fundamental rights, can lead courts and regulators to ignore questions linked to the efficacy of their decisions in reducing a particular harm, and to the costs associated with implementing their decisions.

    19Fortunately, the doctrine of proportionality, examined later in this chapter, requires regulators also to take into account the costs and benefits of their decisions, thereby bringing welfare analysis back into the equation.

    Freedom of expression

    20This section takes a closer look at freedom of expression, which is the fundamental right that is the most affected by regulatory measures targeting internet intermediaries.

    General limitations to freedom of expression

    21Although a fundamental right, freedom of expression is not absolute, and is routinely limited by other laws. Copyright, for example, allows the owner of the copyright to prevent anyone else from reciting his or her poem publicly, or even adapting it to create a new derivative work. This is a direct restriction of freedom of expression. Laws punishing defamation directly restrict freedom of expression, as do laws prohibiting the unauthorized publication of personal data, trade secrets, medical secrets, banking secrets, or state secrets. The Wikileaks and Snowden controversies illustrate the tension between freedom of expression and laws prohibiting the publication of classified information.

    22Dozens of laws limit freedom of expression in democratic societies. The main difference between these laws and the laws in totalitarian regimes is that the laws in democratic societies will not prohibit the publication of content that is critical of a political party, government or religion. The ability to openly criticize government is the underlying reason why freedom of expression is given so much constitutional protection. Without the ability to openly criticize government, democracy cannot exist (Post, 1996).

    23In economic terms, freedom of expression preserves the “marketplace of ideas,” from which socially optimal solutions and democratic governance emerge (Coase, 1974). Mialon and Rubin (2007) refer to freedom of the press and freedom of political speech as the “mother of all rights” because they are essential for creating and enforcing other rights. “If government violates a right but no one can learn of the violation, then there is no cost to the government for the violation.” (Mialon and Rubin, 2007, p. 11).

    24Courts are therefore vigilant when it comes to measures that might have an adverse impact on freedom of expression, particularly when the measures affect the internet. The internet is considered one of the most important enablers of freedom of expression since the invention of the printing press (United Nations, 2011). In the 15th century, kings and clergy attempted to regulate printing presses in order to enforce content policies. The regulation of internet intermediaries is similar, which is why courts approach such measures with caution.

    Is the internet like television?

    25This discussion is about regulating access to harmful content on the internet. A legitimate question is: Why not just apply the same rules as for television? After all, the problem of harmful content is similar, and television rules have been in existence for decades. The answer to this question resides in how courts apply freedom of expression principles to these two media, internet and television.

    26When internet first emerged as a major media, lawmakers’ first reaction was to apply television-like rules. In 1997 the United States Supreme Court invalidated a law that proposed to apply broadcast-like regulation to internet content. The measure was intended to protect children. In the now-famous Reno v. ACLU case3, the Supreme Court found that the internet is equivalent to a tribune in a public square, where citizens can speak or distribute leaflets freely. The court disagreed with the United States Department of Justice, which argued that the internet could be regulated like television. The court found that unlike television, which “pushes” information to passive viewers in a living room, the internet requires users actively to seek out information, as they do when they enter a library. Unlike television or radio, internet users will generally not be shocked by the information they receive, since they are actively seeking the information to begin with.

    27The other reason why the court refused the broadcasting analogy in 1997 is that information on the internet is not limited, whereas information on television is limited due to the scarcity of radio frequencies. Because only a limited number of television broadcasters can be licensed to provide television or radio services, the government is justified in imposing content rules on the license holders. In sum, courts are more tolerant of government efforts to regulate television and radio because of (a) television and radio’s higher influence on public opinion, and (b) the scarcity of broadcasting frequencies.4 The scarcity argument may disappear, because television is increasingly carried over broadband networks. Nevertheless, it is still generally accepted in the United States that television can be regulated more aggressively than the internet without violating the First Amendment of the United States Constitution.

    28European courts have followed the trend set in ACLU v. Reno, by according a high degree of protection to expression on the internet, similar to the protection given to print media.5 In Europe as well, television regulation cannot be transposed as-is to internet content.

    The nature of harms to freedom of expression

    29When examining technical measures affecting internet intermediaries, courts have identified three different kinds of potential harms to freedom of expression. The first and most direct impact is on the publisher of information whose message becomes blocked and thereby inaccessible to internet users. This is the case for example of the publisher of a controversial video, such as “The Innocence of Muslims,” whose video becomes inaccessible to large numbers of people either because the video was taken down from YouTube, or because access to the video was blocked by an internet access provider. This is also the case for the author of an article that has been subject to a delisting request under Europe’s “right to be forgotten” doctrine. In the latter case, the article will be more difficult to find using certain search terms.

    30The second more indirect harm to freedom of expression is when a technical measure is overbroad, and prevents an internet user from obtaining access not only to the information directly targeted by the measure (e.g. material the infringes copyright or that violates laws on child pornography), but also to other information that is not directly targeted by the measure. A clear example of this is where access to the entire YouTube service is blocked in a country because of the presence of a single video that violates local law. However, even measures that are less obviously overbroad attract court scrutiny. A user’s inability to obtain an unauthorized copy of a recent motion picture such as “Gravity” on the internet will not be considered a restriction on that user’s freedom to access information on the internet because the access would violate copyright laws. However, if the system put into place to limit access to “Gravity” might also limit access to another film with the word “gravity” in the title, or a work that parodies “Gravity” but benefits from an exception to copyright, courts will view the measure as a potentially serious restriction to freedom of expression. The Supreme Court in the United States invalidated parts of the Communications Decency Act6 because it was overbroad: the measure targeted “sexually explicit” content, which could limit minors’ ability to search for information on birth control. The European Court of Justice invalidated a measure that might inadvertently block content that does not benefit from copyright protection.7 Any measure that is slightly overboard will generally be struck down as an excessive restriction on freedom of expression.

    31The third and last potential harm identified by courts is the chilling effect that some measures may cause to freedom of expression by causing publishers and intermediaries to limit their activity, to become overly careful in what they say or publish. As explained below, critical and disturbing speech is considered a public good, necessary for the marketplace of ideas to function properly. Such speech may be under produced if not protected. Courts, particularly in the United States, are critical of legal rules that might constrain unpopular speech, even through indirect effects. These indirect effects are often referred to as “chilling effects.”

    32Some laws require ISPs or hosting providers to collect information about the identity of publishers of content so that law enforcement authorities or victims of copyright infringement can identify the author of the content and bring legal action if the content proves illegal. These mechanisms may discourage publishers of controversial content from publishing the content in the first place, or may encourage them to publish on non-public platforms. By discouraging the publication of controversial content, the measure has the effect of weakening open criticism and the exchange of ideas. The concern here is not to protect publishers of manifestly illegal content, but publishers of content that is legal but that might be close to the line, or that could lead to adverse consequences for the author. For example the composer of a musical mash-up may not be 100% sure that her work falls under the fair use exception to copyright. The author of a video revealing unfavorable information about a large corporation may fear reprisal if his or her identity is known. The absence of anonymity creates a chilling effect on free expression.

    33The importance accorded to chilling effects is due to the fact that much speech and ideas are public goods (Posner, 2007, p. 727). As pointed out by Mialon and Rubin (2007):

    “[a]nything that adds to the cost of speech will then have a large suppression effect, since many of the benefits of speech accrue to others anyway. This may be especially true of political speech. Political discourse and debate tend to produce better decisions for society, but individuals often have few incentives to participate, since they do not capture most of the benefits of their own participation and incur opportunity costs, as they could spend their time pursuing personal gain instead.” (Mialon and Rubin, 2007, p. 4)

    Internet intermediary liability and free speech

    34Liability imposed on technical service providers is also a potential source of chilling effects, because such liability prompts ISPs and other technical intermediaries to reduce their potential costs by eliminating risky, albeit legal, content from their service (United Nations, 2011, p. 11).

    35Schruers (2002) shows that holding internet intermediaries liable for the content posted by users would cause the intermediaries to select only low-risk users and content, leading to a general decrease in the amount of content available online, and to an elimination of risky content. This would harm the market for ideas. Laws that limit the liability of internet intermediaries are intended to reduce this chilling effect. Under the notice and takedown approach applied in Europe and the United States, internet intermediaries are generally not liable for the content uploaded by users, but must promptly remove content once they have received notice. But even the notice and takedown regime can lead to excessive removals, thereby harming the market of ideas.

    36Ahlert et al. (2004) conducted a “mystery shopper” experiment. First, they uploaded to several hosting platforms the work “On Liberty” by John Stuart Mill. Written in 1859, the work is no longer protected by copyright. The authors of the study, disguising themselves as representatives of a fictitious “John Stuart Mill Foundation,” then sent notices to the hosting providers asking that the work be removed because of copyright infringement. Most of the hosting providers in Europe complied, without stopping to verify whether the “John Stuart Mill Foundation” actually exists, or whether “On Liberty” is still protected by copyright. The authors of the study use this experiment to show that even under the protective rules of the European E-Commerce Directive8, the fear of liability by internet intermediaries leads to a significant chilling effect, i.e. the removal of legal content that should not be removed.

    37Seltzer (2010) argues that the Digital Millenium Copyright Act in the United States leads to a similar chilling effect, because takedown notices related to copyright infringements are systematically followed, even if the relevant use of the content might qualify for fair use. In its recommendations on internet policymaking, the OECD (2011a) underlines the need to preserve the limited liability of internet intermediaries in order to foster freedom of expression.

    The Dennis formula and its limits

    38The ideology behind freedom of expression in the United States is that a vibrant marketplace of ideas is essential for democracy and economic progress. Expression that does not contribute to the marketplace of ideas, such as threats of violence against an individual, or commercial advertisements, will enjoy a lower level of protection. Expression that contributes to the marketplace of ideas, such as political speech, will be entitled to high protection even if the speech is offensive.

    39Posner (2011) describes how United States courts evaluate measures that limit freedom of expression. On one side of the equation, courts will weigh the harm caused by the relevant content, and the probability of its occurrence. The timeframe within which the harm may occur is also measured.

