Category Archives: Internet

Fake News, Wikipedia and Blockchain (Truth and Consensus)

(this article was intially writen in French)

We must not lie, we are taught at a very young age, and yet we all do it all the time. Provocatively, Meyer (2011) says that you will lie to your wife in one in ten conversations. And if you’re not married, the risk would be even higher. Yet we seem to live with these little lies (to the point of being destabilized when we find ourselves in front of an overly honest person). But the debates on the “Fake News” reminded us that some lies have a price, to the point of creating doubt and permanent distrust. This plurality of words, and the absence of a reference word, is not unlike the philosophy conveyed by crypto-currencies: instead of a centralised mode of governance (validation, certification), it is a global validation by a network, a consensus, which will prevail. Have we changed our definition of what truth is?

The story of “the truth”

In Athens, during the pompous period of Athenian democracy (around the middle of the fifth century BC), every citizen (ὁ βουλεύμενος, “whoever wants it”) has the right to speak, to formulate a speech. We will talk about iségorie (ἰσηγορία) equality of speech. As is so often the case, this equality is in fact very unequal, since not everyone is equal in eloquence. In Gorgias, Plato opposes the latter to Socrates, questioning the basis and purpose of this discourse, which can be put at the service of the common good, of truth, but also be used for purposes of persuasion. Socrates seeks the truth, and the moral value of the stated opinions, fundamentally opposing philosophy and rhetoric. But for Plato, truth is not everything: the notion of the common good seems to prevail over truth, a lie can be lawful if it is just (in the service of this common good). Greek myths are built on this idea, as Veyne (1983) develops, wondering whether the Greeks believed in their myths. This is not a lie, but fiction, allegory. The word μῦθος certainly gave the word “mythomaniac” (by adding μανια, madness), but this word initially meant the speech, the word, the story. We could almost find the concept of “lying true” proposed by poet Louis Aragon.

By radically opposing Good and Evil, Christianity has changed our conception of truth. It is appropriate to confess, to confess, if one aspires to salvation. Truth becomes sacred (and those who do not live with it will pay with their lives). Ecclesiastical authorities then define what is true, Foucault (1994b) spoke of “pastoral” power. “This form of power is oriented towards salvation (…) it is linked to a production of truth“. The opinion of the people is then guided and controlled by official representatives of the truth (ecclesiastical then state).

However, this power will crumble over time. The multiple words will reappear, and with them, the “Fake News”. Just after the First World War, Marc Bloch noted that “false news is always born of collective representations that pre-date its birth; it is fortuitous only in appearance, or, more precisely, all that is fortuitous in it is the initial incident, absolutely unspecified, that triggers the work of imaginations but this setting in motion takes place only because the imaginations are already prepared and ferment deafly. An event, a bad perception for example that would not go in the direction that the minds of all already lean, could at most form the origin of an individual error, but not of a popular and widespread false news“. We find here an idea translated today under the name of “Fake News”. Recently, some have felt compelled to consider legislation, raising the question of freedom of speech, but also probably a form of mistrust by a certain elite towards a people they think incapable of judgment and discernment. But who can afford to decree what is true and what is not?

Science and Truth

There is a form of truth through authority. Following a (painful) fall while skiing, even if I have vague knowledge of anatomy, I did better to consult a doctor, who has expertise from his studies and practice. If he tells me that I have a fracture, it is not the fact that he claims (with authority) that he is telling the truth, but because it is the conclusion he reaches on the basis of solid reasoning, which other doctors might contest. Not having any skills, I rely on his authority. Just as I trust my lawyer if I need legal advice (but perhaps not my doctor). Faced with an expert, the difficulty is to find what weight, what credit, to give him.

Science, as it is taught, seems to state irrefutable truths. If my pen falls off the table, its speed will increase linearly, until it touches the ground. I would be foolish to refute this reality, which stems from the theory of universal gravitation enunciated by Issac Newton in 1687 (in a much more general form than the fall of a pen, or an apple). But this vision is quite dated: “absolute truth” no longer exists, only agreement within the scientific community prevails. As van de Kerchove (2013) notes, mentioning the work of Karl Popper, facts in science are finally established in the same way as “jury proof” in English law. The main difference is the temporal aspect (a judge may set aside elements considered prescribed by law, even if they are scientifically enlightening) and the fact that justice judges a particular case, whereas science seeks general truths. It is the convergence of beliefs, based on collective discussions bringing a consensus which makes it possible to establish scientific truths.
And scientific knowledge evolves over time, even allowing itself to become blurred! Instead of having a “true” or “false” result, we now have true results with a certain probability. In the 2001 IPCC report, it is described as “likely”, with 2 out of 3 chances, that “human activity has been the main cause of observed warming” since the middle of the 20th century. In 2007, the probability exceeded 90%, reaching 95% (then becoming “extremely probable”) in 2014. We will also remember this episode in July 2012, when CERN announced the discovery of an elementary particle (postulated by Higgs in the 1960s, the famous “Higgs boson”) “with a confidence level of 99.99997%”. Science allows itself doubt, and even better, manages to quantify it with amazing precision.

Because doubt still exists in science. This doubt, sometimes called “Cartesian“, allows science to advance, but also, on occasion, to retreat. This is what Oreskes & Conway (2014) points out, highlighting the strategy of certain industries that have funded research projects denying scientific evidence of the dangerousness of tobacco, DDT (insecticide), the reality of the ozone hole, environmental damage from acid rain… etc.. By confusing, discrediting studies – and the scientists who conducted them – doubt has turned against science, while further strengthening the evidence. By questioning the dangers of smoking, numerous studies have established that on the contrary, no doubt was permitted. But the strategy was to buy time (and in that sense the strategy worked).

Justice and truth

But scientists are not the only ones trying to establish the truth. To resume the debate launched by Dagorn (2018), is the tomato a fruit or a vegetable? The fruit is defined in botany as the result of fertilization of the plant: it is derived from the transformation of the pistil once it is fertilized, and will then carry seeds allowing the plant to reproduce. So the tomato is a fruit. But vegetables have no scientific definition, even if they seem to designate the edible parts of plants. That said, chemists who are interested in cooking seem to have decided, using the flavor of the food. The fruit is a sweet food, while the vegetable would be dirty, bitter, or neutral. But if science does not seem to want to impose truth on the subject, justice has examined the question several times. In the United States, at the end of the 19th century, a tax existed on vegetables, but not on fruits. In 1887, John Nix sued New York Harbour treasurer Edward Hedden for taxing his tomato imports (recounted in Sterbenz (2013)). The United States Supreme Court ruled against him in 1893 (unanimously) explaining that the Customs Act referred to the common meaning of the terms “fruit” and “vegetable”, and not to the technical jargon of botanists. But this American vision opposes that adopted in Europe, since a 1988 directive of the Council of the European Union classified tomatoes in the category of fruits (just like carrots or rhubarb branches, moreover).

Justice feels invested with a mission to seek the truth. This is what Jean Domat said in 1745 when he stated that “the Laws want something judged to pass for truth”. The presumption of truth attached to a judgment – “res judicata pro veritate habetur” – is thus found several times in the Civil Code. It may, however, be objected that the raison d’être of this principle is essentially to avoid the repetition of trials ad infinitum (otherwise remedies would not exist: a court decision may be challenged, but according to a very specific procedure). The search for the truth then takes place during an investigation, which has become a “way of authenticating the truth, of acquiring things that will be considered true, and of transmitting them“, as Foucault (1994a) notes.

If science can’t decide, sometimes justice is still asked to make decisions. This is what the decisions say about whether the tomato is a fruit (or not) but it is also the essence of the precautionary principle, evoked in Charpentier (2016). And the situation becomes complex when laws that go against a scientific truth. Baruch (2013) thus returns in detail to the episode of the adoption of the Boyer law tending to repress the denial of genocides, at the turn of the years 2011 and 2012 (the debate on the “memorial laws”), and above all the contestation aroused by article 4 of the law of 23 February – known as the Mekachera law – with numerous positions on the “positive role” of French colonisation. MPs wanted to impose a truth (by law) that went against scientific historical knowledge. And how can we forget the condemnation (following a trial) of Galileo in 1633, following the publication of the Dialogue on the Two Great Systems of the World, by those who supported geocentrism against heliocentrism (previously established by Nicolas Copernicus, among others). In a letter written to the Grand Duchess Christine of Lorraine – mentioned by Gingras, Keating and Limoges (1999) – Galileo notes that the court ruled, “forgetting in a certain way that the multiplication of discoveries contributes to the progress of research, to the development and strengthening of sciences and not to their weakening or destruction, and at the same time showing itself more attached to their own opinions than to the truth“.

Is the majority always right?

In a sitting court, for an accused to be found guilty, the law requires that a (strong) majority of the members of the court of assizes decide it. The majority then imposes its law. But the vote is taken after discussion, after consensus has been reached. According to Wikipedia, “consensus characterizes the existence among the members of a group of a general (tacit or manifest), positive and unanimous agreement that can allow a decision or action to be taken together without a prior vote or special deliberation”. And Wikipedia is built on this idea of consensus. The first principle is openness and transparency: everything is open, everyone can contribute by submitting any content, and a record is kept of the history of interventions. It is required to provide proof of any assertion, with a clear and searchable reference (we find here a basic principle of scientific publication). Then follows a discussion phase. The importance taken today by Wikipedia shows that this principle works.

This delegation of governance is found in the majority of cryptocurrencies. To explain how Lamport works, Shostak and Pease (1982) proposed the fable of the Byzantine generals, to illustrate this concept of “gentium consensus” in computing. Several armies are ready to attack the same city, but the only way to synchronize the different armies, to determine whether to attack or retreat, is to circulate a messenger on horseback. Each general then mandates a knight, to carry the message “attack” or “retreat”, but there may be traitors among the generals who attack. The idea is to bring out a global validation, to obtain a general vote, a consensus, starting from the fact that dishonest people are less numerous than honest people. And indeed, in this case, coordination is possible. As with the blockchain (and all smart contracts), validation is by consensus.
The majority rule, the basis of many democratic systems, seems obvious to us. But the notion of “argumentum ad populum” reminds us that a proposition is not true because most believe it. If this rule seems natural to us, in a democracy, to associate the greatest number with decision-making in the city, in daily life, we quickly see that it leads to many dead ends. Imagine an airline pilot, faced with bad weather conditions, having to make the decision to land (or not) in an emergency. Should it allow passengers to vote? We imagine that it will be inappropriate, if not ineffective, because what the pilot wants is an informed majority, rather than a simple majority. In reality, as we can see in some important democratic decisions, following the choice of the majority is above all a way to avoid bearing responsibility for a choice.

The search for consensus is complicated, even impossible if we believe the literature on voting mechanisms and Arrow’s impossibility theorem. Arrow (1951) – inspired by Condorcet’s paradox – shows that there is no indisputable social choice process that allows a hierarchy of preferences for a group to be expressed from the aggregation of preferences. Also, when a consensus is reached, not everyone is satisfied with the outcome. Some even refer to a “dictatorship of the majority“. To quote Manin’s argument (1985), “the adherence of the greatest number reflects the superior strength of one argument over the others“, because argumentation and discussion are important, “this process makes the appearance of reasonable results more likely“. But this search for consensus is necessarily imperfect: “the true source of legitimacy therefore remains unanimity; the majority will is not legitimate in itself, it is legitimate because it is decided to confer on it all the attributes of the unanimous will. (…) The majority principle is a simple necessity of fact, without any intelligible link with the principle of legitimacy; it is only a convenient convention“, as Minor (2010) recalls.

A “post-truth” world?

