Artificial Intelligence and Predictive Justice: The Negation of Da Mihi Factum, Dabo Tibi Ius and the Danger of Confusing Fact with Prediction

Artificial Intelligence and Predictive Justice:
The Negation of Da Mihi Factum, Dabo Tibi Ius and the Danger of Confusing Fact with Prediction

Alessandro Fusillo¹

June 16, 2026

Property and Freedom Journal

Editor’s note: This paper was presented at the event “Coercion, extortion under the guise of prevention” on June 13, 2026. Participants:

• A. Montagner (Raixe Venete), a Venetian independentist organization
• Father S. Visintin, Abbot of the Benedictine Monastery of Teolo
• Avv. A. Fussillo
• G. Vigni, former president of LIFE TV a liberal-libertarian association

 

Abstract: This article examines the philosophical, legal, and theological implications of predictive-justice systems based on artificial intelligence. Starting from the Misesian axiom of human action and moving through the internal Austrian School debate among Hayek, Hülsmann, and Hoppe, it draws a parallel with the Roman-law principle da mihi factum, dabo tibi ius. The analysis integrates the Catholic conception of sin and retributive justice, the distributist critique of the servile state (Belloc, Chesterton), the historical precedent of English vagrancy laws, the historical parallel with Lombrosianism and algorithmic racial profiling, and the question of a criminal law of intentions, algorithmic opacity, and the risk of a technological thought police.

Preface: “Du musst viel, viel genauer wünschen”

In Paul Maar’s popular series of children’s books Das Sams, the protagonist is an odd little creature with red hair and blue dots on its face. Each dot is a wish: the Sams grants them for its adoptive father, the timid Herr Taschenbier. The mechanism is infallible and unforgiving: the Sams realizes exactly what it is asked, to the letter, without interpretation and without common sense. When Taschenbier wishes for a car, one appears in the living room; when he wants to find himself “on a roof”, he ends up on the wrong one. The Sams, patiently, repeats every time: “Du musst viel, viel genauer wünschen” — you must wish much, much more precisely.[1]

Artificial intelligence resembles the Sams to a striking degree. It is a powerful tool, capable of fulfilling complex wishes — but it does so with the same literal fidelity and the same absence of independent judgment. Its usefulness depends entirely on the quality of the request: on who formulates the wish, with what values, and to what ends. An algorithm trained to “reduce crime” will fulfil that wish in the most efficient way its data suggest — and if that data reflects centuries of structural discrimination, the wish will be granted with algorithmic precision at the expense of the most vulnerable.

This article concerns one of the fields in which the algorithmic Sams risks producing the gravest harm: predictive criminal justice. The argument is not that artificial intelligence is inherently malevolent — no more than the Sams is. The argument is that whoever formulates the wish determines the outcome, and that when such a wish is formulated by opaque power systems run by central planners — whether public or private — the consequences fall on the weakest with the precision of a machine that cannot get it wrong, because it does not know how to wünschen.

1. Introduction: the paradox of anticipatory conviction

Predictive justice — that is, the use of machine-learning algorithms to assess the probability that an individual will commit a future offence or reoffend — poses a radical challenge to the logical structure of judgment in Western law. The Roman-law principle da mihi factum, dabo tibi ius expresses that structure in axiomatic form: the judge ascertains a determinate fact and applies the rule to it. There is no judgment without a fact; there is no culpability without an act.[2]

Predictive-justice systems radically invert this structure: there is no fact — there is a probability distribution; there is no individual — there is a statistical class; there is no rule applied — there is an algorithmic output that purports to replace judgment. The most documented case is the COMPAS system (Correctional Offender Management Profiling for Alternative Sanctions), used in several U.S. states to determine bail and sentence length.[3]

Such an inversion is not a correctable technical distortion: it is the logical consequence of a categorial confusion between fact and prediction, between actus and potentia, which has deep roots in the philosophy of action, in moral theology, and in the history of criminal law.

2. The axiom of human action: Mises and the irreducibility of the future

The soundest theoretical starting point is Ludwig von Mises’s axiom of human action: every human being acts, that is, employs scarce means to attain chosen ends, within a context of radical uncertainty about the future.[4] This entails the necessary acceptance of the concepts of cost, end, means, and the cause-and-effect pairing.