    40On the other side of the equation, courts will evaluate the harm that the measure will cause to the proper functioning of the marketplace of ideas. Measures that suppress a particular point of view, particularly in political debate, will create a large harm to the marketplace of ideas, and will almost never be permitted.

    41Posner summarizes this equation as follows. A measure limiting freedom of expression will be permitted if:

    42B < PL / (1 + i)n, where:

    • B is the total cost to society caused by the proposed measure, including the adverse effect that the measure will have on the marketplace of ideas, in particular through chilling effects;
    • L is the cost linked to the adverse event that the regulatory measure seeks to prevent, e.g. racist crimes or terrorist attacks;
    • P is the probability that the loss “L” will occur in the absence of regulatory intervention;
    • i is the annual discount rate;
    • n is the number of years before the adverse event will occur in the absence of regulatory intervention.

    43The formula as summarized by Posner assumes implicitly that P would be zero in the presence of regulation, i.e. that the regulatory measure would be 100% effective in eliminating the risk of harm. Most regulatory measures are not this effective. They will only reduce P, but not eliminate it entirely. If we assume that the regulatory measure reduces the level of P, but does not eliminate it entirely, then the Dennis formula leads to the following:

    • P is the probability of the loss occurring without the regulatory measure;
    • Po is the probability of the loss occurring with the regulatory measure.
    • P >Po>0

    44The total cost of the situation with regulation would be:

    45For the regulation to be justified, the total cost of the scenario with regulation (B+P0L) would have to be less than the cost of the scenario without regulation B+P0L (1+i)n < PL(1+i)n

    46In other words:

    47Image  

    48Posner’s formula comes from the reasoning of a United States Federal Court of Appeals in the case United States v. Dennis.9 It is sometimes referred to as the “Dennis test.” By applying this test, a regulation prohibiting the publication of information on how to make a chemical weapon at home might be justified, because of the high level of L, the relatively high level of P, and low value of n (the harm could occur soon). Moreover, the recipe for a home-made bomb contributes little to the marketplace of ideas, so B would be low.

    49By contrast, a regulation prohibiting speech that calls for the violent overthrow of the United States government would not be justified. Although the harm caused by the violent overthrow of the government (L) is high, the likelihood of it happening (P) is low. The timeframe within which the harm may occur is also fairly remote, leading to a high value for n. Moreover, unlike the speech in the homemade bomb case, the speech in the violent government overthrow case may contribute to the political marketplace of ideas. Prohibiting it would cause a high level of B.

    50The results of the Dennis test may vary over time. When the United States was a young and fragile democracy, violent overthrow was perceived as a real and immediate risk. P was high, and n was low. A law prohibiting content advocating the violent overthrow of the government would potentially satisfy the Dennis test, even though the effect on the marketplace of ideas (B) would still be high. France recently adopted a law allowing police authorities to order ISPs to block access to sites promoting terrorism.10 The law was intended to limit access to sites that recruit vulnerable young people to join terrorist groups in Syria. Critics of the law argued:

    • that the measure would affect not only websites that promote terrorist acts, but also sites that promote certain religious ideals. In other words, B would be high because of the suppression of speech that is important for the market of ideas.
    • that the law would not decrease P, because young people would find other technical ways to access the sites, and the outlawed sites would gain in popularity because of the blocking measure. Forbidden by the government, the sites would become even more desirable.

    51This last example illustrates an important point relating to the European principle of proportionality, and which is not directly reflected in the first version of the Dennis formula outlined above. If a measure will have a high cost on a fundamental right (a high B value), policymakers and courts should verify whether there are other alternative measures that would be just as effective – or more effective – in lowering P, while carrying a lower cost (B) on fundamental rights. An alternative measure with a lower B should always be preferred where possible over a measure carrying a higher B. Proportionality requires choosing the “least intrusive means.”11 In the case of measures designed to limit the recruitment of young people into terrorist organizations, the use of traditional police investigation tools, and the arrest of group organizers, will surely be more effective than site blocking. However, the question is often not “either or,” but whether the site-blocking measure, in addition to traditional police tools, will make a difference in lowering P. If it does, the measure may still be justified in spite of the high level of B.

    52In the case of the French site blocking law, there was to my knowledge no study estimating the impact of the measure on P. The government’s impact assessment assumed that the sites should be blocked, and only examined whether the blocking should be ordered by a judge, or by officials within the Ministry of Interior.12 For the proponents of the law, and the parliamentary majority, site blocking was assumed to be the right thing to do, but empirical evidence was not examined on the question of whether site blocking would reduce the probability P of the relevant bad event L occurring.

    53It is important to point out here that even if empirical evidence showed the limited utility of the measure in reducing P, lawmakers may enact the law anyway. This is because of the expressive function of law (Sunstein, 1996), i.e. the use of law to send signals to society as to what is acceptable and what is not. In the market for law making, demand comes from voters who expect elected officials to take action with regard to socially unacceptable content (Posner 2011, Stigler 1971). The producers of laws, if they want to be reelected, must make laws that respond to this demand. A law with a high symbolic (or “expressive”) effect will often satisfy this demand, even if the law has a low empirical impact in reducing P.

    54Whether regulators will choose the measure that is “efficient” versus the measure that is “expressive” will depend on a number of factors, including those described by Stigler (1971) and Peltzmann (1976).

    55As I will explain in Chapter 7, political decision-making can result in measures that are not efficient insofar as they do not maximize social welfare. The purpose of my methodology is not to replace sometimes inefficient political decision-making with an infallible scientific formula, but rather slow down the decision-making process by imposing a questionnaire and checklist that require policymakers at least to consider alternatives that tend to maximize social welfare. If policymakers choose another alternative, the choice would at least be more explicit.

    Law and economics explanations for the high protection given to freedom of expression

    56Freedom of expression, particularly in the United States, targets actions taken by the state. According to constitutional theory in the United States, the government is particularly ill-qualified to select “good” and “bad” ideas. The government will have an inherent conflict of interest, favoring ideas that pose the least threat to the political status quo. By influencing the marketplace of ideas, the government can affect the proper functioning of democracy itself, creating an intolerably high cost to society (Breton and Wintrobe, 1992). When it comes to choosing ideas, the government is the “most dangerous monopoly” (Posner, 2011).

    57Coase (1977) takes exception with this reasoning, arguing that the high protection accorded to the marketplace of ideas under United States law results simply from successful lobbying by the intellectual elite, who are the main producers of ideas affected by content regulation. Like any other producer of goods or services, the intellectual elite will lobby for a market in which their production is not limited by regulation. The intellectual elite are also particularly effective at making sure their interests are protected in the legislature and courts. This view echoes Stigler’s (1971) public choice approach.

    58If unregulated, the free marketplace of ideas can create problems of adverse selection:

    “If listeners cannot distinguish ideas in terms of their truth content, they must regard the ideas as containing an average proportion of truth.” (Mialon and Rubin, 2007, p. 6)

    59This could lead to a general decrease in the quality of ideas in the marketplace, which suggests that screening out bad ideas could be efficient. Listeners to ideas also have limited attention, which may lead to congestion. Listeners may not be able to hear good ideas over the noise of bad ideas. Here, too, screening out bad ideas appears attractive. However, the cost of screening, and the risk of error, are high. Moreover, as noted above, the government would not be a reliable entity to perform the screening (Mialon and Rubin, 2007).

    60As noted above, Posner (2011) and Mialon and Rubin (2007) argue that good ideas have the characteristics of public goods, and will be under-produced by the market without support mechanisms. To avoid under-production of ideas, the state must take affirmative measures, including measures that reduce “chilling effects” that might create costs for the creation and exchange of ideas.

    Freedom of expression and self-regulatory measures

    61Although the United States First Amendment targets laws and regulations adopted by the government, Kreimer (2006) explains that government action can take other more indirect forms that affect freedom of expression, including applying pressure on internet intermediaries to contribute voluntarily to content policies. Kreimer draws a parallel with action taken voluntarily by private media companies in the 1950s during the McCarthy era. At the time, United States media companies fired and/or boycotted individuals who refused to testify before the House Committee on Un-American Activities. This resulted not from direct regulation, but from indirect government pressure on companies. Companies responded to the pressure by applying voluntary measures. Kreimer uses this example to illustrate the fact that self-regulatory measures taken by internet intermediaries can result from government constraint and lead to harmful effects on freedom of expression.

    62Garfield (1998) and Benkler (1999) argue that in spite of the state action requirement of the First Amendment of the United States Constitution, freedom of expression principles can apply to private contracts. They argue that state action is present when the government creates laws and courts enforce private property and private contracts, and that the enforcement of a contractual clause restricting expression can also be considered as a government action.

    63This consideration is important when evaluating self- or co-regulatory measures applied by internet intermediaries to enforce national content policies. Although not imposed by direct government constraint, self- or co-regulatory measures can in some cases interfere with freedom of expression. The intensity of this interference will generally be lower than in the case of government regulation.

    Privacy

    64This section examines the second fundamental right that is most frequently affected by technical measures: privacy.

    Privacy and data protection as fundamental rights

    65Privacy and data protection refer to two closely-related rights: privacy is the right to protection of one’s private life against outside interference. It is the right to be “left alone” (Warren and Brandeis, 1890). Data protection is the right to control how personal data is used. The two rights often overlap. Americans often use the term “privacy” to designate both privacy and data protection rights. Europeans often use the term “data protection” to designate both rights. For purposes of this discussion, I will use the terms privacy and data protection interchangeably, to designate all privacy (or data protection)-related rights.