The fact that science proposes vague truths does not facilitate understanding of the world, and the term “post-truth” has been proposed to describe this world where the borderline between lies and truth, honesty and dishonesty, fiction and “non-fiction” is no longer very clear. The open data movement also proposes to introduce a form of transparency, with raw data, so that everyone can decide on a debate. In the United States, several years ago, American citizens were shocked to discover that many politicians had received (sometimes very) large sums of money from private companies. But companies always had the desire to finance politicians in order to secure possible support, if necessary. Between these two contradictory visions, the compromise was to impose transparency: politicians had to keep track of any sum paid by a third party, and this information had to be made public. But what to do with this information? In France, this type of transparency is now mandatory for doctors. The database transparence-santé makes all the information declared by companies on the links of interest they maintain with players in the health sector (including doctors) accessible. Who took the time to go to the site to see if their doctor had a conflict of interest when he prescribed a medication? For this somewhat idyllic vision forgets that “pure” data does not exist (“row data is an oxymoron” to use the title of the book edited by Lisa Gitelman). We suspect that since the law imposes this transparency, practices have changed. Data (and facts) do not exist without narration. In 1936, a publisher asked Georges Orwell to investigate the workers’ condition in northern England, in a mining town during the Depression. In the spring of 1937, The Road to Wigan Pier appeared, often considered a report. Crick (1990) compared the notes taken in Orwell’s diary with the novel, to find out if he was giving raw sensations (facts) or if he had staged again, thus reconstructing his original vision. As he notes, “the bare style of the documentary is in reality a perfectly deliberate artistic creation“. Leys (1984) goes even further in the analysis “what Orwell’s invisible and so effective art illustrates is that’the truth of the facts’ cannot exist in a pure state. Facts by themselves never form anything but meaningless chaos: only artistic creation can invest them with meaning, giving them form and rhythm (…) Literally, truth must be invented“.


Arrow, K. (1951). Social Choice and Individual Values. Wiley.

Baruch, M-O. (2013). Des lois indignes ? Les historiens, la politique et le droit. Tallandier

Bloch, M. (1921) Réflexions d’un historien sur les fausses nouvelles de la guerre. Revue de synthèse historique, 33.

Charpentier, A. (2016). Les dérives du principe de précaution. Risques, 108.

Crick, B. (1990) Georges Orwell: Une Vie. Balland.

Dagorn, G. (2018). Aubergine, tomate, carotte… Savez-vous vraiment distinguer fruits et légumes ?

Domat, J. (1745) Les lois civiles dans leur ordre naturel.

Foucault, M. (1994a) La vérité et les formes juridiques, in « Dits et écrits » tome II texte n°139, Gallimard.

Foucault, M. (1994b) Le sujet et le pouvoir, in « Dits et écrits » tome IV texte n°306, Gallimard.

Gingras, Y., Keating, P. & Limoges, C. (1999) Du scribe au savant : Les Porteurs du savoir de l’Antiquité à la révolution industrielle, Boréal.

Gitelman, L. (2013). Raw Data Is an Oxymoron. MIT Press.

Lamport, L., Shostak, R. & Pease, M. (1982) The Byzantine Generals Problem. ACM Transactions on Programming Languages and Systems, vol. 4, no 3,‎ juillet 1982.

Leys, S. (1984) Orwell, ou, L’horreur de la politique. Plon.

Manin, B. (1985). Volonté générale ou délibération. Le débat.

Meyer, P. (2011). Liespotting.  Saint Martin’s Griffin.

Mineur, D. (2010). Les justifications de la règle de majorité en démocratie moderne. Raisons Politiques, 39, 127—149.

Oreskes, N. & Conway, E.M. (2014). Les marchands de doute : Ou comment une poignée de scientifiques ont masqué la vérité sur des enjeux de société tels que le tabagisme et le réchauffement climatique. Editions le Pommier.

Popper, K. (1973). La logique de la découverte scientifique. Payot.

Sterbenz, C. (2013). The Supreme Court Says The Tomato Is A Vegetable — Not A Fruit. Business Insider,‎ 30 décembre 2013.

van de Kerchove, M. (2013) Vérité judiciaire et para-judiciaire en matière pénale : quelle vérité ?. Droit et Societe, 84, 411-432.

Veyne, P.  (1983) Les Grecs ont-ils cru à leurs mythes ? 
Essai sur l’imagination


Fake News, Wikipedia et Blockchain (Vérité et Consensus)

Il ne faut pas mentir, nous apprend-on tout petit, et pourtant, on le fait tous en permanence. De manière provocatrice, Meyer (2011) affirme que vous allez mentir à votre épouse lors d’une conversation sur dix. Et si vous n’êtes pas marié, le risque serait encore plus élevé. On semble pourtant s’accommoder de ces petits mensonges (au point d’être déstabilisé quand on se retrouve face à une personne trop honnête). Mais les débats sur les « Fake News » nous ont rappelés que certains mensonges ont un prix, au point d’instaurer le doute et la méfiance permanente. Cette pluralité des paroles, et l’absence de parole de référence n’est pas sans rappeler la philosophie véhiculée par les crypto-monnaies : au lieu d’un mode de gouvernance (de validation, de certification) centralisé, c’est une validation globale par un réseau, un consensus, qui fera foi. A-t-on change de définition de ce qu’est la vérité ?

L’histoire de « la vérité »

A Athènes, lors de la période faste de la démocratie athénienne (vers le milieu du Ve siècle avant notre ère), tout citoyen (ὁ βουλεύμενος, « celui qui le veut ») a le droit de prendre la parole, de formuler un discours. On parlera d’iségorie (ἰσηγορία) l’égalité de parole. Comme bien souvent, cette égalité est en fait très inégalitaire, puisque tous ne sont pas égaux en matière d’éloquence. Dans le Gorgias, Platon oppose ce dernier à Socrate, s’interrogeant sur le fondement et la finalité de ce discours, pouvant être mis au service du bien commun, de la vérité, mais aussi être utilise à des fins de persuasion. Socrate cherche la vérité, et la valeur morale des opinions énoncées, opposant fondamentalement philosophie et rhétorique. Mais pour Platon, la vérité n’est pas tout : la notion de bien commun semble primer sur la vérité, un mensonge pouvant être licite s’il est juste (au service de ce bien commun). Les mythes grecs sont d’ailleurs construits sur cette idée, comme le développe Veyne (1983), se demandant si les grecs ont cru en leur mythes. Il ne s’agit pas d’un mensonge, mais de fictions, d’allégorie. Le mot μῦθος a certes donne le mot « mythomane » (en rajoutant μανια, la folie), mais ce mot désignait initialement le discours, la parole, le récit. On pourrait presque retrouver le concept de « mentir vrai » propose par Louis Aragon.

En opposant radicalement le Bien et le Mal, le christianisme a changé notre conception de la vérité. Il convient de se confesser, d’avouer, si on aspire au salut. La vérité devient sacrée (et ceux qui ne s’en accommodent pas le payeront de leur vie). Des autorités ecclésiastiques définissent alors ce qui est vrai, Foucault (1994b) parlera de pouvoir « pastoral ». « Cette forme de pouvoir est orientée vers le salut (…) elle est liée à une production de la vérité ». L’opinion du peuple est alors guidée et contrôlée par des représentants officiels de la vérité (ecclésiastiques puis étatiques).

Ce pouvoir va toutefois s’effriter avec le temps. Les paroles multiples vont resurgir, et avec elles, les « Fake News ». Juste après la première guerre mondiale, Marc Bloch notait qu’ « une fausse nouvelle naît toujours de représentations collectives qui préexistent à sa naissance ; elle n’est fortuite qu’en apparence, ou, plus précisément, tout ce qu’il y a de fortuit en elle c’est l’incident initial, absolument quelconque, qui déclenche le travail des imaginations ; mais cette mise en branle n’a lieu que parce que les imaginations sont déjà préparées et fermentent sourdement. Un événement, une mauvaise perception par exemple qui n’irait pas dans le sens où penchent déjà les esprits de tous, pourrait tout au plus former l’origine d’une erreur individuelle, mais non pas d’une fausse nouvelle populaire et largement répandue ». On retrouve ici une idée traduite aujourd’hui sous le nom de « Fake News ». Récemment, certains se sont senti oblige d’envisager de légiférer, reposant la question de la liberté de parole, mais aussi probablement une forme de défiance d’une certaine élite envers un peuple qu’elle pense incapable de jugement, et de discernement. Mais qui peut se permettre de décréter ce qui est vrai, et ce qui ne l’est pas ?

La science et la vérité

Il existe une forme de vérité par l’autorité. Suite à une chute (douloureuse) au ski, même si j’ai de vagues connaissances en anatomie, j’ai mieux fait de consulter un médecin, qui possède une expertise, venant de ses études, et de sa pratique. S’il me dit que j’ai une fracture, ce n’est pas le fait qu’il prétende (avec autorité) qu’il dit la vérité, mais parce que c’est la conclusion à laquelle il arrive sur la base d’un raisonnement solide, que d’autres médecins pourraient contester. N’ayant pas de compétences, je me fie à son autorité. Tout comme je fais confiance à mon avocat si j’ai besoin de conseils juridiques (mais peut-être pas à mon médecin). Face à un expert, la difficulté est de trouver quel poids, quel crédit, lui accorder.

La science, telle qu’elle est enseignée, semble énoncer des vérités irréfutables. Si mon stylo tombe de la table, sa vitesse va augmenter linéairement, jusqu’à toucher le sol. Je serais fou de réfuter cette réalité, qui découle de la théorie de la gravitation universelle énoncée par Issac Newton en 1687 (sous une forme beaucoup plus générale que la chute d’un stylo, ou d’une pomme). Mais cette vision est passablement datée : la « vérité absolue » n’existe plus, seul l’accord au sein de la communauté scientifique prévaut. Comme le note van de Kerchove (2013), mentionnant les travaux de Karl Popper, les faits en sciences s’établissent finalement de la même manière que la « preuve par le jury » dans le droit anglais. La principale différence est l’aspect temporel (un juge pourra écarter des éléments considères comme prescrits par le droit, même s’ils sont scientifiquement éclairants) et le fait que la justice juge un cas particulier, alors que la science cherche des vérités générales. C’est la convergence des croyances, bases sur des discussions collectives apportant un consensus qui permet d’établir des vérités scientifiques.

Et la connaissance scientifique évolue dans le temps, en s’autorisant même à devenir floue ! Au lieu d’avoir un résultat « vrai » ou « faux » on a aujourd’hui des résultats vrais avec une certaine probabilité. Dans le rapport de 2001 du GIEC, il est qualifié de « probable », avec 2 chances sur 3, que l’« activité humaine est la cause principale du réchauffement observé » depuis le milieu du XXe siècle. En 2007, la probabilité dépassait 90%, pour atteindre 95% (devenant alors « extrêmement probable ») en 2014. On se souviendra aussi de cet épisode en juillet 2012, où le CERN a annoncé la découverte d’une particule élémentaire (postulé par Higgs dans les années 1960, le fameux « boson de Higgs ») « avec un degré de confiance de 99,99997 % ». La science s’autorise le doute, et mieux encore, parvient à le quantifier avec une précision stupéfiante.

Car le doute existe toujours en science. Ce doute, parfois appelé « cartésien », permet à la science d’avancer, mais aussi, à l’occasion, de reculer. C’est ce que rappelle Oreskes & Conway (2014), mettant en avant la stratégie de certaines industries qui ont financé des projets de recherche niant les preuves scientifiques de la dangerosité du tabac, du DDT (l’insecticide), de la réalité du trou de la couche d’ozone, des atteintes environnementales des pluies acides…etc. En semant la confusion, en discréditer des études – et les scientifiques qui les ont menées – le doute s’est retourne contre la science, tout en renforçant encore les preuves. En remettant en cause les dangers du tabac, de nombres études ont permis d’établir qu’au contraire, aucun doute n’était permis. Mais la stratégie était d’acheter du temps (et en ce sens, la stratégie a fonctionné).