Three implications are decisive for our argument. First: action is always future-oriented, but can only be judged a posteriori, on the fact accomplished. Second: the future is radically uncertain, not because knowledge is insufficient, but because human acting is creative — it produces novelty that cannot be deduced from the present state of the world. Third: any attempt to predict an individual’s future action with certainty is categorially impossible, not technically improvable.[5]

Classical law mirrors this ontological structure: da mihi factum is not an arbitrary convention — it is the legal translation of the structure of human action. The rule applies to what has happened, not to what might happen, because only the accomplished fact is ontologically determinate.

3. The problem of knowledge: Hayek and the epistemological limits of calculation

Friedrich August von Hayek delivers the second act of this critique on the epistemological plane. The failure of central planning does not depend on a lack of computational power: it depends on the very nature of human knowledge, which is dispersed, tacit, and contextual.[6]

The knowledge relevant to action cannot be aggregated in a central system because it is embedded in practices, in situated judgments, in individual expectations: it manifests only in the action itself, at the moment it is exercised. No algorithm — however powerful — can capture that knowledge before it manifests.[7]

The parallel with predictive justice is immediate: the system claims to know what an individual will do, but that knowledge does not yet exist — it is by definition pre-factual. Hayek would have recognized in this the categorial error of rationalist constructivism: confusing the model with reality.[8]

4. The radical critique: Hülsmann and Hoppe between property and calculation

The theoretically most original contribution to our argument comes from within the Austrian School itself, from the critique that Jörg Guido Hülsmann and Hans-Hermann Hoppe direct at the Hayekian position. According to these authors, the problem of socialism — and by extension of any system of centralized planning — is not primarily a problem of knowledge, but a problem of institutional structure: without private property there are no prices, and without prices there is no data at all on which to calculate.[9]

Hoppe radicalizes this position ethically: any system that claims to act on grounds other than private property and past individual action commits a logical violation even before an ethical one — it confuses what is with what could be, what the individual has done with what he might do.[10]

The transfer to predictive justice is powerful: the algorithm does not fail merely because it does not know enough, or because it does not know how to use knowledge — it fails because it presupposes a structure in which the future individual fact is already treated as a present datum. It is the dissolution of the ontological distinction between actus and potentia carried out under the authority of mathematical formalization.[11]

5. The theological structure of culpability: sin, will, and retributive justice

5.1 The actus humanus and the conditions of culpability

Catholic moral theology builds judgment upon a structure identical, mutatis mutandis, to the Roman-law principle: there is no sin without actus humanus, that is, without a voluntary act performed by a free and conscious subject. Three elements constitute formal sin: grave matter (the object of the act), full knowledge (awareness), and deliberate consent (freedom of the will).[12]

The Catechism explicitly lists the conditions that diminish or annul culpability: “ignorance, inadvertence, duress, fear, attachments, and other psychological or social factors”.[13]

Thomas Aquinas constructs a doctrine of retributive justice that is a proportionate reaction to a wrong already done, not the prevention of a statistically probable wrong. Punishment makes sense as a response to the fact — as the restoration of an order disturbed by a free and culpable act — not as actuarial risk management.[14]

5.2 The predictive reversal of the structure of culpability

Predictive justice systematically dissolves all three constitutive elements of Catholic culpability. The deliberate act is replaced by a statistical profile; freedom at the moment of the act becomes irrelevant because “future freedom” has already been “calculated”; the proportionality of punishment to the fact gives way to preventive risk management.

This amounts to a pre-moral regression. The subject is not judged for what he has done as a free person, but for what he is — for his statistical nature, his class membership, his socioeconomic origin. This is precisely what the Christian revolution of personal culpability had overcome with respect to ancient naturalistic determinism and tribal collective responsibility.[15]

6. Distributism: property, freedom, and the servile state

6.1 Belloc’s diagnosis

Hilaire Belloc, in The Servile State (1912), constructs one of the most lucid diagnoses of the drift of industrial capitalism: corporatist, mercantilist, and made up of positions of privilege guaranteed by state power to particular entrepreneurs who, for that very reason, are destined for ever more pronounced gigantism.[16] Capitalism, not being in stable equilibrium, tends toward one of two outcomes: socialism (collective property) or the servile state, in which the majority works without property, guaranteed subsistence but deprived of real freedom.[17]

G.K. Chesterton radicalizes the point in The Outline of Sanity (1926): a society in which property is theoretically accessible to all but in fact concentrated in the hands of a few is not a free society. It is a society in which formal freedom masks substantial servitude.[18]