    66In Europe, the protection of one’s private life, and the protection of one’s personal data, are both considered fundamental rights. They are protected by the European Convention on Human Rights and by the EU Charter of Fundamental Rights. In the United States, the right to protection against government intrusion into privacy is a fundamental right, embodied in the Fourth Amendment to the United States Constitution. However, the United States constitutional right is limited to situations that are equivalent to government searches of one’s private home. Over time, United States courts have extended the Fourth Amendment to different situations, including private telephone conversations, the inside (and sometimes outside) of private vehicles, and the outside of a private homes. Mialon and Mialon (2008) studied the effect of the Fourth Amendment on police conduct, and more particularly the effect of the exclusionary rule, which prohibits the use of any evidence obtained from illegal searches. They concluded that the exclusionary rule reduces the number of government searches conducted without sufficient justification (“probable cause”), but also leads to an increase in crime. The effects on welfare are ambiguous. This conclusion also applies to government actions to increase surveillance. Privacy legislation will decrease the level of abusive surveillance by the government of innocent people, but will also increase the risk of actual terrorists escaping surveillance. The weight of these two effects will be difficult to measure with certainty. The purpose of the proportionality test examined later in this chapter is to force regulators to at least try to evaluate the effects (positive and negative) of their measures, and thereby make measures more effective with fewer adverse effects on fundamental rights. The SURVEILLE project described in Chapter 7 attempts to do this by proposing a standard methodology against which to evaluate police surveillance measures. These methodologies do not replace political decision-making, including the adoption of sometimes inefficient but symbolically expressive laws. But the methodologies would obligate policymakers to conduct more meaningful impact assessments prior to proposing new laws or regulations, thereby making choices more explicit.

    67In the United States, the protection of individuals against privacy (or data) intrusions by other individuals or private entities is not a fundamental right. However, the protection is provided either by specific laws, or by common law tort principles (Whitman, 2004; Maxwell, 2014).

    68As is the case for freedom of expression, privacy is a fundamental right that can be interfered with. In Europe, interference with privacy is possible if the measure passes the proportionality test, which will be examined later in this chapter. In the United States, courts will apply a similar balancing test to regulations that affect individuals’ rights to be free from unreasonable surveillance. Most regulations involving internet intermediaries will have an impact (positive or negative) on privacy rights, which is why an understanding of privacy rights, and of the proportionality test, are essential in order to build a methodology for assessing measures affecting those intermediaries.

    Privacy rights in law and economics literature

    69Posner (1981) distinguishes three separate meanings for privacy. The first is the right to conceal information about oneself. The second is the right not to be bothered. The third is privacy as part of a broader right to individual freedom and autonomy. Posner argues that the right to conceal personal information creates inefficiencies that make everyone worse off. If unfavorable information about a product must be disclosed to a prospective buyer, so should unfavorable information about a person, when the person is the “thing” being sold, as is the case in an employment context:

    “The basic point I wish to assert is the symmetry between “selling” oneself and selling a product. If fraud is bad in the latter context (see Michael Darby and Edi Karni) – at least to the extent that one would not think it efficient to allow sellers to invoke the law’s assistance in concealing defects in their goods – it is bad in the former context for the same reasons: it reduces the amount of information in the market, and hence the efficiency with which the market – whether the market for labor, or spouses or friends – allocates resources.” (Posner, 1981, p. 406)

    70Posner concludes that most privacy legislation has a redistributive function, subsidizing classes of individuals whose personal information is unflattering, e.g. persons with a criminal history, but at the expense of other individuals (sellers) and buyers in the market, who would benefit from transparency.

    71United States law recognizes the invasion of privacy as a common law tort, involving four separate privacy rights:

    • the right to object to the use of one’s name or image in advertising;
    • the right not to be portrayed in a “false light;”
    • the right to prevent the collection of personal information using intrusive means (trespass, eavesdropping);
    • the right to prevent the publication of intimate facts about oneself.

    72Posner argues that the first three privacy torts are economically efficient. The first confers a form of property right over an individual’s personal information when used for advertising, which will have the effect of maximizing the value of the information and ensuring its best usage. The second right (“false light”) increases the amount of correct information in the market, thereby making transactions more efficient. The third right (“intrusive means”) is similar to trespass, and has the benefit of encouraging people to speak frankly on the assumption that they will not be listened to. This favors the free exchange of information, and avoids wasteful investment in privacy-enhancing technology, such as high walls around a garden, or building a communications technology that cannot be intercepted.

    73The last tort relates to concealment of true information, which Posner explains can be either efficient or inefficient depending on the transaction. If the concealment is done to mislead a contracting party on a matter that is material for a transaction, Posner believes that concealment should not be protected by law. However, if the concealment relates to information that has no possible value to the transacting parties, then the rule is efficient.

    74Stigler (1980) also asserts that rules allowing individuals to conceal unfavorable information about themselves are inefficient because (i) they will reduce the accuracy of categorization, i.e. people will still be placed in categories, but the categories will be less accurate, and (ii) transacting parties will turn to more costly and potentially less reliable means to obtain the information they seek, thereby increasing search costs and lowering the quality of the information obtained.

    75Varian (1996) highlights the annoyance caused by certain privacy violations, such as being bothered by unwanted telephone or e-mail solicitations. These intrusions create significant costs for consumers by using up the consumer’s scarce attention. But these costs can be reduced if the advertiser has more information about the consumer, not less. If the advertiser knows exactly what the consumer is interested in receiving, the annoyance costs for the consumer will decrease because the consumer will only receive solicitations that he or she is interested in. In this sense, the merchant’s and the consumer’s interests are aligned. The consumer will benefit by letting the merchant know exactly what the consumer is interested in. A mutually profitable bargain can emerge. The situation changes, however, for secondary use of personal data, such as where the original merchant sells its mailing list to a third party. In that case, the purchaser of the mailing list will create costs for the consumer that are not reflected in the price paid for the mailing list, thereby creating negative externalities. To correct this, Varian suggests that consumers have the ability to control secondary use of their information, such as permitting secondary use only in exchange for a payment, or only for certain purposes, or only for a certain time. Varian’s suggestion comes close to the purpose limitation rule that exists under European data protection law, with the important exception that European law has not yet established a means to organize a market for the right to secondary use against payment. European law is also generally hostile to the idea that individuals possess a tradeable right to use personal data.

    76Varian addresses the problem of open data, highlighting privacy objections that have been raised against making public records, previously available through a manual search, available online. Varian asserts that each situation requires a cost-benefit analysis, to determine whether potential benefits from eliminating manual search costs outweigh the additional threat to privacy. One alternative is to impose a fee for online searches that approximates the previous cost of conducting the search manually, including travel and photocopying costs. The fee could be used to help defray the cost of making the information available online, and would result in the same search costs as those that existed before records were available online.

    77Laudon (1996) suggests the creation of a National Information Market that would serve as a clearing house for managing consumers’ authorizations to use personal data. The National Information Market would aggregate payments and distribute the payments to consumers who participate in the collective system. Laudon’s system resembles the collective management of copyright.

    Behavioral economics and privacy

    78Much of the Chicago School’s analysis of privacy is based on the assumption that individuals are able to make rational choices about the use of their personal data, provided they receive sufficient information. This rational choice assumption permeates data privacy laws in the United States and Europe today, both of which focus on full information and individual consent.

    79In the mid-2000s, economists began to question this rational choice assumption. Beales and Muris (2008) challenged the “notice and choice” philosophy of United States consumer privacy laws:

    “The reality that decisions about information sharing are not worth thinking about for the vast majority of consumers contradicts the fundamental premise of the notice approach to privacy. To be an effective approach, some significant number of consumers must not only read privacy notices for the businesses with whom they currently deal, they must also consider the privacy practices of alternative service providers and choose the provider whose practices best match their privacy preferences. There is no reason to think that this is currently happening, or will ever happen.” (Beales and Muris, 2008, p. 114)

    80Acquisti, John and Loewenstein (2009) show how individuals’ choices on privacy vary based on psychological effects studied in behavioral economics. In particular, Acquisti’s experiments show a significant difference between consumers’ willingness to pay (WTP) for additional privacy protection, and their willingness to accept (WTA) payment to give up pre-existing privacy protections. WTP is systematically lower than WTA. These differences are due to several effects, including the endowment effect (i.e. what I currently have is more valuable than what I would pay to acquire it), loss aversion and status quo bias. Individuals’ choices may also depend on the order in which the choices are presented. Brandimante, Acquisti and Loewenstein (2012) also showed that contrary to the assumptions of most privacy laws, individuals behave more recklessly when they have more tools at their disposal to control their privacy.

    81Borgesius (2013) studied the ramifications of behavioral economics on consent, concluding that “insights from behavioural economics cast doubt on the effectiveness of informed consent as a privacy protection measure. Many people click ‘I agree’ to any statement that is presented to them” (Borgesius, 2013, p. 58). Borgesius suggests that the law should impose certain default rules protecting consumers against the most intrusive forms of tracking (e.g. tracking based on health data), and make the default rules “sticky” by adding transaction costs if the consumer wants to opt out. A minimum number of mouse clicks, a telephone call or registered letter would be required to override the default rule for certain intrusive forms of tracking. Other forms of tracking (e.g. tracking of children) might be prohibited altogether.

    82Posner (2008) makes the useful distinction between a person’s “pure” interest in concealment of personal information and a person’s “instrumental” interest, which is based on fear that the information might be used against him. Many individuals will not hesitate to share highly personal information about themselves with strangers, for example when chatting on an airplane. This tends to show that the revelation of highly personal information in itself creates no damage for an individual, and may indeed create a benefit, otherwise the person would not share the information. The benefit is presumably the ability to learn similar information about the other person and improve one’s own information level and well-being. However, if the information is then used to embarrass or blackmail the individual, a harm will materialize. It is the harm from misuse that creates the harm, not the harm from disclosure itself. Posner (2008) emphasizes that the two should not be confused.

    Cost-benefit analysis applied to data protection

    83In the field of privacy law, explicit cost-benefit analyses are rare. One exception is application of the concept of unfair or deceptive practices by the United States Federal Trade Commission (FTC). Section 5 of the FTC Act prohibits “unfair or deceptive acts or practices in or affecting commerce.”13 The FTC has used this provision to enforce data protection principles against a broad range of companies in the United States, including in the internet sector. The FTC has developed what Solove and Hartzog (2014) call a “new common law of privacy.” The FTC’s enforcement actions, guidelines, and settlement agreements provide details on how the FTC applies the broad principles set forth in the FTC Act to particular facts. This process is similar to what courts do when adjudicating common law tort claims. By examining how claims have been dealt with in the past, observers can anticipate how a standard such as fairness will be applied in the future.