La justice et la vérité

Mais les scientifiques ne sont pas les seuls à tenter d’établir la vérité. Pour reprendre le débat lance par Dagorn (2018), la tomate est-elle un fruit ou un légume ? Le fruit est défini en botanique comme le résultat de la fécondation de la plante : il est issu de la transformation du pistil une fois que celui-ci est fécondé, et sera alors porteur des graines permettant à la plante de se reproduire. Donc la tomate est un fruit. Mais le légume n’a pas de définition scientifique, même s’il semble designer les parties comestibles des plantes. Cela dit, les chimistes qui s’intéressent à la cuisine semblent avoir trancher, en utilisant la saveur de l’aliment. Le fruit est un aliment sucré, alors que le légume serait sale, amer, ou neutre. Mais si la science ne semble pas vouloir imposer de vérité sur le sujet, la justice, elle, s’est penchée sur la question a plusieurs reprises. Aux États-Unis, à la fin du XIXème siècle, une taxe sur les légumes existait, mais par sur les fruits. En 1887, John Nix a intenté un procès au trésorier du port de New York, Edward Hedden, qui souhaitait taxer son importation de tomates (raconté dans Sterbenz (2013)). La cour suprême des États-Unis lui donna tort en 1893 (à l’unanimité) expliquant que la loi douanière se référait au sens commun des termes « fruit » et « légume », et non pas au jargon technique des botanistes. Mais cette vision américaine s’oppose à celle adoptée en Europe, puisqu’une directive de 1988 du Conseil de l’Union Européenne a range la tomate dans la catégorie des fruits (tout comme la carotte ou les branches de rhubarbe, d’ailleurs).

La justice se sent investi d’une mission, de recherche de la vérité. C’est ce que disait Jean Domat en 1745 lorsqu’il affirmait que « les Lois veulent qu’une chose jugée passe pour vérité ». La présomption de vérité attachée a un jugement – « res judicata pro veritate habetur » – se retrouve ainsi à plusieurs reprises dans le Code Civil. On pourra toutefois objecter que la raison d’être de ce principe est essentiellement d’éviter le renouvellement de procès à l’infini (sinon les voies de recours n’existeraient pas : on peut contester une décision de justice, mais selon une procédure bien précise). La recherche de la vérité se fait alors lors d’une enquête, devenue une « manière d’authentifier la vérité, d’acquérir des choses qui vont être considérées comme vraies, et de les transmettre », comme le note Foucault (1994a).

Si la science n’arrive pas à trancher, on demande parfois malgré tout à la justice de prendre des décisions. C’est ce que disent les décisions pour savoir si la tomate est un fruit (ou pas) mais c’est aussi l’essence du principe de précaution, évoqué dans Charpentier (2016). Et la situation devient complexe lorsque des lois qui vont à l’encontre d’une vérité scientifique. Baruch (2013) revient ainsi en détails sur l’épisode de l’adoption de la loi Boyer tendant à réprimer la négation des génocides, au tournant des années 2011 et 2012 (le débat sur les « lois mémorielles »), et surtout la contestation suscitée par l’article 4 de la loi du 23 février – dite loi Mekachera – avec de nombreuses prises de position sur le « rôle positif » de la colonisation française. Les députés ont voulu imposer une vérité (par la loi) qui allait à l’encontre de la connaissance historique scientifique. Et comment oublier la condamnation (suite à un procès) de Galilée en 1633, consécutive à la publication du Dialogue sur les deux grands systèmes du monde, par ceux qui soutenaient le géocentrisme contre l’héliocentrisme (établi auparavant par Nicolas Copernic, entre autres). Dans une lettre écrite a la grande-duchesse Christine de Lorraine – mentionnée par Gingras, Keating et Limoges (1999) – Galilée note que le tribunal a tranché, « oubliant d’une certaine manière que la multiplication des découvertes concourt au progrès de la recherche, au développement et à l’affermissement des sciences et non pas à leur affaiblissement ou à leur destruction, et se montrant dans le même temps plus attachés à leurs propres opinions qu’à la vérité ».

La majorité a-t-elle toujours raison ?

Dans une cour d’assise, pour qu’un accusé soit déclaré coupable, la loi exige qu’une (forte) majorité de membres de la cour d’assises le décident. La majorité impose alors sa loi. Mais le vote se fait après discussion, après recherche d’un consensus. Selon Wikipedia, « un consensus caractérise l’existence parmi les membres d’un groupe d’un accord général (tacite ou manifeste), positif et unanime pouvant permettre de prendre une décision ou d’agir ensemble sans vote préalable ou délibération particulière ». Et justement, Wikipedia est construit sur cette idée de consensus. Le premier principe est l’ouverture et la transparence : tout est ouvert, chacun peut contribuer en soumettant n’importe quel contenu, et une trace est garde de l’historique des interventions. Il est demande d’apporter des preuves de toute affirmation, avec une référence claire et consultable (on retrouve ici un principe de base de la publication scientifique). Suit alors une phase de discussion. L’importance prise aujourd’hui par Wikipedia montre que ce principe fonctionne.

Cette délégation de la gouvernance se retrouve dans la majorité des crypto-monnaies. Pour expliquer le fonctionnement Lamport, Shostak et Pease (1982) ont proposé la fable des généraux byzantins, pour illustrer ce concept de « consensus gentium » en informatique. Plusieurs armées sont prêtes à attaquer une même ville, mais le seul moyen pour synchroniser les différentes armées, pour déterminer s’il faut attaquer ou battre en retraite, est de faire circuler un messager à cheval. Chaque général mandate alors un chevalier, pour porter le message « attaque » ou « retraite », mais il peut y avoir des traitres parmi les généraux qui attaquent. L’idée est de faire émerger une validation globale, d’obtenir un vote général, un consensus, en partant du fait que les personnes malhonnêtes sont moins nombreuses que les personnes honnêtes. Et effectivement, dans ce cas, une coordination est possible. Pareil pour la blockchain (et tous les « smart contracts »), la validation se fait par consensus.

La règle de la majorité, base de nombre de système démocratique, nous semble comme une évidence. Mais la notion d’« argumentum ad populum »  nous rappelle toutefois que qu’une proposition n’est pas vraie parce que la plupart y croit. Si cette règle nous semble naturelle, en démocratie, pour associer le plus grand nombre à la prise de décisions dans la cité, dans la vie quotidienne, on voit rapidement qu’elle aboutit à de nombreuses impasses. Imaginons un pilote de ligne, face à de mauvaises conditions atmosphériques, devant prendre la décision d’atterrir (ou pas) en urgence. Doit-il permettre aux passagers de voter ? On imagine que ça sera inapproprié, pour ne pas dire inefficace, car ce que souhaite le pilote, ça serait plutôt une majorité informée, plus qu’une simple majorité. En réalité, comme on peut le voir dans certaines décisions importantes en démocratie, suivre le choix de la majorité est surtout un moyen d’éviter de porter la responsabilité d’un choix[ii].

La recherche d’un consensus est compliquée, voire impossible si on en croit la littérature sur les mécanismes de vote, et le théorème d’impossibilité de Arrow. Arrow (1951) – s’inspirant du paradoxe de Condorcet – montrer qu’il n’existe pas de processus de choix social indiscutable, qui permette d’exprimer une hiérarchie des préférences pour un groupe à partir de l’agrégation des préférences. Aussi, lors d’un consensus, tout le monde n’est pas satisfait du résultat. Certains évoquent même une « dictature de la majorité ». Pour reprendre l’argumentation de Manin (1985), « l’adhésion du plus grand nombre reflète la force supérieure d’une argumentation par rapport aux autres », car l’argumentation et la discussion sont importantes, « ce processus rend plus probable l’apparition de résultats raisonnables ». Mais cette recherche de consensus est forcément imparfaite : « la véritable source de la légitimité demeure donc l’unanimité ; la volonté majoritaire n’est pas légitime par elle-même, elle est légitimée parce qu’on décide de lui conférer tous les attributs de la volonté unanime. (…) Le principe majoritaire est une simple nécessité de fait, sans lien intelligible avec le principe de légitimité ; il n’est qu’une convention commode », comme le rappelle Mineur (2010).

Un monde « post-truth » ?

Le fait que la science propose des vérités floues[iii] ne facilite pas la compréhension du monde, et le terme « post-truth » a été proposé pour décrire ce monde où la frontière entre mensonge et vérité, honnêteté et déshonnêteté, fiction et « non-fiction » n’est plus très nette. Le mouvement pour les données ouvertes – ou « open data » – propose d’ailleurs d’instaurer une forme de transparence, avec des données brutes, pour que tout le monde puisse trancher sur un débat. Aux États-Unis, il y a plusieurs années, les citoyens américains ont été choqué de découvrir que nombres d’hommes politiques avaient perçu des sommes (parfois très) importantes de la part d’entreprises privées. Mais les entreprises avaient toujours le désir de financer des hommes politiques afin de s’assurer un éventuel soutien, si besoin. Entre ces deux visions antinomiques, le compromis a été d’imposer une transparence : les hommes politiques devaient garder une trace de toute somme versée par un tiers, et cette information devait être rendu publique. Mais que faire de cette information ? En France, ce type de transparence est aujourd’hui obligatoire pour les médecins. La base transparence-santé rend accessible l’ensemble des informations déclarées par les entreprises sur les liens d’intérêts qu’elles entretiennent avec les acteurs du secteur de la santé (dont les médecins). Qui a pris le temps d’aller sur le site pour regarder si son médecin avait un conflit d’intérêt lorsqu’il a prescrit un médicament ? Car cette vision quelque peu idyllique oublie que les données « pures » n’existent pas (« raw data is an oxymoron » pour reprendre le titre du livre édité par Lisa Gitelman). On se doute que depuis que la loi impose cette transparence, les pratiques ont changée. Les données (et les faits) n’existent pas sans narration. En 1936, un éditeur demande à Georges Orwell une enquête sur la condition ouvrière dans le nord de l’Angleterre, dans une cité minière en pleine période de Dépression. Au printemps 1937 paraîtra The Road to Wigan Pier, souvent considéré comme un reportage. Crick (1990) a comparé les notes prises dans le journal d’Orwell et le roman, pour savoir s’il donnait des sensations brutes (des faits) ou s’il avait remis en scène, reconstruisant ainsi sa vision première. Comme il le note « le style dépouillé du documentaire est en réalité une création artistique parfaitement délibérée ». Leys (1984) va encore plus loin dans l’analyse « ce que l’art invisible et si efficace d’Orwell illustre, c’est que la ‘vérité des faits’ ne saurait exister à l’état pur. Les faits par eux-mêmes ne forment jamais qu’un chaos dénué de sens : seule la création artistique peut les investir de signification, en leur conférant forme et rythme (….) Littéralement, il faut inventer la vérité ».


Arrow, K. (1951). Social Choice and Individual Values. Wiley.

Baruch, M-O. (2013). Des lois indignes ? Les historiens, la politique et le droit. Tallandier

Bloch, M. (1921) Réflexions d’un historien sur les fausses nouvelles de la guerre. Revue de synthèse historique, 33.

Charpentier, A. (2016). Les dérives du principe de précaution. Risques, 108.

Crick, B. (1990) Georges Orwell: Une Vie. Balland.

Dagorn, G. (2018). Aubergine, tomate, carotte… Savez-vous vraiment distinguer fruits et légumes ?

Domat, J. (1745) Les lois civiles dans leur ordre naturel.

Foucault, M. (1994a) La vérité et les formes juridiques, in « Dits et écrits » tome II texte n°139, Gallimard.