Both draw on Catholic social doctrine: Leo XIII’s Rerum Novarum (1891) and Pius XI’s Quadragesimo Anno (1931) affirm the natural right to widely distributed private property as a condition for the dignity and freedom of the person.[19]

6.2 Algorithmic profiling as the codification of poverty

The connection with predictive justice is tight. If property is concentrated and the social conditions of the propertyless are structurally precarious, a system of algorithmic profiling will inevitably tend to encode poverty as a criminal risk factor. This is not an accidental distortion: it is the mathematical translation of an unjust social structure.

The algorithm does not create class prejudice — it formalizes it, automates it, and confers upon it the authority of scientific objectivity. Belloc’s propertyless man becomes, in the predictive system, the presumed offender by condition, not by act. Socioeconomic profiling is the new status criminis.[20]

7. The historical precedent: enclosures, vagrancy, and transportation

7.1 The structural genesis of English pauperism

The causal sequence begins with the enclosures: the progressive fencing-off of common lands, accelerated by the Enclosure Act of 1773 and by the hundreds of Private Enclosure Acts of the eighteenth and nineteenth centuries, deprived English peasants of access to the common lands that had guaranteed their subsistence for centuries. This was not a natural process: it was a coercive redistribution of property upward, sanctioned by a Parliament dominated by landowners.[21]

The person expelled from the land was not a criminal: he was the victim of a socioeconomic transformation planned from above. Belloc, who reconstructs exactly this sequence in The Servile State, identifies it as the historical origin of the modern servile condition.[22]

7.2 Repressive legislation as a response to a structural problem

The response of English criminal law to the question of vagrancy is one of the most eloquent historical illustrations of status criminis. The Statute of Laborers (1351) was the first attempt to halt the mobility of labourers after the Black Death. The Vagabonds Act (1494) established vagrancy as an autonomous criminal category. The Act for the Punishment of Sturdy Vagabonds and Beggars (1536) prescribed flogging, branding, and temporary enslavement. The Transportation Act of 1718 systematized transportation as the penal response to poverty.[23]

The Vagrancy Act of 1824, which remained in force until the formal repeal provided for by section 81 of the Police, Crime, Sentencing and Courts Act of 2022 (entry into force conditional on the adoption of replacement legislation), is proof of the exceptional longevity of this mechanism.[24]

7.3 Transportation as predictive justice ante litteram

The logic of transportation is exactly that of predictive justice applied to a social class: I do not condemn you for what you have done, I remove you for what you are and for the risk you represent to the social order. This logical structure is identical to that of contemporary predictive justice. The difference is purely technical: in place of the magistrate who empirically assessed the condition of the vagrant, we have an algorithm that assigns a risk score on a statistical basis.[25]

8. The criminal law of intentions and algorithmic morality

8.1 Thoughts, words, deeds, and omissions: the map of culpability

The formula of the Confiteor — thoughts, words, deeds, and omissions — is no accident: it mirrors the complete topography of moral agency. Catholic moral theology punishes all four dimensions because divine judgment is omniscient and knows intention. Inner sin (of thought or desire) is real sin — as the Sermon on the Mount teaches (“whoever looks at a woman with lust has already committed adultery with her in his heart”, Mt 5:28). Human criminal law, by contrast, performs a deliberate structural reduction: it punishes only what is externally verifiable and causally relevant.[26]

Classical criminal law reduces culpability to only two elements of the liturgical formula: deeds (the constitutive fact, a necessary foundation) and — within strictly defined limits — omissions in positions of legal duty. Thoughts are irrelevant except as a subjective element qualifying an act already performed (mens rea). Ulpian’s principle cogitationis poenam nemo patitur (Digest 48.19.18) sums up two thousand years of tradition: no one is punished for his thoughts.[27]

8.2 Predictive justice as a criminal law of the offender

Predictive justice does precisely what liberal criminal law had excluded: it punishes presumed intentions and attitudes, not acts committed. The algorithmic profile infers from past behaviour a “character” or “disposition” of the individual, from which it infers the probability of a future act, on which it bases a restrictive measure. What is punished is not an act: it is a psychological structure inferred algorithmically.