    84Beales (2003) describes how the fairness test has been applied by the FTC from 1938 to present. In the 1970s, the FTC used its authority to prohibit unfair practices in a broad variety of circumstances, relying in part on broad public policy criteria. Critics – and in particular the United States Congress – became concerned that the unfairness standard was too subjective. In 1980, the FTC clarified its approach by adopting its Unfairness Policy Statement (FTC, 1980). Congress then inserted the FTC’s methodology into the FTC Act itself in 1994. The United States Congress wanted to make sure that the FTC would limit itself to an objective methodology when evaluating fairness, and not rely solely on public policy considerations.

    85The 1994 revision to the FTC Act creates a balancing mechanism to determine whether a practice is unfair:

    “The Commission shall have no authority under this section or section 57a of this title to declare unlawful an act or practice on the grounds that such act or practice is unfair unless the act or practice causes or is likely to cause substantial injury to consumers which is not reasonably avoidable by consumers themselves and not outweighed by countervailing benefits to consumers or to competition. In determining whether an act or practice is unfair, the Commission may consider established public policies as evidence to be considered with all other evidence. Such public policy considerations may not serve as a primary basis for such determination.”14

    86United States law therefore requires that the FTC conduct a cost-benefit test to determine whether a practice is unfair. If the practice causes substantial injury to consumers that consumers cannot reasonably avoid, and the injury is not offset by corresponding consumer benefits, then the practice will be deemed unfair. The unfairness test is separate from the FTC’s analysis of whether a practice is deceptive. According to Beales (2003), a deceptive practice is a subset of the larger category of unfair practices. Under the FTC’s methodology, a deceptive practice would not require a cost-benefit analysis, and would be presumed to be unfair. This is understandable because a deceptive practice is tantamount to lying to consumers, and such conduct is not likely to have any offsetting consumer benefits.

    87In the field of data protection, the FTC has used the theory of deceptive practices to sanction companies that do not honor their own privacy policies. In the case where a company has not broken any of its own promises, the FTC will not be able to punish the company for being deceptive. The FTC will have to show that the practice is unfair. To do this, the FTC must first find that the practice causes or is likely to cause substantial injury to consumers. A substantial injury can result from a large injury to a small number of consumers or a small injury to a large number of consumers.

    88Consumer injury for privacy violations is often difficult to measure and has been the focus of much debate in the United States, where the concept of privacy as a fundamental human right is not as ingrained as in Europe (Whitman, 2004). The injury to each consumer taken individually can be extremely small. For example, the excessive collection of data may marginally increase the risk that a given consumer will fall victim to identity theft or receive unwanted advertisements. The individual injury in these situations would be difficult to quantify. Nevertheless, the FTC has stated its belief that these practices may create a substantial injury to consumers (Solove and Hartzog, 2014). Also, any practice that limits a consumers’ autonomy and choice may be considered to create a substantial injury. For example, a default setting in software that leads to unexpected sharing of personal computer files was held to be unfair because it hindered consumer choice.15

    89The injury must also be one that cannot be reasonably avoided by consumers. This element of the equation captures the cost of accident avoidance present in the Hand formula for torts: where the cost for the victim of taking measures to avoid the accident is sufficiently low, it is economically efficient to allow the activity to continue and impose the cost of avoidance on the potential victim. This ties in with the FTC’s mission to ensure that consumers are sufficiently informed and have the opportunity to make choices relating to their privacy. Any hidden or unexpected collection or uses of personal data could be deemed unfair because the consumer did not have a reasonable opportunity to make a choice in the matter.

    90The last step in the unfairness test requires that the FTC evaluate any countervailing benefits. This step requires that the FTC inquire whether the practice in question generates new valuable services, or lower prices, for consumers. In this connection, the FTC must compare the situation that would exist in the absence of any regulation by the FTC to the situation that would exist if the practice were stopped or regulated. The difference represents the costs associated with the FTC’s own regulatory action, and conversely, the benefits associated with leaving the practice unregulated.

    91In summary, the practice would be prohibited if and only if H-HA ˃ WA-WP

    92Where:

    • H is total aggregate consumer harm created by the practice;
    • HA is aggregate harm that consumers can reasonably avoid;
    • WA is total consumer welfare when the practice is allowed;
    • WP is total consumer welfare when the practice is prohibited.
    • In each case W would be calculated without deducting H.

    93The FTC’s unfairness test can best be understood through an example. Imagine that the FTC is considering the practice of setting third-party advertising cookies on users’ computers when the users open a webpage. Is there a substantial consumer injury (H)? There may be, because the third-party ad cookies could lead to embarrassing situations such as when a user is presented an advertisement that is related to his or her browsing history and the user would prefer to keep the browsing history secret. The user may also find such tracking “creepy” making the consumer less inclined to use certain internet services in the future (Tene and Polonetsky, 2013). The FTC could view reduced consumer trust in online transactions as a form of harm. Consumer harm (H) is certainly present, even if its quantification will prove challenging.

    94Can the injury be reasonably avoided by the consumer? This may depend on the level of disclosure provided to the consumer and the availability of easy-to-use tools to block third-party advertising cookies. Good disclosure and an easy, one-click, blocking tool might cause HA to approach H. The cost of avoidance would be low.

    95Finally, is the consumer injury offset by consumer benefits? This step would require the FTC to evaluate the benefits that flow to consumers from the widespread use of third-party advertising cookies. These benefits would consist principally of the wider availability of free online services, which in turn increases consumer choice and freedom of expression. The FTC would have to consider the costs associated with a prohibition of third-party cookies or the imposition of a consumer opt-in mechanism. These costs would be the difference between consumer welfare when the practice is allowed (WA) and consumer welfare when the practice is prohibited (WP). If the costs associated with these regulatory remedies (WA-WP) exceed the costs associated with the consumer injury that cannot reasonably be avoided (H-HA), then the relevant practice would not be considered unfair.

    96To date, the FTC has more readily alleged unfairness in data security-related enforcements (for example, data breaches where companies are alleged to have had unreasonable security practices that put personal information at risk of misuse) than it has in pure privacy-related enforcement actions (for example, where the issue is not security but a company’s decision to share personal information or to target ads to consumers in alleged unexpected ways). Despite this, the FTC has expressed an increased willingness to utilize unfairness even for privacy enforcement.

    97The FTC’s fairness test relies explicitly on a cost-benefit analysis, comparing the aggregate harm caused by a given practice to its aggregate benefits. Where the aggregate harm outweighs the benefits, and avoidance costs are high for the victim, the practice is unfair. Conversely, where the aggregate benefits outweigh the injury (after taking into account reasonable injury prevention measures taken by the victim), the practice is fair. This approach reflects the traditional law and economics approach to tort law, based on the so-called Hand formula (Posner, 2011). Under the Hand formula (named after Justice Learned Hand), a person is negligent if he or she expends costs on injury prevention (B) in an amount less than the amount of the injury L multiplied by its probability P. When calculating P, the injuring party can assume that the victim will also take reasonable steps to avoid injury. Under this approach, not all injuries are prevented, only a reasonable level of injuries. In its simplest form, the Hand formula means that a person will be negligent if, but only if, B < PL. A more refined expression of the Hand formula compares expected costs of harm and expected costs of prevention at the margin. The optimal level of prevention costs occur where an additional dollar of prevention (B) would yield at least a dollar of reduction in harm (PL).

    98From a social welfare standpoint, the objective is to minimize the sum of total costs of prevention (B) plus the total costs of harm (PL).

    99Graphically, this occurs where the lines PL and B intersect:

    Figure 4: showing the changes in costs in harm (PL) and in prevention (B) associated with different levels of care. C* represents the optimal level of care. (Source: Posner, 2011)

    Image

    100This approach comes as close as possible to a negotiated outcome if there were a perfect market for buying and selling risks and injury prevention measures (Coase, 1960). The total costs of injury, plus the total costs of injury prevention spent by the injuring party and the victim are minimized, thereby achieving an efficient outcome from a welfare economics standpoint.

    101The European approach to fair processing in the privacy field focuses not on a cost-benefit analysis, but on the level of information provided to the data subject and the data subject’s ability to exercise his or her individual autonomy. The FTC’s approach is a welfare economics approach, whereas the European approach is an individual rights approach. These two approaches are not necessarily incompatible, although the subject is hotly debated (Kaplow and Shavell, 2006).

    How to measure costs and benefits in privacy

    102Let us focus on the particular problem of conducting a cost-benefit test in data protection. How should costs and benefits be measured? Alessandro Acquisti (2010) and Adam Thierer (2013) explore this difficulty. Thierer’s focus is on conducting cost-benefit analyses in the context of regulatory proposals, following the United States rules on good regulation that I examine in Chapter 5.

    103Acquisti (2010) and Thierer (2013) point out that privacy is an intangible – and in many cases immeasurable – right, similar to the right to pursue happiness. Privacy is often based on consumer emotions, not economic considerations, making economic evaluation difficult. Individuals say that privacy is important, but traditional economic measurement tools, such as willingness to pay (WTP) and willingness to accept (WTA), show that individuals in fact attach a low value to privacy in practice. There is a considerable gap between what people say, and how they actually behave when given a choice to acquire (or forego) privacy protection in exchange for a price (ENISA, 2012). This paradox may lead to an under-valuation of privacy harms, if the harms are measured using traditional willingness to pay tests.

    104In some cases privacy violations can lead to measurable harm, such as when a company loses credit card records. A loss of credit card information requires banks and consumers to take steps to avoid fraud. Those steps create costs that can be measured. The receipt of unwanted spam also creates harm that can be quantified, as does the loss of data that might facilitate identity theft. Even if an actual case of identity theft cannot be traced to a given data breach, the data breach increases the probability of identity theft, and that probability can be estimated. Moreover, the increased risk of identity theft may require that consumers take preventive action to address the increased risk, and the cost of those measures can be quantified.

    105The most difficult harms to quantify are those associated with the feeling that certain data practices are “creepy” (Tene and Polonetsky, 2013). Another way of looking at “creepy” data practices is to call them practices that go beyond what a consumer would reasonably expect. “Creepy” data practices may cause consumers to reduce the level of their activity online, thereby creating a social cost similar to the cost created by excessive government surveillance, which would cause people to communicate less (Posner, 1978).