Foucault, M. (1994b) Le sujet et le pouvoir, in « Dits et écrits » tome IV texte n°306, Gallimard.

Gingras, Y., Keating, P. & Limoges, C. (1999) Du scribe au savant : Les Porteurs du savoir de l’Antiquité à la révolution industrielle, Boréal.

Gitelman, L. (2013). Raw Data Is an Oxymoron. MIT Press.

Lamport, L., Shostak, R. & Pease, M. (1982) The Byzantine Generals Problem. ACM Transactions on Programming Languages and Systems, vol. 4, no 3,‎ juillet 1982.

Leys, S. (1984) Orwell, ou, L’horreur de la politique. Plon.

Manin, B. (1985). Volonté générale ou délibération. Le débat.

Meyer, P. (2011). Liespotting.  Saint Martin’s Griffin.

Mineur, D. (2010). Les justifications de la règle de majorité en démocratie moderne. Raisons Politiques, 39, 127—149.

Oreskes, N. & Conway, E.M. (2014). Les marchands de doute : Ou comment une poignée de scientifiques ont masqué la vérité sur des enjeux de société tels que le tabagisme et le réchauffement climatique. Editions le Pommier.

Popper, K. (1973). La logique de la découverte scientifique. Payot.

Sterbenz, C. (2013). The Supreme Court Says The Tomato Is A Vegetable — Not A Fruit. Business Insider,‎ 30 décembre 2013.

van de Kerchove, M. (2013) Vérité judiciaire et para-judiciaire en matière pénale : quelle vérité ?. Droit et Societe, 84, 411-432.

Veyne, P.  (1983) Les Grecs ont-ils cru à leurs mythes ? 
Essai sur l’imagination


[i] Publiée en ligne dans (historiquement, cette lettre permet aussi de mieux comprendre les rapports entre sciences et religions, Galilée tentant d’expliquer que l’héliocentrisme n’est pas contraire aux théories de Saint Augustin).

[ii] Parmi les exemples récents, on pourra penser aux turbulences entre 1965 et 2018 autour de la construction de Notre Dame des Landes, et la difficulté de prendre des décisions (avec une enquête publique en 2006 refusant l’aéroport a 67% puis le referendum départemental de 2016 où le refus est cette fois de 45%)

[iii] Au sens de la logique floue (“fuzzy logic” en anglais) où au lieu d’avoir des opérateurs « vrai » (1) et « faux » (0), on a des valeurs réelles dans l’intervalle [0,1].

The ethics of modelling in a world where normality no longer exists

(this article was originaly writen in French – part one and two – and published in Risques)

The mechanism for covering natural disasters, in France, was created to compensate “direct uninsurable material damage caused by the abnormal intensity of a natural agent” (article L. 125-1 paragraph 3 of the Insurance Code). Still on the legal level, the Court of Cassation formulated, in November 1986, a principle according to which “no one must cause others an abnormal neighbourhood disturbance”. And in order to be entitled to compensation following pre-trial detention, it is necessary for the plaintiff to prove that the detention caused him “manifestly abnormal and particularly serious harm” (Article 149 of the Code of Civil Procedure). But what does this “abnormality” in all these articles mean? According to the dictionary, abnormality is defined as “contrary to the usual order of things” (one could see there an empirical, statistical notion), “contrary to the just order of things” (this notion of “just” probably calls for a normative definition) but also “not in conformity with the model”. Defining a standard is already not simple if we are only interested in the descriptive, empirical aspect, as actuaries can do when they construct a model (especially in large dimensions, where, as we shall see, normality no longer exists), but if we also integrate a dimension of justice and ethics, we wonder if the task is not impossible…

The average man from Quetelet and Galton

In the 19th century, if several astronomers measured the speed of the same celestial object, they obtained (often) several different measurements. In order to know which one to use in their calculations, the idea of using the “averages method” was quickly imposed – as Stahl[2006] recalls, and especially Sheynin[1973] – this average having a greater precision than any other quantity (or would now say statistically). From a set of observations \{x_1,...,x_n\}, we set
\bar x=\frac{(x_1+⋯+x_n)}{n}=\frac{1}{n}\sum_{i=1}^nx_iWe can note that this size is also solution of the optimization problem
\bar x=\text{argmin}\lbrace\sum_{i=1}^n(x_i-m)^2\rbracewhich shows the importance of “least squares”. Adolphe Quételet was, it seems, the first to apply this calculation of averages to human measures, introducing his famous concept of the “average man”. If we define the mean using a quadratic error minimization, we have an interpretation in terms of forecast: the mean size is the size that a randomly drawn person should measure (up to a random – and unpredictable – variation). In 1846, in a letter Adolphe Quételet used the image of the gladiator statue to explain what the average man might be:

Suppose a thousand statues were used to copy the gladiator with all the care imaginable. Your Highness certainly does not think that the thousand copies that will have been made will each reproduce exactly the model, and that by measuring them successively, the thousand measures that I would obtain would be as concordant as if I had taken them all on the statue of the gladiator himself. The first chances of error would be joined by the inaccuracies of the copyists; so that the probable error would perhaps be very great. Despite this, if the copyists have not worked with preconceived ideas, exaggerating or reducing certain proportions according to school prejudices, and if their inaccuracies are only accidental, the thousand measures, grouped in order of magnitude, will still present a remarkable regularity and will follow one another in the order assigned to them by the law of possibility. I see Your Highness smiling; she will no doubt tell me that such assertions will not compromise me, since we will not be willing to try the experiment. And why not? Perhaps I will surprise him by saying that the experience is ready-made. Yes really, more than a thousand copies of a statue have been measured, which I will not guarantee to be that of the gladiator, but which, in any case, is not far from it: these copies were even alive, so that the measures were taken with every possible chance of error: I would add, moreover, that the copies could have been distorted by a host of accidental causes. One must therefore expect, here, to find a very sensitive probable error.

This average man liked a lot at the time, especially within the English eugenicist school, directed by Francis Galton, even if the latter is mainly interested in deviations from this norm (upward deviation and downward deviation). As Bulmer recalls[2004], “the deviations from that average – upwards towards genius, and downwards towards stupidity – must follow the law that governs deviations from all true averages”. Galton’s work was aimed at understanding these deviations. If Florence Nightingale stated that “the average man is God’s will”, Galton was more interested in the hereditary character of the deviation than in the average. But does that mean anything to this average man?

Looking for the “average” person

Rose[2016] presents two examples in her book Tyrany of the NormThe End of Average . The first is drawn from problems encountered by the US military in the 1950s. When designing the cockpits of fighter aircraft, engineers had used the dimensions of more than 4,000 pilots to optimally position the seat relative to the pedals, the joystick, the height of the windscreen, but also the shape of the seat, the helmet, etc. These measurements made it possible to calculate the measurements of the “median” pilot in about ten dimensions. For example, the average pilot size was 179 cm, which allowed the average pilot size to be defined between 175 and 185 cm. While a majority of the pilots were medium in size, none of the 4,000 pilots was “average” in all dimensions. As Daniels[1952] stated, “designing a cockpit for the average pilot was in fact not designing one for any pilot.

The second example is linked to two statues, those of Norma and Normann (historically on display in Cleveland, now in the Harvard Library). The artist Abram Belskie and the obstetrician Robert Latou Dickinson made these statues together in 1943. Their particularity is that no model has been represented. In fact, it was to represent a woman and a man who had the average measurements of the time (from measurements made on thousands of subjects). Once these statues were made, a contest was held to find out who these statues could represent. Several thousand people from Ohio sent their measurements, but none matched those of the statues. Of course, several hundred were the same size. Several hundred had the same chest circumference. But none had all the right measurements. Because as Todd Rose explains, man is not unidimensional: it is on several dimensions that we measure it in several dimensions. And trying to summarize it in a one-dimensional size is far too reductive. This is what he shows in his book on intelligence tests, for example, where the same IQ can be associated with two very different people. It makes no sense to focus on a single indicator when deciding to recruit someone. The concern when working in a multivariate context is that the average loses its meaning. In fact, from a probabilistic point of view, being average can be extraordinary.

The curse of the dimensionality

In fact, this problem is well known to statisticians as the “scourge of the dimension”. Let’s take a simple example: suppose that a quantity of interest follows a normal law N(\mu,\sigma^2), for example weight, height, chest circumference, etc.. One could say that being in the norm is being in an interval [\mu\pm1.5\sigma]. If we have a normal law, this situation occurs in 85% of cases. And the 15% that are not in this range will be seen as “abnormal”. The sizes can be abnormally small, or abnormally large. It is the drawing of figure 1, on the top. We can now look at two dimensions, weight and height, for example. The norm here would be that in both dimensions, we are in the interval [\mu\pm1.5\sigma]. If the quantities are independent, the probability that both quantities are “normal” is 75%, since 0.85^2\sim 0.75

In other words, in dimension two, 75% of the observations are globally normal, and 25% are then abnormal. In dimension 3, we pass to 65 %, that is to say more than a third of abnormal observations (on the bottom on figure 1, the red points being the abnormal points).

Figure 1 Proportion of “average” individuals in dimensions 1, 2 and 3

In dimension five, we go below 50%, in other words, being in the norm in the five dimensions is no longer the case of the majority. And in dimension twenty, those which are normal are rather atypical, with a proportion of the order of 5 %. Thus, in large dimensions, normality is no longer associated with the idea of a majority. This is the problem that actuaries face today when using very large data, in pricing for example: it becomes very difficult to characterize a rate class (by saying what the average insured in that class looks like).

Normality, statistics and standards

From an empirical, descriptive point of view, being within the norm means nothing other than being within the average, not getting too far from that average. We will then tend to define the norm as the frequency of what happens most often, as the attitude most frequently encountered or the preference most regularly expressed. But this normality is not normativity, and “to be in the norm”, to be exemplary, is then a different dimension, which this time no longer relates to a description of reality but to an identification of what it should tend towards. So we move from the register of being to that of being, from “is” to “ought” to use Hume’s terminology[1739]. It is indeed difficult to envisage the model (or normality) without sliding towards the second meaning that can be found in the concept of standard, which in turn has a strictly normative dimension. This vision leads to confusion between norms and laws, even if not all normativity is exhausted by laws. Hume thus notes that, in all moral systems, authors move from statements of fact, that is, statements of the “there is” type, to proposals that include a normative expression, such as “one must”, “one must”. What Hume disputes is the shift from one type of statement to another: for him, these are two types of statements that have nothing to do with each other, and that cannot therefore be logically linked with each other, in particular from an empirical norm to a normative rule. For Hume, an assertion that is not normative cannot give rise to a normative conclusion. Hume’s assertion has given rise to numerous comments and interpretations, particularly because, as it stands, it seems to be an obstacle to any attempt at naturalization of morality – as McIntyre[1959] or Rescher[1990] detail. In this sense, there is a strong distinction between the norm in regularity (normality) and the rule (normativity).

Statistical laws, from micro to macro

The statistical law is about what “is” because it has been observed (for example, “men are taller than dogs”). Human law (divine, or judicial) is what “is” because it has been decreed, and therefore “must be” (“Men are free and equal” or “Man is good”). Finally, the physical law is about what “is” because we can show it (“The planets are attracted to each other”), often within the framework of hypotheses. We see that the three concepts can be linked. For example, Kepler’s law was historically established using observations (and historically fell into the first category), before being demonstrated in the Copernican model (and then moved on to the third). A concept of balance can also be associated with this law, this “norm”. However, as Hilpinen[1971] points out, however, probabilistic laws ask many questions, one need only think of dice throws or expectations: what is meant by “it is normal to wait five minutes for the bus to stop”, or more ethically disturbing, “it is normal for a person remanded in custody to be imprisoned for eighteen months”?