In technical terms, this is a criminal law of the offender (Täterstrafrecht) — the most dangerous form of criminal law, historically associated with totalitarian regimes. The criminal law of the offender punishes the subject for what he is, not for what he has done. Nazism theorized it explicitly (the criminal as a biological-racial “type”); Stalinism practiced it under the category of the “class enemy”. Predictive justice revives it in statistical and technocratic form, with the further aggravating factor of algorithmic opacity.[28]

8.3 Who chooses the moral principles of the algorithm?

Every predictive-justice algorithm embodies a value choice about what counts as a “risk factor”. That choice is opaque (proprietary systems such as COMPAS are protected by trade secrecy), lacking logical grounding and, above all, not shared by the subjects of control (the criteria are chosen by private developers, not by parliaments), self-referential (the algorithm trains on historical data reflecting past punitive choices), and exempt from individual review (the algorithm cannot be cross-examined).

In Loomis v. Wisconsin (2016), the Wisconsin Supreme Court upheld a sentence based on COMPAS, even though the defendant had been unable to examine the algorithm’s criteria. The U.S. Supreme Court declined to hear the case. This is the most significant judicial precedent on the question: it establishes that an algorithm’s trade secrecy can override a defendant’s right to challenge the basis of his own conviction.[29]

Cathy O’Neil has systematically shown how algorithms of this kind do not reduce biases but formalize and multiply them, lending them the authority of scientific objectivity. Frank Pasquale has documented how algorithmic opacity constitutes a structural problem, not a correctable technical imperfection.[30]

9. Lombroso, racial profiling, and the return of the born criminal

9.1 The Positivist School and the shift from the offence to the offender

Algorithmic predictive justice is not the first time in the history of criminal law that judgment shifts from the criminal act to the personality of the criminal. Cesare Lombroso, with L’uomo delinquente (1876), founded criminal anthropology on a radical thesis: the criminal is recognizable by atavistic somatic anomalies — physical stigmata (cranial conformation, prognathism, facial asymmetries) supposedly revealing an innate biological predisposition to crime. The “born criminal” is not punished for what he has done: he is declared dangerous for what he is, for his biological make-up.[31]

The Positivist School — with Lombroso, Enrico Ferri, and Raffaele Garofalo — proposed an explicit reversal of the foundations of Francesco Carrara’s Classical School: no longer the analysis of the offence as an abstract legal category and a judgment on the moral responsibility of a free subject, but the study of the “natural history of the criminal man” and the calibration of punishment not to the gravity of the offence but to the social dangerousness of the offender, understood as an empirically detectable probability of committing future offences. Ferri developed in parallel his Sociologia criminale, denying free will and grounding criminal law on the deterministic prevention of crime.[32]

Lombrosian doctrine was opposed by the Catholic tradition precisely because it undermined the dogma of free will: if crime is biologically determined, there is no moral responsibility, no culpability, no possibility of redemption. The Thomistic critique of actus humanus as the foundation of judgment converged with the classical legal critique: both rejected the reduction of the individual to a predetermined natural type.

9.2 Racial profiling as algorithmic Lombrosianism

The connection between Lombrosian theory and contemporary predictive justice is not merely historical: it is structural. In both cases, the subject is classified as dangerous not for an act committed but for characteristics deemed indicative of a tendency toward crime. In Lombroso, those characteristics were somatic and biological; in predictive justice, they are statistical and socioeconomic. But the logic is identical: from nature (biological or statistical) future behaviour is inferred; from future behaviour the present restrictive measure is justified.

Racial profiling — the use of race or ethnicity as a risk factor in policing and the criminal justice system — is the most explicit contemporary form of this nexus. In predictive-justice systems, race is not necessarily an explicit variable in the model: it is a proxy variable embedded in the historical data on which the algorithm is trained. If the criminal justice system has historically arrested and convicted minorities at disproportionate rates, the data reflects that history; the algorithm learns it as the norm and reproduces it as prediction. As the UN Special Rapporteur on racism has documented, “bias from the past leads to bias in the future”: socioeconomic variables (area of residence, education level, economic conditions) function as proxies for race, perpetuating historical discrimination in the neutral language of statistics.[33]

This vicious circle has been empirically documented: predictive-policing algorithms increase surveillance in minority-majority neighbourhoods; increased surveillance produces more arrests; more arrests update the model, confirming the “dangerousness” of the neighbourhood; the model directs further surveillance. This is not a correctable technical error: it is a self-fulfilling prophecy structurally embedded in the logic of the system.[34]

9.3 From biology to statistics: a structural continuity

The difference between Lombroso and algorithmic predictive justice is one of language and method, not of logical structure. Lombroso measured skulls and catalogued physical stigmata; the algorithm calculates risk scores on socioeconomic variables. But both perform the same fundamental operation: they infer future dangerousness from the subject’s present nature, bypassing the mediation of the free and culpable act. Both claim the authority of science: Lombroso that of positivist anthropology, the algorithm that of machine learning.