    106In many cases, the data protection harm can be linked to inadequate information provided to the data subject. Lack of information reduces consumer choice, and is a frequent justification for privacy regulation. Solove and Hartzog (2014) examine several cases where the FTC has based its unfairness findings on inadequate information to consumers, including cases involving non-obvious default settings in software.

    107After looking at the costs, and if possible quantifying them, regulators must look at the benefits of the relevant practice. Benefits of a potentially unfair practice are equal to the costs associated with stopping or regulating the practice. This means that regulators must consider two situations: a situation where the practice is unregulated, and a situation where the practice is regulated or prohibited, and compare the two situations. The difference between these two situations is the opportunity cost of the regulation, or put differently, the benefit of no regulation. Like privacy harms, benefits are difficult to quantify. Widespread use of advertising cookies generates increased advertising revenues through targeted advertising, which in turn brings more free services and information to consumers. Goldfarb and Tucker (2011) attempted to measure the effect of the EU cookie regulation on the effectiveness of online advertising. They found that Europe’s opt-in rule for cookies had a significant adverse effect on the online advertising market:

    “First, privacy protection will likely limit the scope of the advertising-supported internet. However, it also crucially suggests that the types of content and service provided on the internet may change. In particular, without the ability to target, website publishers may find it necessary to adjust their content to be more easily monetizable. Rather than focusing on political news, they may focus on travel or parenting news because the target demographic is more obvious. Furthermore, without targeting it may be the case that publishers and advertisers switch to more intentionally disruptive, intrusive, and larger ads.” (Goldfarb and Tucker, 2011, p. 18)

    108Goldfarb and Tucker (2011) also argue that privacy regulation has an effect on innovation, which should be considered in any cost-benefit exercise. Thierer (2013) points out that privacy regulation can also affect other individual liberties, such as freedom of expression. Like harm to innovation, harm to freedom of expression is difficult to quantify. But the existence of these harms should be considered, at least from a qualitative standpoint.

    Fundamental rights and proportionality

    109Both in the United States and in Europe, fundamental rights are routinely balanced in order to adopt socially optimal rules. In Europe, the balancing is done in the context of the so-called proportionality test, which we will examine in this section. Any measure imposed on internet intermediaries in Europe will have to satisfy the proportionality test if the measure affects one or more fundamental rights.

    The three-criteria test of the European Court of Human Rights

    110The European Court of Human Rights has developed three criteria that must be cumulatively satisfied whenever governments introduce measures to limit fundamental rights such as freedom of expression or privacy (Callanan et al., 2009). These three criteria appear in one form or another in all the court decisions relating to limitations of fundamental rights, including technical measures that address content policies such as fighting online copyright infringement. This proportionality test must form part of any methodology used to assess regulatory measures designed to limit access to illegal content on the internet.

    111First criterion: the measure must be provided for in a law that is understandable and has been adopted pursuant to democratic procedures.16 This is the first safeguard for individual rights. When the legislature has adopted a law that specifically allows for a technical measure that could have an impact on a fundamental right, there is a presumption that the measure has been subject to a democratic debate and that the outcome of that debate between elected officials already represents a balance of competing interests and rights. The law must specifically envisage the restrictive measure in question in order to pass this first test. A law that gives broad but unspecified powers to the courts, to the government or to an administrative agency to impose restrictive measures is less likely to pass this first test. Consequently any law adopted to put into place a regulatory framework for fighting illegal content on the internet must explicitly identify the measures that the regulator may apply and how the regulator should apply the measures. In other words, the law must provide criteria to be applied by the decision-maker and a framework for determining when the technical measures are justified. The regulator in charge of applying the measure cannot be given a blank check. The law must be clear, and easy to understand, to avoid the risk of arbitrary application.

    112The first criterion is explained by the European Court of Human Rights in a well-known decision dealing with the blocking of the Google Sites service in Turkey17:

    “In matters affecting fundamental rights it would be contrary to the rule of law, one of the basic principles of a democratic society enshrined in the Convention, for a legal discretion granted to the executive to be expressed in terms of an unfettered power. Consequently, the law must indicate with sufficient clarity the scope of any such discretion and the manner of its exercise.”18

    “[J]udicial review of such a measure, based on a weighing-up of the competing interests at stake and designed to strike a balance between them, is inconceivable without a framework establishing precise and specific rules regarding the application of preventive restrictions on freedom of expression (see RTBF v. Belgium, cited above, § 114).”19

    113Second criterion: the measure must seek to achieve a legitimate objective. This second test will in most cases be satisfied. Restrictive measures adopted in democratic societies generally seek to promote a legitimate objective such as protection of youth, protection of property rights, privacy, cultural diversity or public security. For example, the protection of copyright is recognized as a form of protection of property, which is itself a constitutionally recognized right. Consequently, any measure adopted to limit online copyright infringement will necessarily pursue a legitimate objective and satisfy this second test. Measures intended to protect privacy, to protect children against exploitation, or to protect the public against terrorist attacks, will also satisfy the second test.

    114Third criterion: the measure must be necessary in a democratic society. This test is the most difficult to pass, and goes to the heart of the proportionality review. To satisfy the third test, the measure must be narrowly tailored to achieve the desired objective without affecting more than absolutely necessary other fundamental rights. It is this third test that must form the core of the analysis for any measure taken by an internet intermediary designed to implement a content policy on the internet.

    115These three criteria appear in various forms in court decisions and in international documents dealing with fundamental rights, such as a recent United Nations report on the promotion and protection of the right to freedom of opinion and expression (United Nations, 2011), which summarizes the three-step test as follows:

    1. “The restriction must be provided by law, which is clear and accessible to everyone (principles of predictability and transparency);
    2. The restriction must pursue one of the purposes cited in Article 19(3) of the International Covenant on Civil and Political Rights, i.e. respect of the rights or reputations of others, protection of national security or of public order, or of public health or morals (principal of legitimacy);20
    3. The restriction must be proven as necessary and the least restrictive means required to achieve the purported aim (principles of necessity and proportionality).”

    116The best way to understand how the proportionality test works is to examine a court decision explaining each step in the process.

    117One of the most interesting decisions is that of the British High Court in the BT/TalkTalk case.21 Two British ISPs, BT and TalkTalk, challenged the legality of the UK Digital Economy Act (DEA) on several grounds, including violation of several European directives. What is significant for purposes of this chapter is the claim that the DEA failed the proportionality test. According to BT and TalkTalk, the restrictive measures envisaged by the DEA affected more than necessary other fundamental rights and therefore failed the third branch of the proportionality test. The British government prevailed on this issue: the lower court held that the DEA satisfied the proportionality test. On appeal, BT and TalkTalk did not raise the proportionality argument again. Consequently, the lower court’s analysis remains valid. Below is a short analysis of the lower court’s reasoning on the question of necessity.

    Should a court give deference to lawmakers’ balancing?

    118An important threshold question raised by the court in the BT/TalkTalk case was to what extent the court should second-guess balancing that was already done by the legislature. Should the court approach the question anew, or should the court give deference to the balancing that was already done by lawmakers? This point connects back to the first step in the three-part test, i.e. that the restrictive measure be adopted in a law that has been subject to democratic debate. If the balancing test was done via the compromises adopted through the legislative process, a court should hesitate before second-guessing the outcome of that balancing.

    119In the BT/TalkTalk case, the court found that where the balancing of interests relates to broad social values and benefits, the court should defer to the legislature because: (i) the legislature is more accountable to citizens than courts are22, and (ii) the legislature will generally have better access to relevant information on societal balances than the court will.23

    120The court said it would apply closer scrutiny if the relevant measure consisted in restricting a fundamental right without a clear countervailing fundamental right being promoted on the other side. This ties back with the second step in the three-part test, i.e. that the restrictive measure be adopted for the purpose of promoting a legitimate objective. Protecting another fundamental right is a legitimate objective, and therefore satisfies this test.

    121Giving some deference to the legislature does not mean blindly accepting its conclusions. It means instead that where the balancing test is close, the court will accept the legislature’s conclusions and not substitute its own. The level of deference might be lower if the institution doing the balancing was a regulatory authority instead of a legislature. A regulatory authority is less accountable to citizens than are elected members of parliament. Consequently the court’s scrutiny of the regulator’s balancing might be more intense than the scrutiny of the parliament’s balancing. In appeals of regulatory decisions, courts generally give some deference to the technical findings of regulatory authorities, particularly in areas where regulatory authorities have special expertise. However, for questions relating to the balancing of fundamental rights, courts will generally apply their own balancing anew, without giving deference to the conclusions of the regulator.

    Identification of the conflicting rights and interests

    122After determining what level of scrutiny should be applied to the measure (high, medium or low), the court in the BT/TalkTalk case proceeded to explore the content of the balancing test as applied to the DEA. The court listed the relevant rights and interests in competition with each other. The court identified three different kinds of rights and interests at stake in the DEA:

    123First, the rights of content owners to protect their copyright against unlawful activity on the internet;

    124Second, rights of internet intermediaries to enjoy exemptions from liability and freedom of regulatory burdens in connection with their activity;

    125Third, the rights of users to enjoy unrestricted access to information on the internet.

    126The court stressed that the rights to be balanced were all recognized as fundamental rights:

    “…this is not a case where, on the one side, there is a human right, or a fundamental EU freedom, and on the other side the State is seeking to restrict or interfere with that right on the grounds of general utility or welfare.”24

    Balancing the relevant interests

    127Citing the Promusicae judgment of the CJEU25 and case law of the European Court of Human Rights, the British High Court concluded that copyright is an important right of property and that its enjoyment and exploitation is recognized as a fundamental right. The court stated that some fundamental rights suffer no restriction, such as the right to life or the prohibition of torture and inhuman or degrading treatment or punishment. However, the right of freedom of expression, although it is a fundamental pillar of a democratic society, is subject to some restrictions. The right to the protection of privacy may also be restricted using a balancing test.