The norm can be seen as a regularity of cases, observed using frequencies (or averages), for example, on the size of individuals, the duration of sleep, in other words the data that constitute the description of individuals. Anthropometric data have thus made it possible to define an average size of individuals in a given population, according to their age; compared to this average size, a difference of 20% more or less determines gigantism or dwarfism. If we think of road accidents, it can be considered abnormal to have a road accident in a given year, at an individual (micro) level, because the majority of drivers do not have an accident. Nevertheless, from the insurer’s (macro) point of view, the norm is that 10% of drivers have an accident. It would therefore be abnormal for no one to have an accident.

Nevertheless, from the insurer’s (macro) point of view, the norm is that 10% of drivers have an accident. It would therefore be abnormal for no one to have an accident. This is the argument found in Durkheim[1897]. From the singular act that is suicide, if it is considered from the point of view of the individual who commits it, Durkheim tries to see it as a social act, then a real regularity, within a given society. From then on, according to Durkheim, suicide became a normal phenomenon. Statistics then make it possible to quantify the tendency to suicide in a given society, as soon as we no longer observe the irregularity that appears in the singularity of an individual story, but a social normality of suicide.

Standard, convention and ethical aspects

If we take an evolutionary view, what is normal is what is most capable of adapting, of responding to needs, of providing a model for the resolution of situations (nature making abnormality disappear), and normality tends towards normativity, and it becomes difficult to distinguish between the two aspects. In fact, David Hume addresses this point in the well-known example of rowers, who get into the same boat to cross a river and row in rhythm (this example is discussed at length in Mackie[1980]). The two rowers gradually adjust their rowing strokes, one in relation to the other, and it is not necessary to obtain an explicit agreement (which would formulate the standard) that they would respect. The law, which consists in imposing a standard can be useful in case of conflict (if one of the rowers refuses to row, or two rowers of very different physical capacities), but very often, it is not necessary to formulate explicitly this standard inherent to their conduct. The external observer will observe a regularity (when the cruising rhythm is reached) that he can model, but this normal observed rhythm is not necessarily imposed by a law. In the case of rowers, we find the notion of balance mentioned previously. To build a model is to extract the signal from the noise (to use Silver’s distinction[2015]), it is to look for a standard, in the statistical sense. But this goes further if a predictive model is constructed, and reality must then conform to the model, as actuaries often hope.

Patrick Blackburn, Maarten de Rijke & Yde Venema, Modal Logic, Cambridge University Press, 2002.

Bulmer M., Francis Galton: Pioneer of Heredity and Biometry. Johns Hopkins University Press, 2004.

Daniels G., “The Average Man”, Air Force Aerospace Medical Research Lab, vol. 53, n° 7, 1952.

Durkheim E., Le suicide, 1897.

Hilpinen R., Deontic Logic: Introductory and Systematic Readings, 1971, Dordrecht, D. Reidel Publishing Company.

Hume D., Traité de la nature humaine. Tome III : de la morale, 1739.

McIntyre D.C., “Hume on ‘Is’ and ‘Ought’”, The Philosophical Review, vol. 68, n° 4, 1959, pp. 451-468, Duke University Press.

Mackie J.L., Hume’s Moral Theory, Routledge & Kegan Paul Books, 1980.

Rescher N., “How Wide Is the Gap Between Facts and Values?”, Philosophy and Phenomenological Research, vol. 50, 1990, pp. 297-319.

Silver N., The Signal and the Noise: Why So Many Predictions Fail – But Some Don’t, Penguin Press, 2015.

Rose T., The End of Average: How We Succeed in a World That Values Sameness, HarperOne, 2016.

Sheynin O., “Mathematical Treatment of Astronomical Observations (A Historical Essay)”. Archive for History of Exact Sciences, vol. 11, 1973, pp. 97-126.

Stahl S., “The Evolution of the Normal Distribution”, Mathematics Magazine, vol. 79, 2006, pp. 96-113.

Excess of precautionary principle

(this article was intially writen in French, and published in Risques)

« Dans le doute, abstiens-toi » (“When in doubt, abstain yourself”) says popular wisdom. The precautionary principle (in German “Vorsorgeprinzip“) arose from the idea that it is appropriate to accept that there is doubt, or (scientific) uncertainty, in the knowledge of risks. A little over 20 years ago, in France, Barnier’s law introduced the precautionary principle into French law for the “risk of serious and irreversible damage to the environment”, commonly known as “environmental risk”. A little bit more than 10 years ago, it was enshrined in the French Constitution, approved by 531 Members, thus expressing a very broad political consensus, probably also social. Today, however, the precautionary principle is mentioned in contexts as diverse as the risk of terrorist acts, but also in civil or criminal law procedures. What are the consequences of this drift in the use of the precautionary principle?

Prudence, prevention, precaution

It is often accepted that precaution distinguishes prevention by the absence of risk identification. Prevention could be associated with protection against identified risks, while the notion of precaution questions possible actions in the face of risks not yet identified. As Ewald, Gollier and de Sadeleer (2009) point out, the precautionary principle in environmental matters involves three imperatives: reducing risks and avoiding emissions even when there are no short-term effects, formulating environmental quality objectives, and defining an ecological approach to environmental management. Therefore, even in the absence of certainty (and whatever the scientific knowledge at the time), the precautionary principle aims at not delaying the adoption of effective and proportionate measures to prevent irreversible damage at an acceptable cost. Other principles can complement the latter, such as the principle of information, or participation, which postulates that every citizen must have access to information relating to his environment.

As Hunyadi (2004) noted, these three notions are closely related, but different,

  • Prudence refers to proven risks, those whose existence is demonstrated or known empirically to such an extent that the frequency of occurrence can be estimated. Probability makes the risk insurable. This category includes alcohol consumption, or playing Russian roulette.
  • Prevention targets proven risks, those whose existence is demonstrated or known empirically without however being able to estimate the frequency of occurrence. Nuclear risk probably falls into this category. Uncertainty is not about the risk, but about its probability of occurrence. The absence of probabilities normally makes the risk uninsurable by the traditional insurance industry.
  • Precaution refers to risks for which neither the magnitude nor the probability of occurrence can be calculated with certainty, based on current knowledge. One example that has been much debated is genetically modified organisms, but we can include all the risks associated with nanotechnologies.

Precaution, prevention and forecasts

In the words of Hohmann (1994), precaution arises from “the will to free oneself from the assimilative approach in order to replace it with an anticipatory approach”. Anticipation and forecasting are then at the core of precaution: the precautionary principle implies anticipating risks, with the dimension of uncertainty that will accompany it. Like insurance. Except that the latter generally assumes that the hazard is beyond the control of agents, whereas precaution is concerned with endogenous risks, which should be anticipated.

And when precaution becomes a legal principle, we can fall into excesses well known by science fiction fans.” I’m placing you under arrest for the future murder of Sarah Marks and Donald Doobin that was to take place today,” police officer John Anderton told the potential murderer in the 1956 Philip K. Dick short story Minority Report. Can one arrest someone preventively, or worse, cautiously, in the name of the precautionary principle?

Yet we are not far from this science fiction scenario when we look at how the fight against terrorism works. In November 2015, as a result of the state of emergency, “administrative” searches were conducted.” The objective is not to be accused of doing nothing when we had information. It is a kind of precautionary principle applied to terrorism,” said a police officer in Libération. Before the summer of 2016, Eric Ciotti, Member of Parliament for Alpes-Maritimesin France, said, with regard to people with “S-type records“, that “these people today must have a precautionary principle, they must be placed in a situation of detention. They can no longer be free, because they constitute a threat” (reported by Le Monde (2016)). Is that really the precautionary principle?

Justice, uncertainty, hazard and precaution

The precautionary principle requires the practice of doubt, which may give the impression of “hyperbolic doubt”, which lawyers sometimes attach to the concept of the inversion of the burden of proof. The adage “guilty until proven innocent” is replacing the age-old legal maxim “innocent until proven guilty”, as van den Belt (2003), or Flückiger (2003), questioning the concept of proof in the test of the precautionary principle, evoked.

Strangely, the words proof and probability derive from the same Latin adjective, probus. Probus ager is the field where seeds germinate and grow. What is probus is then what meets an expectation, produces, nourishes. Applied to man, it will imply goodness, honesty (probity). It is then a positive response to expectation (on the contrary, chance would be associated with the improbable, with a negative response). Probus will give probatio, the proof, later called proba. Approbare then means “prove”, or even “approve”, and the probabilis derivative would mean “approvable”. It is this meaning which seems to have been retained by Cicero when he used probabilia to evoke possible conjectures. What is likely then contains a large psychological part, with an ethical judgment (we will approve), but also a truth judgment (we will prove). As Gaston Bachelard said, “it is through confusion between the psychological domain and the real domain that we incorporate the notion of probability” (Bachelard (1927)). The probability, however complicated the calculation behind the value may be, measures only our expectation.

The right decision in an uncertain world

We are asking the courts to rule on problems. Sometimes to establish a truth that scientists cannot establish (as in the context of environmental risk). But in everyday life, judges make many decisions. And make mistakes. In decision theory, this can be summarized in Table 1. In criminal justice, the judge – or a jury – must decide whether a person is innocent or not. Two types of errors are possible: false negatives (those found innocent) and false positives (those sent to prison).

Table 1: Decision-making mechanism and errors (in French)

In children’s stories, false positives, it’s Pierre who howls at the wolf when there’s nothing (and who ends up tiring everyone, and when the wolf actually arrives, nobody believes it). When it comes to justice, a false positive is an innocent man sent to prison. It is not uncommon for a type of error to be “economically” more interesting. A cost-benefit analysis in marketing, for example, shows that soliciting someone who will refuse a product is generally less expensive than missing a good customer. In other applications, one type of error will be “socially preferable”, as in public health. After a quarantine, we may prefer to convict innocent people than to have a contaminated person who leaves the contained area (and then contaminate the rest of the population). The great difficulty in decision theory is precisely to choose your errors (or rather your error acceptance rates).

The right decision and mistakes

Voltaire wrote in 1747, about Zadig, chosen by the King to become Prime Minister, “it is from him that nations derive this great principle: that it is better to chance to save a guilty than to condemn an innocent”. This principle has long seemed correct (this is the second example in Table 2).

Table 2: Example of decisions rendered 1) Realistic case 2) All innocent 3) All guilty

As Table 2 shows, let us start from a real (or supposedly real) situation, with a judicial system that makes some mistakes: out of the 100 people tried, 10 guilty were acquitted, while 15 innocent people were sent to prison. That is 25% errors overall, even if we can imagine that the two errors are not of the same order. By declaring everyone innocent, 70% mistakes are made, which is obviously too high. But if you find everyone guilty, you make 30% mistakes. To some extent, the risks of the two types of errors are inversely related. Reducing one of them will generally be to the detriment of the other. We must recognize that there is an inevitable compromise in the design of our criminal justice system. In economics, the balance depends on our estimate of the economic (and other) costs associated with each of the two types of error. But in judicial matters, one can imagine that the moral aspect is important.

In a civilized justice system, the risks of error of the first type are minimized to the extent possible, at least that is what is meant by the term “innocent until proven guilty”. There is a price to pay for this cautious and civilized approach, namely that eventually many wrongdoers must be acquitted because of a “lack of sufficient evidence”.