The critique that the Classical School and the Catholic tradition directed at Lombroso applies with equal force to predictive justice. Francesco Carrara had identified with precision the core of the problem: criminal law cannot dispense with moral culpability, that is, with the subject’s capacity to understand and to will at the moment of the act. Shifting judgment from responsibility for the fact accomplished to dangerousness based on the nature or characteristics of the subject means abandoning criminal law and replacing it with a biological or statistical police — with the difference that Lombroso’s was at least explicit about its premises, whereas the algorithmic version presents itself as neutral and objective.[35]

10. Freedom of thought, the chilling effect, and thought police

10.1 Freedom of expression and its foundations

Freedom of thought and expression — art. 21 of the Italian Constitution; art. 10 ECHR; art. 19 of the 1966 International Covenant on Civil and Political Rights — presupposes that thought is legally irrelevant until it translates into acts. John Stuart Mill, in the foundational On Liberty (1859), states the harm principle: power may restrict an individual’s liberty only to prevent harm to others, not to manage his statistical probability of future behaviour.[36]

The European Court of Human Rights has consistently held that freedom of expression covers even ideas that “offend, shock or disturb”: its essential core cannot be sacrificed in the name of preventive security.[37]

10.2 The chilling effect as a constitutional harm

The doctrine of the chilling effect — developed by the U.S. Supreme Court from the 1950s onward — recognizes that the threat of legal consequences can suppress the exercise of fundamental rights even without an explicit prohibition. If predictive justice associates certain kinds of expression, certain associations, or certain online activity with criminal risk profiles, it produces exactly this effect: it does not forbid thought, but renders it statistically costly.[38]

Stefano Rodotà identified with great clarity how pervasive digital control operates not only as an ex post instrument of repression, but as an ex ante device of conformism: people modify their behaviour — including their intellectual expression — when they know or fear they are being watched. The digital panopticon does not need to intervene to be effective: it is enough that its presence be perceived as possible.[39]

10.3 Pre-crime and thought police as a logical outcome

George Orwell in Nineteen Eighty-Four (1949) describes a thought police that intervenes on crime before it manifests itself: crimethink is punished not for what it produces but for what it is. Algorithmic predictive justice reproduces this structure with one added element: the authority of science. Orwell’s Big Brother was explicit about its arbitrariness; the algorithm presents itself as neutral, objective, inevitable.[40]

The issue is not merely theoretical. Four levels of usurpation overlap and interpenetrate: on the ontological plane, what has not yet happened is punished; on the subjective plane, what one is, not what one has done, is punished; on the cognitive plane, inferred intentions and attitudes are punished; on the political plane, a morality is imposed without that imposition corresponding to any form of consent on the part of its victims, not even in the very imperfect form of so-called “democratic legitimacy”. Predictive justice is not a correctable technical error: it is the simultaneous subversion of four foundational principles of liberal law.[41]

11. Synthesis: the historical line of criminalization by condition

The thread running through this article can now be stated in its complete form. The enclosures structurally created the condition of the propertyless. Vagrancy laws criminalized that condition. Transportation offered a risk-management solution to it ante litteram. Belloc and Chesterton identified this sequence as the origin of the modern servile state. Catholic moral theology has always insisted on personal culpability for the act as the foundation of judgment. The Austrian School, from Mises to Hoppe, has shown why the prediction of future individual behaviour is categorially impossible. Classical law codified all of this in the principle da mihi factum, dabo tibi ius.

Predictive justice based on artificial intelligence systematically dissolves all these bulwarks. It is not a technical novelty: it is the re-emergence, by computational means, of a logic of criminalization by social condition. And it adds a new level of danger: its vocation toward algorithmic morality — the imposition of behavioural and cognitive standards chosen by private actors with no possibility of control by the victims of the system — potentially transforms it into an instrument of thought control, not merely of action control.