    128The court then proceeded to balance the various competing interests. To quote the court, a measure designed to limit copyright infringement may restrict fundamental rights:

    “provided that the restrictions in fact correspond to objectives of general interest and do not, taking account of the aim of the restrictions, constitute disproportionate and unacceptable interference, impairing the very substance of the rights guaranteed.”26

    129The court here restates the second and third parts of the three-step test discussed above. The terms “disproportionate and unacceptable” used by the court appear redundant: A restriction that is “disproportionate” would necessarily be “unacceptable.” The phrase “impairing the very substance of the rights guaranteed” provides an example of when a restriction would be disproportionate (and therefore unacceptable): Any restriction that makes the exercise of a fundamental right difficult or impossible, and not just less convenient, would be disproportionate, and therefore unacceptable. Respect for the essence of the right is also reflected in Article 52(1) of the European Charter of Fundamental Rights, which provides that:

    “Any limitation on the exercise of the rights and freedoms recognised by this Charter must be provided for by law and respect the essence of those rights and freedoms. Subject to the principle of proportionality, limitations may be made only if they are necessary and genuinely meet objectives of general interest recognised by the Union or the need to protect the rights and freedoms of others.”

    130The test was also well summarized by the High Court of Ireland, in its June 27, 2012 decision EMI Records v. Data Protection Commissioner:

    “…the nature of the injunction sought; the limitation to and the duration of any monitoring; the breadth or narrowness of scope of any order; the nature of the equipment to be used; the potential for the interference of that equipment with the proper use of the existing systems of the intermediary; the balance of burden as to equipment and personnel and cost; the intrusiveness of any remedy into legitimate privacy and entitlement to communicate; and any potential data protection impingements, together constitute the main factors in a court determining where the proportionality of a remedy to the mischief of the improper use of intellectual property online is to be struck or whether an injunction application is to be refused, despite legal compliance, on discretionary grounds.” 27

    Absolute versus relative proportionality, cost-benefit analysis

    131Portuese (2013) demonstrates that the proportionality principle as applied by European courts is in reality a form of cost-benefit analysis. Portuese argues that the case law of the United States Supreme Court also reflects similar principles, although they are not known by the name “proportionality.” Portuese points out that the European proportionality test includes two aspects: an absolute efficiency test, under which the benefits of the rule should outweigh the costs for all stakeholders involved. In addition, proportionality includes a comparative efficiency test, under which the net benefits derived from the measure should be higher than those from all possible alternative measures. In terms of balancing fundamental rights, Portuese characterizes the cost-benefit analysis as ensuring that the marginal benefit reaped from the increased enjoyment of a fundamental right is greater than the marginal cost of restricting another human right. In addition, the measure that is selected should be the one that generates the highest net social benefits after taking into consideration all other legally and factually possible alternatives (Portuese, 2013).

    132Hickman (2008) distinguishes between overall proportionality and relative proportionality. Overall proportionality is an overall cost-benefit analysis to verify that the costs of the measure in interfering with individual rights are more than offset by benefits to society. Relative proportionality consists of comparing several alternative measures and choosing the one with the highest net social benefit. The focus here is in choosing the alternative with the highest marginal utility, in a Kaldor-Hicks sense. Rivers (2006) argues that the focus on relative proportionality should result in Pareto optimal result, whereas Hickman argues that the optimal choice will be Kaldor-Hicks efficient.

    133Hickman points out that a measure that satisfies the overall proportionality test may not satisfy the relative proportionality test, and vice versa. The two tests should be cumulatively satisfied. Hickman regrets that the current approach to proportionality does not clearly recognize these two tests, resulting in an absence of a clear methodology. According to Hickman (2013), “proportionality can either become the fig leaf for unstructured judicial decision-making or it can become a powerful normative and predictive tool in public law” (p. 716). Hickman refers to the relative proportionality test as the “minimum impairment” or “least injurious means” test (Hickman, 2013, p. 701).

    134The difference between “overall” and “relative” proportionality lies in whether the costs of a given measure include opportunity costs, i.e. the costs associated with the best other alternative. When opportunity costs are taken into account, the two tests yield the same results. To illustrate, imagine two regulatory measures 1 and 2 designed to fight online copyright infringement. Let us assume that the benchmark scenario, the scenario of no regulation, corresponds to a level of protection of copyright of 10, and a level of protection of privacy of 10.

    135Regulatory alternative 1 increases the level of protection of copyright from 10 to 15, but reduces the protection of privacy to 8, yielding a net benefit of 3 (I assume here for simplicity that each unit of copyright protection has the same value as a unit of privacy protection).

    136Regulatory alternative 2 increases the level of copyright protection from 10 to 16, but reduces the protection of privacy to 5, yielding a net benefit of 1.

    137Both regulatory alternatives appear to yield positive net benefits compared to the baseline scenario. Both would appear to satisfy the overall proportionality test. If opportunity costs are considered, however, regulatory alternative 2 would fail the test. The costs of regulatory alternative 2 in that case must include the forgone net benefits flowing from regulatory alternative 1, the best other alternative. The net benefits of alternative 1 are equal to 3. When opportunity costs are included, the costs of regulatory alternative 2 increase from 5 to 8, thereby exceeding the benefit flowing from alternative 2 (+6). This yields a negative result, meaning that alternative 2 would fail the cost-benefit test when opportunity costs are considered.

    138Proportionality is a form of cost-benefit analysis, requiring the selection of the measure that generates the lowest impairment of fundamental rights while permitting a reasonable level of enforcement of the desired policy objective. Chapter 6 proposes a system to include proportionality in the overall methodology used by policymakers when evaluating possible measures affecting internet intermediaries.

    Proportionality and the “least injurious means” test

    139The “least injurious means” test was articulated in the European Court of Justice’s Queen v. Ministry of Agriculture case:

    “The Court has consistently held that the principle of proportionality is one of the general principles of Community law. By virtue of that principle, the lawfulness of the prohibition of an … activity is subject to the condition that the prohibitory measures are appropriate and necessary in order to achieve the objectives legitimately pursued by the legislation in question; when there is a choice between several appropriate measures, recourse must be had to the least onerous, and the disadvantages caused must not be disproportionate to the aims pursued.”28

    140Tranberg (2011) analyses several cases of the European Court of Justice, concluding that the court’s rule of proportionality requires national authorities to consider several alternatives and select the one that has the lowest adverse impact on fundamental rights while still attaining the desired objective. This methodology is strikingly similar to the one used in regulatory impact assessments examined in Chapter 5. For example, in the Hüber case29, the ECJ considered German legislation requiring that data regarding immigrants be held in a centralized database. The purpose of the database was to help protect immigrants against discrimination. However, the database also created risks for privacy, particularly due to its centralized character. In his opinion, Advocate General Maduro indicated that a centralized database would be proportionate only if it is the only effective method of applying the national provisions on migration and residence. If there are less risky alternatives, such as a decentralized database, those alternatives should be used, even if they are marginally less convenient and effective.

    141In the Schecke case30, the CJEU considered EU legislation designed to enhance government transparency by publishing the names of recipients of agricultural subsidies. Here the court found that the European institutions had failed to apply the proportionality test because they did not consider alternative measures that would have achieved the desired objective of transparency with a lower interference with individuals’ rights to privacy. This case shows that proportionality is not only concerned with the outcome, but also the process: authorities must consider several alternatives and evaluate whether less intrusive measures are available that would still attain the desired objective.

    142The “least intrusive means” approach to proportionality could be expressed in the form of an algorithm. The algorithm would identify the measure that has the lowest aggregate level of interference with individual rights while still leading to results that fall within a range of acceptable outcomes. For the algorithm to work, policymakers must first define a range of acceptable outcomes, bearing in mind that a level of 100% enforcement of a content policy will be neither achievable nor desirable as it would come at an inordinate cost in enforcement resources and in terms of individual rights. If policymakers define a range of outcomes going from a perfect outcome (which will generally not be achievable) to second best, third best, and fourth best outcomes, it will then be possible to measure different enforcement measures against these outcomes. If an enforcement measure required to achieve the second best outcome creates large additional costs on fundamental rights compared to a measure that achieves the third best outcome, policymakers should prefer the measure that achieves the third best outcome. On the other end, if the difference in cost is negligible between achieving the second best and third best outcome, then it would be reasonable to adopt the measure that achieves the second best outcome.

    143The difficulty at the outset would be to define the range of acceptable outcomes. When important content policies are at stake, such as protection against child pornography, it may be difficult for policymakers to admit that anything less than perfect protection is acceptable. The reality is, however, that for any form of enforcement, including enforcement measures to prevent serious crimes such as murder, the optimal level of enforcement will not be complete enforcement (Shavell, 1993).

    144When proportionality is viewed as a cost-benefit test (Portuese, 2013) requiring consideration of several alternatives in order to choose the least intrusive (Tranberg, 2011), the proportionality test begins to look like a regulatory impact assessment. Chapter 5 will present the principles governing regulatory impact assessments, and Chapter 6 will attempt to merge the proportionality principle for fundamental rights into a broad impact assessment for measures designed to limit access to harmful content. The main lesson from this Chapter 3 is that proportionality tests for fundamental rights are not necessarily incompatible with better regulation principles and methodology used for conducting cost-benefit analyses and regulatory impact assessments. Harms and benefits to fundamental rights can rarely be quantified in monetary terms, but the relative benefits and harms of different measures can be compared, allowing a regulator to select the measure that creates the lowest relative harm while still achieving the desired objective.

    Robert Alexy’s balancing test

    145Alexy (2012) explains that when one fundamental right is impaired, the impairment must be more than counterbalanced by the benefit derived from satisfying another countervailing right. Alexy created a weighting formula to measure the level of impairment and the level of benefit:

    146Image  

    147Where:

    • i and j relate to two competing rights or principles (Pi, Pj) that must be compared.
    • I is the intensity of the interference with, or the promotion of, the relevant principle Pi or Pj.
    • W is the abstract weight of the relevant principles Pi or Pj.
    • R is the level of reliability of the assumptions leading to I and W.

    148For example, if Pj represents a content policy relating to the protection of children against sexual exploitation, and Pi represents individuals’ right to privacy, the denominator in the equation would attempt to capture:

    • the relative weight of the principle of fighting child pornography (Wj);
    • the expected level of attainment of that principle resulting from the relevant measure (Ij);
    • the uncertainty (Rj) relating to Ij and Wj, in other words the degree of reliability of the empirical assumptions concerning what the measure in question means for the non-realisation of Pi. Put more simply, R is the risk of error.