Error and liability

The judge has experience built on a sub-population that is probably not representative of the French population. This selection bias is caused by the fact that the police had to convince a prosecutor to continue the investigations, at least to reach the investigation stage. What could almost legitimize that a judge – or in a way any person gravitating in the world of justice – has such a negative a priori, that in doubt, for fear of committing an error, he made his instinct, built from a biased population. Worse still: if a judge gets into the habit of convicting (based on the fact that there is no smoke without fire), he or she may be convinced that he or she has only been confronted with culprits. Yet it is normal to make mistakes.” Errare humanum est” said Saint Augustine. The danger is that the decision-maker often confuses’fault’ and’error’, and for fear of committing a fault, for which he could be held responsible, he does not dare to decide. The difficulty is admitting her mistakes, and learning from her. For it must be remembered that the error of the second kind is a double error: if there is an innocent man languishing in prison, there is also – probably – a criminal still at large.

Victims at the centre of justice

With the precautionary principle, power makes society the potential victim, and invites us to objectify ourselves as such. After the Second World War, as Rechtmann (2005) showed, psychiatrists’ attention shifted from trauma to victims. And this idea was then imposed on society as a whole: everyone must, in order to exist, express his suffering and arouse compassion. While “victimology” has thus made its entry into the world of psychiatry, at the same time the place of the victim has been strengthened in law. Levy (2004) has shown that today, victims are granted excessive rights, especially when they fall into certain categories (children, victims of sexual violence, acts of terrorism, etc). In this case, the victim’s word is sacredized, the accused’s defense therefore becomes impossible, everyone being unconsciously convinced of his guilt. In La Société des Victimes, Guillaume Erner denounces a new moral order which is being established and confers on the victim an almost sacred status, since it becomes a “secularized version of martyrs and saints”.

Lévy (2004) recalled that psychiatry seems to think that psychological disorders resulting from certain offenses are perfectly compatible with normal personalities. Thus, “it is considered that the absence of any visible sign of trauma not only does not exclude the crime, but in some cases, may constitute an additional indication of its reality. This is Edgar Hoover’s idea, as legend has it: after being tapped, if the wiretaps confirmed suspicions, individuals were classified as “subversive”, whereas if the wiretaps were inconclusive, they were classified as “subversive malicious”.

We are a long way from the time when Voltaire could claim to prefer having a guilty man in nature than having an innocent man in prison, demanding the lowest possible first hope error rate. By placing the victim at the centre of justice, we are now asking for a zero second species error rate. This amounts to denying the presumption of innocence: “guilty until proven innocent”, relying abusively on the sacrosanct “precautionary principle”.

The precautionary principle is also a risk

It is now clear that the precautionary principle has set a new standard for judging responsibility and extended its ethical space. As François Ewald noted, “the one who introduces the risk must foresee it”. By not taking sufficient precaution, in particular abstention, he can be held responsible. In the name of this principle, certain festive events are canceled, as reported by La Voix du Nord and France 3 during the cancellation of (major) sales last September. The precautionary principle frightens decision-makers and imposes an inertia, an unlimited conservative. Perhaps it is time to be more cautious in applying this principle.

Bachelard, G. 1927, Essai sur la connaissance approchée, Vrin.

Erner, G. 2006. La Société des victimes, La Découverte.

Ewald, F. Gollier, C. et de Sadeleer, N. 2009, Le principe de précaution, Que sais-je ?

Flückiger, A. 2003. La preuve juridique à l’épreuve du principe de précaution. Revue Européenne des Sciences Sociales, 41, 107-127.

France 3. 2016. Lille : après la braderie, le semi-marathon et le 10 km annulés

Hunyadi, M. 2004. La logique du raisonnement de précaution. Revue Européenne des Sciences Sociales, 42, 9-33.

La Voix du Nord, 2016. Le principe de précaution appliqué à la lettre à la braderie du centre

Le Monde, 2016. Interner tous les djihadistes présumés « fichés S », le retour d’une proposition inapplicable. 14 juin 2016,

Lévy, T. 2004. Éloge de la barbarie judiciaire, Odile Jacob.

Libération, 2015. Les perquisitions, « un principe de précaution ». 19 novembre 2015,

Rechtmann, R. 2005. « Du traumatisme à la victime », in D. Fassin et P. Bourdelais (dir.), Les Constructions de l’intolérable. Études d’anthropologie et d’histoire sur les frontières de l’espace moral, La Découverte.

Van den Belt, H. 2003. Debating the Precautionary Principle: “Guilty until Proven Innocent” or “Innocent until Proven Guilty”? Plant Physiology, 132(3), 1122–1126.

Voltaire, 1747. Zadig ou la destinée.


Machines, procedures and avoiding responsibility

Some people are trying to make us believe that artificial intelligence is a “revolution”. What if it wasn’t? Can we not simply see the logic of a process that goes back at least fifty years ago? Bureaucracy has pushed us to put in place simple procedures in all areas of everyday life, allowing everyone to avoid any responsibility, to no longer have to think and to be smart. Algorithms are scary, we wonder where the “human” is in these decision-making procedures… What if he had already disappeared long ago?

Continue reading Machines, procedures and avoiding responsibility

Can predictive models be fair?

In Nosedive the first episode of season 3 of the television series Black Mirror, we discover the dystopia of a society governed by a “personal rating”, a score, a score ranging from 0 to 5. In this world, each person rates the others, the best rated having access to better services (priority in services, better rates, better prices, etc). Will this tendency to construct scores in all sorts of fields (historically on credits but today on criminal, even civic aspects in some countries) not lead to a world that would be an endless popularity contest? And how would it be compatible with social justice, a priori desirable?

Credit scores and social networks

A credit score is, from an actuarial point of view, a quantity proportional to the probability of not honouring its commitments as a creditor. It may also be able to pay the due dates for three consecutive months, or just be late. In real life, as always, it’s a little more complicated. In the United States or Great Britain, it is not uncommon for students to go into debt for decades in order to have the opportunity to follow the courses that interest them (even if the motivation is mainly to obtain a degree at the end of the course). But above all, as soon as they reach the age of 18, credit rating companies will monitor all their movements. Often without their knowledge. And if one day, a consumer credit or mortgage is refused, the reasons are never motivated. Is it a delay in paying rent? Forgotten library fines? An unpaid water bill, years old?

Credit rating companies in the United States, but also in China, are beginning to explore the use of social media data to improve credit scores. Can’t counting the number of times a user uses the word “wasted” in what they post online reveal information about debt repayment? This is at least what the American credit analyst FICO claims: “If you look at how many times a person says’wasted’ in their profile, it has some value in predicting whether they’re going to repay their debt (…) It’s not much, but it’s more than zero” (quoted in McLannahan (2015)). In China, peer-to-peer lender Jubao revealed that he was more likely to give “bonuses” to borrowers if they were Facebook friends with celebrities, as Botsman (2017) tells us.

For the moment, credit rating companies still use the data they know well (utility bills and credit cards), but they imagine that a lot of interesting information must be accessible (in one way or another) on social networks. But data are still scarce, and difficult to analyze. What about the sarcastic or humorous component in a tweet using the word “wasted”? As is often the case, the difficulty is that truly relevant data are difficult to obtain. If it is possible to have information on the payment of rent when a tenant goes through an agency, what to do for transactions between two individuals? And if that were possible, how would you handle the roommate case? Not getting credit because a former roommate didn’t pay on time becomes disturbing. All the more so if it is perhaps about a cellular telephone bill claimed abusively by the telephone company, whereas the subscription had been cancelled.

But the big “malus” in the credit score is often the fact of never having had a credit card. One might think that a person who did not need a credit card (and was satisfied with a debit card, allowing him to buy from a merchant, like most bank cards in France) is a prudent person, who does not need credit for daily expenses. But for credit institutions, this person is not reliable because we don’t know him. And it is up to it to prove that it is (we return to the recurrent practice of reversing the burden of proof mentioned in Charpentier(2016)). This is strangely what happens today when you want to enter the American soil without having a Facebook page.

In a world of widespread surveillance

What if credit institutions aren’t the only ones interested in our lives? What would a world be if, in addition to knowing if I pay my bills on time, some people wanted to know about my networks of friends, which newspapers I read, whether I prefer to buy whole milk or semi-skimmed milk? When we visit the Stasi Museum in Berlin, we discover that this world existed, that 1 person out of 63 was an agent (or indicator) of the Stasi (counting the occasional indicators, the proportion can reach one person out of 6). The museum describes a total panoptism, each being observed permanently, as described by Foucault (1975). But doesn’t this nightmare correspond to today’s world of permanent surveillance, more or less consented. Surveillance via cellular phones (geolocation for the most common function, but sometimes also audio recordings without the user’s knowledge by certain applications), via connected objects, but also surveillance cameras coupled with increasingly powerful facial recognition algorithms. At the end of 2017, 170 million cameras were installed in China, and the 300 million mark should be reached by 2020. During an experiment attempted by the BBC[1], it took 7 minutes to find the journalist John Sudworth walking in the streets.

The danger is that you never know who’s in control. More and more private security companies have partnered with governments. Email providers read our messages to detect spam, but also to resell certain information. For example, in the Privacy Policy attached to Gmail’s Terms of Use (Google) we read “Our automated systems analyze your content (including email) to provide you with custom product features, such as (…) custom advertising. Insurers are increasingly considering the installation of GPS boxes in cars, but through external service providers. Beyond the ownership of data (mentioned in Charpentier & Suire (2016)) we can wonder about their resale, and their use. Knowing that someone regularly visits a blood transfusion centre is potentially interesting information, especially coupled with others.

Since 2014, the Chinese government has been working on an evaluation system for its own citizens, scheduled to be implemented in 2020, as Trujillo (2017) tells us. This “social credit system” aims to create a “citizen score” (to use the expression of Galeon & Bergan (2017)), in order to predict and prevent potential dangers, normalizing individual behaviour through panoptic devices (e.g. video surveillance), inducing self-defence and self-control reflexes. As Foucault (1975) said, it is a question of “ensuring that surveillance is permanent in its effects, even if it is discontinuous in its action; that the perfection of power tends to make the topicality of its exercise useless” (even today, it is more and more continuous in its action). Some of these scores are used by police to find out where to patrol to reduce crime, such as PredPol. But when we look more closely, the predictions say, in substance, that the crimes will take place (in majority) in the (historically) most criminogenic areas of the city. The boundary between banality and tautology is narrow. And the real danger is that scores often transform probabilities into near-certainties, and suspicion becomes proof, as Supiot (2015) noted.

Predictive justice and actuarial methods

In June 2010, a report from the Academy of Medicine called for “improving the practice of expert sex offender dangerousness by teaching and disseminating actuarial methods. These “actuarial methods” are quite simply scoring techniques, “profiling” as defined in the European regulation on personal data of 27 April 2016 (RGDP). Angèle Christin was interested in algorithms that estimate the probability of recidivism in the American criminal justice system. As she has shown, these techniques raise many questions, particularly discriminatory biases, the opacity that makes recourse difficult, and especially the difficulty of understanding what is actually calculated. In the State of Virginia, a score between 1 and 10 is returned, an agreement taken over by Compas (Correctional Offender Management Profiling Alternative Sanctions) which also offers a colour code that predicts the risk of violent recidivism. It is then a decision-making tool, a machine that cannot place a person in detention alone (Christin et al. (2015)).

The conclusions of a predictive score depend on two key elements: the model used, and the data. In the majority of cases in the United States, model codes remain opaque (and therefore impossible to attack), and few have seen the data used to calibrate these models. But one can ask oneself if the court decisions are not also relatively opaque? Judges must certainly give reasons for their decisions, which makes them open to criticism and attack, but if the process were so transparent, shouldn’t the outcome of a (human) trial then be more predictable? Finally, the different biases are quite simple to understand. Suppose being rich means having a good lawyer, and having a good lawyer means not having certain convictions. In this case, a wealth variable (the type of vehicle owned for example) will be positively related to not being guilty (convicted coupagle), and will lower the dangerousness score. The other danger in selection biases is that they are sometimes complex to understand, even paradoxical. A classic example is shown in Figure 1. During World War II, engineers and statisticians were asked how to reinforce bombers who were facing enemy fire.