The gravest risk is not the technical error — that the algorithm gets a single case wrong. The gravest risk is the naturalization of this logic: that mathematical formalization confers upon the socioeconomic equation the authority of scientific objectivity, rendering invisible the value choice that underlies it.[42]

12. Conclusions: toward a justice that remains faithful to the fact

The analysis offered here suggests some methodological conclusions for the debate on the use of artificial intelligence in the criminal justice system. These conclusions are all the more important given the recently announced upcoming introduction of the systematic use of AI in police activity in Italy, as the government has recently announced.

First, the distinction between fact and prediction is not technical but ontological. No improvement in algorithms can bridge this gap, because future human action is not a hidden datum to be discovered but a reality that does not yet exist. Da mihi factum is not a surmountable limitation: it is the logical structure of judgment.

Second, the use of AI in criminal justice is not neutral. In a society in which property is concentrated and socioeconomic conditions are radically unequal owing to situations of privilege created by fiat money and state regulation, any system of statistical profiling will tend to encode that inequality into a presumption of dangerousness. This problem cannot be corrected by technical adjustments: it requires reflection on the distributive structure of society.

Third, tradition — legal, philosophical, and theological — offers critical resources that it would be a grave error to ignore. The Roman-law principle of culpability for the fact, the Catholic doctrine of actus humanus and of personal culpability, the Austrian critique of the impossibility of calculation without private property, the distributist denunciation of the servile state: all these strands converge in indicating that a justice faithful to the dignity of the person can only be justice of the fact, not of prediction.

Fourth, the risk of an algorithmic thought police is not science fiction but a coherent logical extension of the principles on which predictive justice already operates. When a system begins to infer dangerousness from expressions, associations, and online activity, the boundary between punishing an act and surveilling a thought dissolves. Maar’s Sams grants wishes; the algorithm represses thoughts — with the same literal precision, with the same indifference to consequences.

Endnotes

[1]Paul Maar, Eine Woche voller Samstage, Verlag Friedrich Oetinger, Hamburg, 1973 (first volume of the Das Sams series). The quotation “Du musst viel, viel genauer wünschen” appears in the second volume: Paul Maar, Am Samstag kam das Sams zurück, Verlag Friedrich Oetinger, Hamburg, 1980, p. 37. The series comprises eleven novels and has sold over six million copies; it is considered a classic of German-language children’s literature.

[2]The principle da mihi factum, dabo tibi ius is of well-established Roman-law derivation and summarizes the structure of classical legal reasoning: the rule applies to the fact, it does not precede it in terms of individual imputation. Cf. Charles Perelman, Logique juridique. Nouvelle rhétorique, Dalloz, Paris, 1976.

[3]Julia Angwin, Jeff Larson, Surya Mattu, Lauren Kirchner, “Machine Bias”, ProPublica, 23 May 2016. The investigation demonstrated that COMPAS systematically overestimated the risk of recidivism for African-American defendants compared to white defendants, effectively encoding pre-existing socioeconomic inequalities into an apparently objective verdict.

[4]Ludwig von Mises, Human Action: A Treatise on Economics, Yale University Press, New Haven, 1949, ch. I (“Acting Man”). Italian edition: L’azione umana. Trattato di economia, Rubbettino, Soveria Mannelli, 2021. The axiom is also set out in Ludwig von Mises, Nationalökonomie: Theorie des Handelns und Wirtschaftens, Editions Union, Genf, 1940.

[5]Ludwig von Mises, Theory and History: An Interpretation of Social and Economic Evolution, Yale University Press, New Haven, 1957. Mises rigorously distinguishes between the natural sciences, which deal with constants and regularities, and the sciences of action, which deal with irreducibly free individual choices.

[6]F.A. Hayek, “The Use of Knowledge in Society”, American Economic Review, vol. 35, no. 4, 1945, pp. 519-530. This essay, together with “Economics and Knowledge” (Economica, vol. 4, no. 13, 1937, pp. 33-54), constitutes the epistemological foundation of the Hayekian critique of socialism.

[7]F.A. Hayek, The Counter-Revolution of Science: Studies on the Abuse of Reason, The Free Press of Glencoe, New York, 1952. Hayek there develops the critique of “scientism” — that is, the attempt to apply the methods of the natural sciences to the social sciences.