    149In this formula, the denominator represents the right that is being protected and the numerator is the right that is being interfered with. The numerator would attempt to capture:

    • the relative weight of the principle of protecting privacy (Wi );
    • the expected level of interference with that principle through the relevant measure (Ij );
    • the uncertainty (Ri) relating to Ii and Wj, i.e., the risk of error.

    150R would generally be between 0.5 and 1; W would be 1, 2, 3 or 4; I would be 1, 2, 3 or 4. Alexy suggested using a three-level scale, based on a geometric progression. A “low” level of interference would be given the number 20 (i.e., 1); a “medium” level of interference would be given the number 21 (i.e., 2); a “high” level of interference would be given the number 22 (i.e., 4). Similar values would be given to the weighting variable W, corresponding to the relative importance of the fundamental right. This would permit high levels of interference (I = 4) with important fundamental rights (W = 4) to stand out in the numerator, by creating a product of 16.

    151A low value of Wi,j, and in any event a value less than 1, would be necessary to justify a measure.

    152In the practical examples given by Alexy (2014), the “R” factor rarely comes into play. The essence of Alexy’s formula boils down to attributing a score to the importance of the right in the abstract (W), and then attributing a score to the level of interference with the right (I).

    153In the context of an EU-funded research project called “SURVEILLE,” researchers adapted the Alexy formula and applied it to surveillance measures designed to fight crime and terrorism (Scheinin and Sorell, 2015). The objective of the SURVEILLE project is similar to this book’s: define a standard methodology against which proposed measures can be assessed, particularly in light of their efficacy in attaining the intended objective, and the level of interference with fundamental rights.

    154In the context of the SURVEILLE project, Grazia (2013) examines each fundamental right potentially affected by government surveillance measures, dividing those rights into their key attributes. Grazia then attempts to distinguish between the essence of the right, which may not be interfered with, and the peripheral attributes of the right, which in most cases can be interfered with, provided there is a good reason for doing so.

    155The SURVEILLE approach will be examined in more detail in Chapter 7.

    Nussbaum’s ethical filter

    156Nussbaum (2002) argues that certain basic entitlements cannot be balanced in a cost-benefit analysis:

    “[S]ome costs have a distinctive nature; they are bad in a distinctive way. No citizen should have to bear them.” (Nussbaum, 2002, p. 1036)

    157The definition of the minimum level of entitlements can prove difficult, but conceptually, for each fundamental right, there is a red line that cannot be crossed, regardless of the countervailing benefit. In European proportionality reviews, this concept is reflected in the requirement that interference with fundamental rights must not destroy the “essence” of the right. Nussbaum warns that cost-benefit analyses can lead to morally wrong decisions. One suggestion is to attribute an infinitely high cost (Nussbaum refers to a “tragedy tax”) to certain interferences with fundamental rights, so that any proposals involving unacceptably high interferences are eliminated.

    158In the methodology I propose in Chapter 6, the “tragedy tax” would be reflected in the additional constraints applied after the initial cost-benefit analysis is completed. Any proposals that include a “severe” or “extremely high” interference with a fundamental right would be eliminated, regardless of the level of countervailing benefits.

    Fundamental rights and the Hand formula

    159As we have seen, most fundamental rights are not absolute, and can be balanced against other rights and interests. The balancing is called the “proportionality test” and is applied by courts to regulatory measures that limit access to information on the internet.

    160The proportionality test represents a form of cost-benefit analysis, where the costs and benefits correspond to impacts on fundamental rights. These costs and benefits cannot generally be reduced to monetary values. However, a scoring mechanism or other techniques can be used as a rough substitute. (I will propose methods in Chapter 6.) Under the proportionality test, policymakers must choose the least intrusive means to achieve the desired objective. This involves not only comparing the costs and benefits of the proposed measure, but also taking into account opportunity costs, i.e. the net benefit that is forgone from not choosing the best available alternative.

    161If we compare harm to fundamental rights as a kind of accident that generates monetary damages, then the Hand formula could apply:

    Figure 5: illustrating the Hand formula and the optimal level of accident prevention measures. (Source: Posner, 2011)

    Image

    162PL would represent the costs associated with violation of the fundamental right that is being protected by the relevant policy measure, e.g. the right to privacy. The more units of prevention that are devoted to protecting the right to privacy, the lower the costs associated with privacy-right violations. This is illustrated by the downward slope of PL.

    163B would represent the costs resulting from the regulatory measure designed to protect privacy, e.g. harm to freedom of expression. The more units of prevention that are devoted to protecting privacy, the higher the costs associated with violation of freedom of expression. This is illustrated by the upward slope of B.

    164Under the Hand formula, total accident costs should be minimized, i.e. the sum of B and PL. When applied to fundamental rights, the test leads to a similar result, i.e. the total costs of interference with fundamental rights, B + PL, should be minimized.

    165Conceptually this appears simple. In practice, it will be difficult to reach any consensus on where to place the PL and B curves on the graph. However, keeping the graph in mind will help policymakers avoid the mistake of assuming that measures designed to protect a fundamental right (e.g. privacy) must necessarily be situated on the far right-hand side of the x-axis. Moreover, the use of a scoring mechanism similar to Alexy’s can give a general indication on how the respective curves should be placed, leading to the emergence of one or two alternatives that are not too far from the theoretical optimum.

    Notes de bas de page

    1  Article 20, EU Charter of Fundamental Rights.

    2  Article 2, EU Charter of Fundamental Rights.

    3  Reno v. ACLU, 521 U.S. 844 (1997).

    4  Red Lion Broadcasting v. FCC, 395 U.S. 367 (1969). See also French Constitutional Court decision n° 82-141 DC of July 27, 1982, recital 5.

    5  Yildirim v. Turkey, European Court of Human Rights decision n° 3111/10 of March 18, 2013; see also, French Constitutional Council decision n° 2009-580 DC of June 10, 2009, par. 12.

    6  Reno v. ACLU, 521 U.S. 844 (1997).

    7  Scarlet Extended v. SABAM, ECJ, Case C-70/10, 24 November 2011.

    8  Directive 2000/31/EC.

    9  183 F.2d 201 (2d Cir. 1950), aff’d, 341 U.S. 494 (1951).

    10  French Law n° 2014-1353 of November 13, 2014.

    11  See Section 5 below.

    12  Impact assessment (Etude d’impact) relating to the proposed law strengthening the fight against terrorism (projet de loi renforçant les dispositions relatives à la lutte contre le terrorisme) NOR : INTX1414166L/Bleue-1, July 8, 2014.

    13  15 U.S.C. 45.

    14  15 U.S.C. §45(n).

    15  In re Sony BMG Music Entertainment, FTC complaint n° C-4195, 28 June 2007.

    16  Opinion of the Advocate General in the ECJ, Scarlet v. SABAM case n° 70/10.

    17  Ahmet Yildirim v. Turkey, ECtHR n° 31111/10, December 18, 2012.

    18  Id., at para. 59.

    19  Id., at para. 64.

    20  Article 19 of the United Nations International Covenant on Civil and Political Rights (ICCPR) provides as follows:

    1. Everyone shall have the right to hold opinions without interference.

    2. Everyone shall have the right to freedom of expression; this right shall include freedom to seek, receive and impart information and ideas of all kinds, regardless of frontiers, either orally, in writing or in print, in the form of art, or through any other media of his choice.

    3. The exercise of the rights provided for in paragraph 2 of this article carries with it special duties and responsibilities. It may therefore be subject to certain restrictions, but these shall only be such as are provided by law and are necessary:

    a. For respect of the rights or reputations of others;

    b. for the protection of national security or of public order (ordre public), or of public health or morals.

    21  British Telecommunication and TalkTalk Telecom Group v. Secretary of State for Business, Innovation and Skills, High Court of Justice, Queen’s Bench division [2011] EWHC 1021 (Admin), April 20, 2011.

    22  “First, there is considerable support in the case law for the proposition that the Courts should afford particular deference to elected and accountable decision makers where the decision concerns subject matters that are regarded as within the particular province of the political branches…. ‘greater deference will be due to the democratic powers where the subject matter in hand is peculiarly within their constitutional responsibility.’ ” Id., at para 210.

    23  “Secondly, Parliament struck the challenged balance after a lengthy process of consultation of all interested parties, which took account of the representations made by those parties, and after a voluntary, non-legislative scheme was tried out. That process is likely to have provided the decision maker with an insight and capacity that the court is unlikely to enjoy.” Id, at para. 212.

    24  Id., at para. 215.

    25  Promusicae v. Telefonica, Case C-276/06, CJEC Jan. 29, 2008.

    26  BT/TalkTalk, op. cit., at para. 215.

    27  EMI Records v. Data Protection Commissioner, High Court of Ireland, 2012/167 JR, June 27, 2012, IEHC 264, June 27, 2012, para. 810.

    28  CJEU, The Queen v. Ministry of Agriculture, Fisheries and food, ex parte FEDESA and others, Case C-331/88, paragraph 13, quoted in Tranberg (2011).

    29  European Court of Justice, Hüber, Case C-524/06, Advocate General Opinion, para 16.

    30  European Court of Justice, Volker and Marcus Schecke Eifert, Cases C-92 and C-93/09, November 9, 2010, para. 81.

    Précédent Suivant
    Table des matières

    Le texte seul est utilisable sous licence Licence OpenEdition Books. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.

    Voir plus de livres
    Abécédaire des architectures distribuées

    Abécédaire des architectures distribuées

    Cécile Méadel et Francesca Musiani (dir.)

    2015

    Héros

    Héros

    Action, innovation, interaction dans les organisations et au cinéma

    Olivier Fournout

    2014

    Nouvelles énergies pour la ville du futur

    Nouvelles énergies pour la ville du futur

    Eva Boxenbaum, Brice Laurent et Annalivia Lacoste (dir.)