Figure 1: Damaged locations of returned aircraft (source: McGeddon 2016)

Statistician Abraham Wald began collecting data on impacts in the cabin, as reported by Mangel & Samaniego (1984). To everyone’s surprise, he recommended shielding the areas of the aircraft that showed the least damage. Indeed, the aircraft used in the sample had a significant bias: only returned aircraft were taken into account. If they were able to return with holes at the tips of the wings, it is because these parts are sufficiently solid. And since no aircraft returned with holes in the propeller engines, these were the parts that needed to be reinforced.

Another danger is where causal relationships are reversed. What about this doctor who prescribes a powerful neuroleptic to a patient under investigation, lest justice reproach him for not having seen the dangerousness of his patient, and conversely, justice relies on this prescription to prove that the patient is dangerous? A poorly designed algorithm could misunderstand the meaning of causal relationships.

But predictive models in judicial matters are not only on the side of judges. In the event of a road traffic accident, the Badinter Act (of 5 July 1985) provides for a “right to compensation” for any victim of a traffic accident involving a land motor vehicle. When the driver’s insurance company offers compensation, the victim makes a quick cost/benefit analysis to find out if he goes to court. If it does not formally construct a predictive model, it tries to see, from some elements to its knowledge, the costs of asking a judge to decide on the amount of compensation, and its (potential) benefits.

Another important point is that lawyers call these “predictive” models “actuarial” models. The first function of actuaries was to discount, to calculate the value of time. And judicial time often has disastrous consequences. How would a human decision, imperfect, taken after 5 years of procedure be “better” than an automatic decision taken in 15 days by a machine? Many people who have known proceedings of several years, resulting in a dismissal, dream of accelerated procedures. Because “lost time” has a value, actuaries know it well.

What then of this efficiency of algorithmic models? Justice must be effective, but this constraint must not make us forget the central objective, which is to render justice. What happens if efficiency becomes an objective, not to say the main objective? This is the question posed by predictive models: what is the objective that we are trying to maximize? And how is it formulated in a simple way?

Decision support, or justification for decision making?

In the United States, many judges have been accused of motivating a judgment using decision support tools, which leaves some doubt as to the real function of these tools. The original idea was to help. Recently, several systems put in place in the past years have been questioned. For example, in Australia, the STMP (Suspect Targeting Management Plan) proposed to identify whether or not pre-adolescents should be monitored. This model is similar to any actuarial model, i.e. a risk assessment and prediction tool, focusing either on repeat offenders or on those suspected of committing a future crime. However, a recent report showed that its use had “no observable impact on crime prevention” [2]. At the same time in the United States, Compas (Correctional Offender Management Profiling Alternative Sanctions) has been criticized in Dressel & Farid (2017) : “Advocates of these systems argue that data and advanced automated learning make these analyses more accurate and less biased than those of humans. However, we show that the widely used Compas risk assessment software is no more accurate or fair than predictions made by people with little or no criminal justice expertise. By questioning people recruited on the Internet, without legal skills, it was a question of predicting whether or not people would commit another crime within the next two years. Compas was wrong in 34.8% of cases, and Internet users in 33% of cases. That said, one may wonder here what “to be wrong” means. In this case, recidivism is not measured here, but conviction for recidivism. What if the models (or the people) hadn’t been wrong, but the judges, on the other hand, had?

Predict and make mistakes

And if one of the worries did not come in what one asks a predictive tool? To predict is (basically) to establish a probability for a future fact. As was pointed out in a debate on polls and elections, can we say that we are wrong if we announce that an event can happen with a 5% chance, and that it actually happens? To know if a forecasting technique is good, you need to collect a forecast set, and compare them to observations. This is what meteorologists have been doing for about fifteen years, and which has been formalised by Gneiting et al. (2007). Their idea is that a set of predictive distributions is obtained by a \{\hat F_t,\hat F_{t+1},..,\hat F_{t+h}\} model and it is appropriate to compare these distributions to \{y_t,y_{t+1},..,y_{t+h}\} – and not \{\hat y_t,\hat y_{t+1},..,\hat y_{t+h}\}. It is then necessary to introduce a distance between the predictive distributions, and the observations. In a physical system, it is possible to imagine understanding the different causal relationships, and thus to predict. But in human relations (and justice is a perfect example), nothing is as simple, as automatic as the laws of fluid mechanics that make it possible to model meteorological phenomena.


Binet, Jacques-Louis, 2010, La prévention médicale de la récidive chez les délinquants sexuels. Académie de Médecine.

Botsman, Rachel. 2017. Who Can You Trust?: How Technology Brought Us Together – and Why It Could Drive Us Apart. Portfolio Penguin

Charpentier, Arthur & Suire, Raphaël 2016. Données et santé: valeurs, acteurs et santé. Risques, 107

Charpentier, Arthur. 2016. Les dérives du principe de précaution. Risques. 108

Christin, Agnèle, Rosenblat, Alex & Boyd, Danah 2015. Courts and Predictive Algorithms. Datacivilrights

Dressel, Julie & Farid, Hany 2018. The accuracy, fairness, and limits of predicting recidivism. Science Adavances

Foucault, Michel 1975 Surveiller et punir, naissance de la prison. Gallimard

Galeon, Dom & Bergan, Brad 2017. China’s “Social Credit System” Will Rate How Valuable You Are as a Human. Futurism

Gneiting, Balabdaoui & Raftery 2007. Probabilistic forecasts, calibration and sharpness. JRRS-B, 69, 243–268.

Mangel, Marc & Samaniego, Francisco 1984.  Abraham Wald’s work on aircraft survivability , Journal of the American Statistical Association, vol. 79, no 386,‎ 259–267

McLannahan, 2015 Being ‘wasted’ on Facebook may damage your credit score (Octobre 2015, Financial Times)

Supiot, Alain 2015. La gouvernance par les nombres : cours au Collège de France, 2012-2014. Fayard.

Trujillo, Elsa 2017. La Chine met en place un système de notation de ses citoyens pour 2020. Le Figaro, décembre 2017

[1] In « In Your Face: China’s all-seeing state »




Supprimer des données ?

Il y a une dizaine de jours, je participais à la conférence Sciences XXL, à l’INED, où nous parlions des utilisations des données massives en Sciences Sociales, et au même moment, j’ai découvert – en discutant avec plusieurs assureurs – qu’ils envisageaient de supprimer des données, de manière également massive. Quand on veut faire de l’économétrie de la finance, on peut utiliser des données de cours boursiers, qu’on peut trouver en ligne, ou – pour une granularité plus faible – acheter à des fournisseurs de données. Mais quand on fait de l’économétrie de l’assurance, avoir des données est plus compliqué. Dans mon tout premier cours d’assurance dommage, j’enseignais la théorie, faute de données. Depuis, j’ai réussi à récupérer des données. On a mis ces données dans un package R (casdatasets), et mieux: avec des jeux de données (prêtées par des assureurs) on a pu lancer notre troisième pricing game, qui devrait nous permettre de mieux comprendre la dynamique des marchés d’assurance.

Bref, j’ai été particulièrement surpris d’apprendre que des assureurs envisageaient de détruire des données (individuelles). J’avais entendu de tels propos lors qu’avec Arnaud on commencé à travaillé sur le risque de catastrophes naturelles, et qu’on essayait de mieux comprendre l’impact des tempêtes Lothar et Martin, en décembre 2009. En particulier, au commencement de nos recherches, nous voulions avoir un ordre de grandeur du nombre de polices d’assurance (multirisques habitation) dans un portefeuille, ou plutôt le taux de polices touchées, si possible avec une granularité spatiale assez faible. Des assureurs ont pu nous aider à avoir des ordres de grandeur, par département

mais d’autres nous ont dit qu’il serait difficile de retrouver les données, qui remontent à près de 20 ans. Voire impossible. Et c’est là qu’on a commencé à entendre parler de suppression de données personnelles. Et si les données n’ont pas été agrégées au niveau que l’on souhaite, c’est impossible de faire la requête, 20 ans après.

Continue reading Supprimer des données ?

L’esprit critique, c’est tellement old-school

I’m sorry. I did not start it, but I may have contributed to it” disait S.M. Ong au début du mois en évoquant les “fake-news”.

Ces dernières heures, j’ai pas mal échangé sur twitter – ce qui n’est pourtant pas dans mes habitudes – suite à un tweet que j’avais posté, en début de semaine,

J’avais trouvé l’article amusant, vraiment bien écrit, donc je l’ai mentionné sur Twitter. Tout comme je peux mentionner une carte intéressante sur le site du New York Times ou une statistique intéressante sur le site du Guardian. Après à peine une heure, j’ai été surpris du nombre de personnes qui semblaient avoir pris ça au premier degré. Et ça ne s’est pas démenti avec le temps, même si j’ai posté rapidement un tweet rappelant que était un site parodique…

J’ai reçu un tweet cet après-midi où on tentait de m’apprendre que c’était un “hoax”

ce qui m’a interpelé car pour moi, un “hoax” et un article sur un site parodique, ce sont des choses différentes. Puis on m’a dit que je n’avais pas à twitter des choses pareils, des “fakes”, ayant un compte twitter certifié (tout comme Jean-Marc Morandini, ou…. Le Gorafi)

J’avoue que tous ces échanges m’ont interpelé. Surtout dans le contexte des débats sur les “fake news” qui fleurissent ces jours-ci. Mais comme s’interrogeait S.M. Ong, “what’s fake news and what’s satirical fake news these days“, la distinction est-elle simple à faire ?

Hoax et “fake news”

Les “hoax”, parfois traduit “infaux“, est un mot anglais qui désigne un canular créé notamment à des fins malveillantes, comme le rappelle wikipedia. Ces “hoax”, ou “fake news” nous ont fait entrer dans une ère de post-vérité, “post-truth”, désigné mot de l’année 2016 par l’Oxford Dictionary, comme le notait Alison Flood.

Benoît avait une réflexion intéressante à ce sujet

Quand on lit un article sur son téléphone cellulaire, on en voit pas le site, le contexte général, on ne voit plus que l’article. On ne voit pas que le site a tendance à publier des articles satiriques. On ne voit même pas le site sur lequel on est arrivé….

Les sites parodiques (ou satiriques)

Ce qui m’a surpris, suite à mon tweet, c’est que si je mentionne un article publié dans Le Gorafi, personne ne crie au scandale, personne ne me rappelle que “c’est un hoax, attention”. Ici, nous ne sommes pas dans le “hoax”, nous sommes dans l’humour, comme beaucoup de journaux le pratiquent, ainsi que le rappelle Isabel Ermida dans “Language and Humour in the Media” (qui revient d’ailleurs sur l’origine du mot “hoax”). L’exemple historique le plus connu est The Onion, en ligne depuis 20 ans (mais qui existait avant sous format papier, depuis 1988). Elle Hunt mentionne d’ailleurs que The Onion n’est pas un site de “fake news”. En France, la référence depuis 2012 est Le Gorafi. Comme le rappelle le site “Tous les articles relatés ici sont faux (jusqu’à preuve du contraire) et rédigés dans un but humoristique”. Quand on lit “Toulouse: il se fait abattre de 46 balles dans le corps pour avoir demandé un ‘pain au chocolat’”  on se doute que ce n’est pas sérieux (ou sinon, ça mérite réflexion…) ou encore “il parvient enfin à utiliser le théorème de Pythagore dans sa vie professionnelle“. Peut-on parler de “hoax” ?

En 1998, Lyle Zapato avait fait un test auprès de collégiens qui avaient cru, à l’unanimité (certes, il n’y avait que 25 enfants), un article sur internet qui mentionnait une espèce animale malheurement menacée, les “octopus paxarbolis”, c’est à dire des pieuvres qui vivent dans les arbres, raconté dans “Help Save The Endangered Pacific Northwest Tree Octopus From Extinction!“.

Il est étonnant qu’il soit aussi simple de croire vraie une histoire manifestement fausse. Mais comme le disait Jean-Noël, oui, ça fait sérieux, il y a même une photo de train, et de raclette !

Les sites “sérieux”… et l’esprit critique

C’est marrant parce que l’an passé, mon fils avait justement une introduction critique aux médias, au collège. On lui expliquait qu’il y avait des sites “sérieux”. Mais force est de constater que publier dans un journal renommé ne veut rien dire… Par exemple, le même jour que mon tweet sur la raclette, publiait “le français se hisserait au 3e rang des langues les plus parlées au monde

L’article en question parlait d’une “étude” (non citée) mais qui renvoyait vers un article paru dans, ce dernier étant un peu plus prudent en annonçant “le français serait la troisième langue du monde selon une nouvelle étude“. Tout en restant assez flou. Mais le titre de l’article de est complètement faux ! Le Français reste (relativement) peu parlé, mais reste une langue “puissante”, selon un terme qui n’est pas expliqué (il faut chercher un peu pour trouver plus de détails, sur Au même moment, publiait “on s’est aperçu que Google Maps mentait“, repris sur Comme le notait Philippe Reka, dire “on a demandé à des chercheurs chinois, japonais, indiens de faire des tests, et on a pu voir que si vous êtes à Pékin, vous avez une certaine frontière dans l’Himalaya et qu’à Delhi, vous en avez une autre“, c’est un peu fort de café… Il suffit en effet d’aller (depuis son canapé) sur ou Pour reprendre l’animation de Philippe Rivière, on voit assez simplement que les tracés des frontières diffèrent (comme tout ceux qui jouent avec google maps le savent depuis longtemps)

Mais je m’arrête là, revient très régulièrement sur les “vraies news”, qui n’en sont souvent pas…

Comme le rappellait Benjamin Herold le meilleur antidote à la gratuité de l’information en ligne (mais je rajouterais à la course au buzz que l’on retrouve partout, y compris dans les sites – et les journaux – “sérieux”) est une culture de la pensée critique. C’est également ce que rappelle “Evaluating Information: The Cornerstone of Civic Online Reasoning” paru voilà déjà 5 ans.

Je vais continuer à publier sur mon compte twitter des liens vers des informations sérieuses, mais aussi des articles qui m’ont amusés. Tout comme sur mon blog, lorsque je publie “coffee and productivity” montrant que le café augmente la productivité,

ou “cigarette and life expectancy” montrant qu’avec le même raisonnement, fumer fait vivre plus longtemps…

J’ai envie de croire que les personnes qui suivent mon fil twitter ou qui viennent sur mon blog savent faire preuve d’esprit critique. Et apprécient que l’on soit sérieux, mais aussi satirique, à l’occasion, voie que l’on fasse preuve d’humour…  Cela dit, la frontière entre “humour” et “sérieux” est complexe à définir… spurious-correlations c’est sérieux ?

Retour sur 20 ans de numérique et d’Internet

Ce mois-ci, je vais être éditeur d’une série d’articles sur le thème des données, qui seront publiés sur, le blog d’ENSAE Alumni. Pour le premier article, Philippe Tassi propose un retour sur 20 ans de numérique et d’Internet.

Depuis le bouleversement apporté, à la fin du XIXème siècle et varile début du XXème, par l’accès du grand public à l’électricité, aucune autre révolution de même nature n’a eu lieu jusqu’à la mise à disposition commerciale d’Internet, au milieu des années 90, et donc l’entrée de nos civilisations dans le monde numérique.

D’abord, des faits : même si certaines données chiffrées actuelles seront rapidement caduques, elles décrivent aisément la vitesse de diffusion d’internet à la fois en termes d’accession et d’usage. A l’automne 1997, moins de 1 % des foyers vivant sur le territoire français sont connectés à Internet. Cette proportion passe à 4,7 % en 1999, 27,4 % en 2003, 35,5 % en 2005. Dix ans plus tard, en septembre-octobre 2015, 85 % des ménages y ont accès. Et au-delà de l’accès au réseau, sa pratique est devenue massive : 45 millions de français se connectent à Internet au moins une fois par mois, tout écran.

Internet ne s’est pas construit en un jour

La conception d’Internet n’est pas récente. Il a fallu du temps entre l’apparition du socle technologique – Ray Tomlinson, décédé en mars 2016, a inventé l’e-mail en 1971 – et la mise en œuvre de services adaptés au public. Par comparaison, on a su transporter du son via les ondes hertziennes dès les années 1890, alors que la radio, en tant que média, n’existera qu’à partir de 1922-1923. Le transfert d’une image mobile par ces mêmes ondes date des années 30, le média télévision étant officiellement créé en France en 1949.

Internet n’échappe pas à cette règle. Son point de départ est la crise des missiles de Cuba, en octobre 1962, en pleine guerre froide entre Etats-Unis et URSS. Elle révèle au président Kennedy la faiblesse d’un système centralisé. En 1964 apparaît l’idée de réseau décentralisé, moins vulnérable. Une première ébauche est conçue en 1969, dénommé Arpanet (Advanced Research Projects Agency) ; elle relie les universités de Stanford, UCLA, Santa Barbara et Utah. Le courrier électronique existe dès 1971. Les bases techniques des protocoles TCP et TCP/IP datent des années 70. Dans cette même décennie naissent Microsoft (1975) et Apple (1976). En 1983, Arpanet est scindé en Milnet, intégré au réseau militaire américain, et un nouvel Arpanet universitaire, renommé Internet en 1986. 1990 voit l’émergence du protocole http et du langage html, du concept de www. Les créations des futurs acteurs majeurs se multiplient : Yahoo! et Amazon en 1994, Google en 1998, FaceBook en 2004, Twitter en 2006. Fin des années 2000, les voies menant au monde digital se sont diversifiées : à l’historique micro-ordinateur se sont ajoutés les nouveaux écrans : smartphone depuis 2007, tablette depuis 2010, favorisant la mobilité. L’individu devient ATAWAD : Any Time, Any Where, Any Device. En 2015, le ménage français moyen dispose de 6,4 écrans.

[à suivre..]

Monthly Review

A (nonexhaustive) list of writings worth reading, discovered in September.

Continue reading Monthly Review

Weapons of Math Destruction: How Big Data Increases Inequality

Weapons of Math Destruction est un livre passionnant, qui, comme le sous-titre l’indique parle de “big data”. Beaucoup de livres parus ces jours-ci parlent de big data. Le point de vue est celui de Cathy, ancienne professeur de mathématiques (après des études à Berkeley, elle a obtenu un doctorat à Harvard et enseigné au MIT lors de son postdoc), qui a travaillé sur les marchés financiers au moment de la crise de 2008 dans un hedge fund, avant de rejoindre RiskMetrics en 2009, comprenant qu’il était important de modéliser les risques financiers. Après avoir côtoyé le mouvement Occupy Wall Street, elle décide reprendre sa vie en main, en devenant Data Scientist. Elle a ainsi dirigé un des cursus de l’école de journalisme de Columbia tout en étant consultante en Data Science.

Son point de vue est particulièrement intéressant, car elle met en avant le lien entre la gestion des risques et le monde qui s’ouvre, autour du big data. Le premier est que les deux ont attiré massivement l’élite des étudiants, la finance entre 1995 et 2010, le monde de la donnée ensuite. Pour les deux, le “vrai monde” semble déconnecté de ce qu’ils manipulent: “the real world, with all its messiness, sits apart“. Ce qui pose la question centrale du livre, qu’est-ce qui sera l’analogue de la crise du crédit dans le monde de la donnée? Et pour Cathy, la réponse est simple. Dans la dystopie que nous promet le Big Data, les inégalités sont augmenter.

Il y a déjà les inégalités du côté des plateformes, avec des compagnies qui finissent par occuper des positions monopolistiques. Comme le note Cathy, il n’y a rien d’incroyablement intelligent dans le fonctionnement de Google, ou Facebook. Ces compagnies sont énormes parce qu’elles ont été là, au bon moment, mais le danger est leur voracité insatiable. Car à l’heure actuelle, la même entreprise gère nos courriers électroniques, nos recherches (de voyage, de musique, etc), nos réseaux sociaux, et peut ainsi évaluer, mais également contrôler notre monde.

Et au delà des plateformes, ce sont surtout les inégalités au sein de la population qui vont exploser. Cathy O’Neil illustre sa thèse à l’aide d’exemples particulièrement bien choisis. Par exemple la publicité ciblée permet à des publicitaires prédateurs d’atteindre de manière encore plus efficace les personnes les plus vulnérables à l’escroquerie. Les algorithmes utilisés par les ressources humaines créent des discriminations pires que celles qui pouvaient déjà exister. Et c’est exactement le même phénomène que l’on observe avec les police prédictive, où l’égalité devant la loi n’a plus de sens dans la réalité quotidienne.

La logique de ces algorithmes prédictifs est simple: les gagnants vont gagner, et les perdants vont perdre. Mais de manière sournoise, car invisible. Il n’y a pas de volonté délibérée et affichée, il n’y a plus forcément de policier raciste, ou de recruteur misogyne. Les inégalités se créent sans que personne ne s’en rende vraiment compte. Considérons une RH qui veuille définir un score de “succès” pour proposer des promotions, peut être basé sur un historique de succès observés dans les années précédentes. Si les données ne contiennent que peu de femmes (car peu avaient, historiquement, des postes à responsabilité) l’algorithme pénalisera les femmes. Et la spirale va commencer. Il y en aura de moins en moins dans la base, et elles seront de facto davantage pénalisées. Le livre foisonne de petits exemples qui illustrent parfaitement son propos, car ce phénomène existe partout. Les programmes visant à réduire la criminalité font en fait augmenter la criminalité dans les régions les plus pauvres. Si les quartiers pauvres ont plus de crimes, il y a aura davantage de policiers, qui feront alors davantage d’arrestation, etc. Les inégalités sont exacerbées, car les modèles manquent de boucles de rétroactions.

Une fois posés tous ces problèmes, on est en attente de solutions. Et le tableau que dresse Cathy n’est pas très réjouissant. On retrouve le phénomène décrit par la loi de Goodhart, “quand une mesure devient un objectif, elle cesse d’être une mesure”. Les plus malins comprennent la logique de l’algorithme, et finissent par pervertir l’évaluation. C’est le cas des évaluations des élèves dans les écoles américaines, où les instituteurs sont incités (par l’algorithme) à tricher pour aider leurs élèves à avoir des bonnes notes, pour faire remonter l’école dans le classement. Mais il n’est pas surprenant qu’il n’y ait pas de recette miracle, car le problème est profond. Comme le note Cathy O’Neil, “the model is optimized for efficiency and profitability, not for justice or the good of the ‘team.’ This is, of course, the nature of capitalism“.

A la lecture du livre, on comprend la logique dans le parcours de Cathy, sa logique de rejoindre Occupy après avoir vécu de l’intérieur la crise du crédit. On y entend un appel à la conscience des Data Scientist. Leur rappelant que la construction d’un “bon” modèle ne doit pas se faire sur des critères statistiques, mais doit aussi tenir compte d’une notion d’équité. Ce qui rejoint les débats que nous avons au sein de la chaire, lorsque nous essayons de réintégrer la notion de mutualisation lorsque les actuaires font de la segmentation tarifaire. Bref, ce livre est indéniablement le livre qu’il convient de lire en cette rentrée.