    2013

    Recharger les véhicules électriques et hybrides

    Recharger les véhicules électriques et hybrides

    Matthieu Glachant, Marie Laure Thibault et Laurent Faucheux

    2013

    « Moi je lui donne 5/5 »

    « Moi je lui donne 5/5 »

    Paradoxes de la critique amateur en ligne

    Dominique Pasquier, Valérie Beaudouin et Tomas Legon

    2014

    Les technologies numériques de santé

    Les technologies numériques de santé

    Examen prospectif et critique

    Valérie Fernandez, Laurent Gille et Thomas Houy

    2015

    Le phénomène « pro-ana »

    Le phénomène « pro-ana »

    Troubles alimentaires et réseaux sociaux

    Antonio A. Casilli et Paola Tubaro

    2016

    Diversifier le recrutement public

    Diversifier le recrutement public

    Le cas des magistrats

    Florence Audier, Maya Bacache-Beauvallet et Éric Mathias

    2016

    Smart(er) Internet Regulation Through Cost-Benefit Analysis

    Smart(er) Internet Regulation Through Cost-Benefit Analysis

    Measuring harms to privacy, freedom of expression, and the internet ecosystem

    Winston J. Maxwell

    2017

    Voir plus de livres
    1 / 9
    Abécédaire des architectures distribuées

    Abécédaire des architectures distribuées

    Cécile Méadel et Francesca Musiani (dir.)

    2015

    Héros

    Héros

    Action, innovation, interaction dans les organisations et au cinéma

    Olivier Fournout

    2014

    Nouvelles énergies pour la ville du futur

    Nouvelles énergies pour la ville du futur

    Eva Boxenbaum, Brice Laurent et Annalivia Lacoste (dir.)

    2013

    Recharger les véhicules électriques et hybrides

    Recharger les véhicules électriques et hybrides

    Matthieu Glachant, Marie Laure Thibault et Laurent Faucheux

    2013

    « Moi je lui donne 5/5 »

    « Moi je lui donne 5/5 »

    Paradoxes de la critique amateur en ligne

    Dominique Pasquier, Valérie Beaudouin et Tomas Legon

    2014

    Les technologies numériques de santé

    Les technologies numériques de santé

    Examen prospectif et critique

    Valérie Fernandez, Laurent Gille et Thomas Houy

    2015

    Le phénomène « pro-ana »

    Le phénomène « pro-ana »

    Troubles alimentaires et réseaux sociaux

    Antonio A. Casilli et Paola Tubaro

    2016

    Diversifier le recrutement public

    Diversifier le recrutement public

    Le cas des magistrats

    Florence Audier, Maya Bacache-Beauvallet et Éric Mathias

    2016

    Smart(er) Internet Regulation Through Cost-Benefit Analysis

    Smart(er) Internet Regulation Through Cost-Benefit Analysis

    Measuring harms to privacy, freedom of expression, and the internet ecosystem

    Winston J. Maxwell

    2017

    Accès ouvert

    Accès ouvert freemium

    ePub

    PDF

    PDF du chapitre

    Suggérer l’acquisition à votre bibliothèque

    Acheter

    Édition imprimée

    Presses des Mines
    • amazon.fr
    • mollat.com
    • leslibraires.fr
    ePub / PDF

    1  Article 20, EU Charter of Fundamental Rights.

    2  Article 2, EU Charter of Fundamental Rights.

    3  Reno v. ACLU, 521 U.S. 844 (1997).

    4  Red Lion Broadcasting v. FCC, 395 U.S. 367 (1969). See also French Constitutional Court decision n° 82-141 DC of July 27, 1982, recital 5.

    5  Yildirim v. Turkey, European Court of Human Rights decision n° 3111/10 of March 18, 2013; see also, French Constitutional Council decision n° 2009-580 DC of June 10, 2009, par. 12.

    6  Reno v. ACLU, 521 U.S. 844 (1997).

    7  Scarlet Extended v. SABAM, ECJ, Case C-70/10, 24 November 2011.

    8  Directive 2000/31/EC.

    9  183 F.2d 201 (2d Cir. 1950), aff’d, 341 U.S. 494 (1951).

    10  French Law n° 2014-1353 of November 13, 2014.

    11  See Section 5 below.

    12  Impact assessment (Etude d’impact) relating to the proposed law strengthening the fight against terrorism (projet de loi renforçant les dispositions relatives à la lutte contre le terrorisme) NOR : INTX1414166L/Bleue-1, July 8, 2014.

    13  15 U.S.C. 45.

    14  15 U.S.C. §45(n).

    15  In re Sony BMG Music Entertainment, FTC complaint n° C-4195, 28 June 2007.

    16  Opinion of the Advocate General in the ECJ, Scarlet v. SABAM case n° 70/10.

    17  Ahmet Yildirim v. Turkey, ECtHR n° 31111/10, December 18, 2012.

    18  Id., at para. 59.

    19  Id., at para. 64.

    20  Article 19 of the United Nations International Covenant on Civil and Political Rights (ICCPR) provides as follows:

    1. Everyone shall have the right to hold opinions without interference.

    2. Everyone shall have the right to freedom of expression; this right shall include freedom to seek, receive and impart information and ideas of all kinds, regardless of frontiers, either orally, in writing or in print, in the form of art, or through any other media of his choice.

    3. The exercise of the rights provided for in paragraph 2 of this article carries with it special duties and responsibilities. It may therefore be subject to certain restrictions, but these shall only be such as are provided by law and are necessary:

    a. For respect of the rights or reputations of others;

    b. for the protection of national security or of public order (ordre public), or of public health or morals.

    21  British Telecommunication and TalkTalk Telecom Group v. Secretary of State for Business, Innovation and Skills, High Court of Justice, Queen’s Bench division [2011] EWHC 1021 (Admin), April 20, 2011.

    22  “First, there is considerable support in the case law for the proposition that the Courts should afford particular deference to elected and accountable decision makers where the decision concerns subject matters that are regarded as within the particular province of the political branches…. ‘greater deference will be due to the democratic powers where the subject matter in hand is peculiarly within their constitutional responsibility.’ ” Id., at para 210.

    23  “Secondly, Parliament struck the challenged balance after a lengthy process of consultation of all interested parties, which took account of the representations made by those parties, and after a voluntary, non-legislative scheme was tried out. That process is likely to have provided the decision maker with an insight and capacity that the court is unlikely to enjoy.” Id, at para. 212.

    24  Id., at para. 215.

    25  Promusicae v. Telefonica, Case C-276/06, CJEC Jan. 29, 2008.

    26  BT/TalkTalk, op. cit., at para. 215.

    27  EMI Records v. Data Protection Commissioner, High Court of Ireland, 2012/167 JR, June 27, 2012, IEHC 264, June 27, 2012, para. 810.

    28  CJEU, The Queen v. Ministry of Agriculture, Fisheries and food, ex parte FEDESA and others, Case C-331/88, paragraph 13, quoted in Tranberg (2011).

    29  European Court of Justice, Hüber, Case C-524/06, Advocate General Opinion, para 16.

    30  European Court of Justice, Volker and Marcus Schecke Eifert, Cases C-92 and C-93/09, November 9, 2010, para. 81.

    Smart(er) Internet Regulation Through Cost-Benefit Analysis

    X Facebook Email

    Smart(er) Internet Regulation Through Cost-Benefit Analysis

    Ce livre est diffusé en accès ouvert freemium. L’accès à la lecture en ligne est disponible. L’accès aux versions PDF et ePub est réservé aux bibliothèques l’ayant acquis. Vous pouvez vous connecter à votre bibliothèque à l’adresse suivante : https://freemium.openedition.org/oebooks

    Suggérer l’acquisition à votre bibliothèque Acheter ce livre aux formats PDF et ePub

    Si vous avez des questions, vous pouvez nous écrire à access[at]openedition.org

    Smart(er) Internet Regulation Through Cost-Benefit Analysis

    Vérifiez si votre bibliothèque a déjà acquis ce livre : authentifiez-vous à OpenEdition Freemium for Books.

    Vous pouvez suggérer à votre bibliothèque d’acquérir un ou plusieurs livres publiés sur OpenEdition Books. N’hésitez pas à lui indiquer nos coordonnées : access[at]openedition.org

    Vous pouvez également nous indiquer, à l’aide du formulaire suivant, les coordonnées de votre bibliothèque afin que nous la contactions pour lui suggérer l’achat de ce livre. Les champs suivis de (*) sont obligatoires.

    Veuillez, s’il vous plaît, remplir tous les champs.

    La syntaxe de l’email est incorrecte.

    Référence numérique du chapitre

    Format

    J. Maxwell, W. (2017). Balancing fundamental rights. In Smart(er) Internet Regulation Through Cost-Benefit Analysis. Paris: Presses des Mines. https://doi.org/10.4000/books.pressesmines.3703
    J. Maxwell, Winston. « Balancing Fundamental Rights ». In Smart(er) Internet Regulation Through Cost-Benefit Analysis. Paris: Presses des Mines, 2017. doi:10.4000/books.pressesmines.3703.
    J. Maxwell, Winston. « Balancing Fundamental Rights ». Smart(er) Internet Regulation Through Cost-Benefit Analysis, Presses des Mines, 2017, https://doi.org/10.4000/books.pressesmines.3703.

    Référence numérique du livre

    Format

    J. Maxwell, W. (2017). Smart(er) Internet Regulation Through Cost-Benefit Analysis. Paris: Presses des Mines. https://doi.org/10.4000/books.pressesmines.3697
    J. Maxwell, Winston. Smart(er) Internet Regulation Through Cost-Benefit Analysis. Paris: Presses des Mines, 2017. doi:10.4000/books.pressesmines.3697.
    J. Maxwell, Winston. Smart(er) Internet Regulation Through Cost-Benefit Analysis. Presses des Mines, 2017, https://doi.org/10.4000/books.pressesmines.3697.
    Compatible avec Zotero Zotero

    1 / 3

    Presses des Mines

    Presses des Mines

    • Plan du site
    • Se connecter

    Suivez-nous

    • Flux RSS

    URL : http://www.pressesdesmines.com

    Email : presses@mines-paristech.fr

    OpenEdition
    • Candidater à OpenEdition Books
    • Connaître le programme OpenEdition Freemium
    • Commander des livres
    • S’abonner à la lettre d’OpenEdition
    • CGU d’OpenEdition Books
    • Accessibilité : partiellement conforme
    • Données personnelles
    • Gestion des cookies
    • Système de signalement