[8]F.A. Hayek, Law, Legislation and Liberty, vol. I (Rules and Order), Routledge & Kegan Paul, London, 1973. Hayek distinguishes between thesis (deliberately constructed order) and nomos (spontaneous order emerging from individual action). Predictive justice belongs to the logic of thesis, claiming to construct ex ante what classical law can only judge ex post.

[9]Jörg Guido Hülsmann, “Knowledge, Judgment, and the Use of Property”, Review of Austrian Economics, vol. 10, no. 1, 1997, pp. 23-48. Hülsmann argues that the problem of socialism is unsolvable not because information is lacking, but because without private property rational economic calculation is structurally impossible.

[10]Hans-Hermann Hoppe, A Theory of Socialism and Capitalism: Economics, Politics, and Ethics, Kluwer Academic Publishers, Boston, 1989, ch. V-VI. Cf. also Hans-Hermann Hoppe, The Economics and Ethics of Private Property, Kluwer Academic Publishers, Boston, 1993 (2nd ed., Ludwig von Mises Institute, Auburn, 2006).

[11]Hans-Hermann Hoppe, “Of Private, Common, and Public Property and the Rationale for Total Privatization”, Libertarian Papers, vol. 3, no. 1, 2011. The theme of property as the logical foundation of any system of rational judgment is developed in this essay in a way particularly relevant to our argument.

[12]Catechism of the Catholic Church, Libreria Editrice Vaticana, Vatican City, 1992, §§1730-1748 (freedom and moral responsibility) and §§1849-1876 (sin). On the Thomistic foundation, cf. Thomas Aquinas, Summa Theologiae, I-II, qq. 6-21 (De voluntario et involuntario).

[13]Catechism of the Catholic Church, cit., §1735. The doctrine is developed by Servais Pinckaers O.P., Le fonti della morale cristiana. Metodo, contenuto, storia, Ares, Milan, 1992 (first Italian edition of the original work Les sources de la morale chrétienne, Fribourg, 1985), in particular ch. X-XI on the structure of the human act.

[14]Thomas Aquinas, Summa Theologiae, II-II, q. 61 (De iustitia) and q. 108 (De vindicatione). Cf. also Giuseppe Graneris, Contributi tomistici alla filosofia del diritto, SEI, Turin, 1949.

[15]Augustine of Hippo, De libero arbitrio, introductory study, text, translation and commentary by Franco De Capitani, Vita e Pensiero, Milan, 1987 (reprinted 1994). Cf. also Francesco Compagnoni, Giannino Piana (eds.), Nuovo Dizionario di Teologia Morale, Paoline, Cinisello Balsamo, 1990, entries Imputabilità and Peccato.

[16]Cf. also Kevin Carson, Organization Theory, 2008, ch. II on economies of scale as the result of the distortion caused by state intervention.

[17]Hilaire Belloc, The Servile State, T.N. Foulis, London-Edinburgh, 1912. The text is available in the archive of the Ludwig von Mises Institute. The central thesis is that industrial capitalism is not stable: it inevitably produces the concentration of property and the dependence of the majority.

[18]G.K. Chesterton, The Outline of Sanity, Methuen & Co., London, 1926. Cf. also G.K. Chesterton, What’s Wrong with the World, Cassell and Company, London, 1910. For a systematic analysis of distributism, cf. John Médaille, Toward a Truly Free Market: A Distributist Perspective on the Role of Government, Taxes, Health Care, Deficits, and More, ISI Books, Wilmington (DE), 2010.

[19]Leo XIII, Encyclical Rerum Novarum, 15 May 1891, §§4-14. Pius XI, Encyclical Quadragesimo Anno, 15 May 1931, §§105-110 on the concentration of wealth and the principle of subsidiarity.

[20]The concept of status criminis — criminal responsibility grounded not in the act but in the condition of the subject — is critically analyzed in Bernard Harcourt, Against Prediction: Profiling, Policing, and Punishing in an Actuarial Age, University of Chicago Press, Chicago, 2007. Harcourt demonstrates empirically how actuarial systems tend to concentrate surveillance on the most socioeconomically vulnerable groups.

[21]Karl Polanyi, The Great Transformation: The Political and Economic Origins of Our Time, Farrar & Rinehart, New York, 1944. Cf. also E.P. Thompson, The Making of the English Working Class, Victor Gollancz, London, 1963.

[22]Hilaire Belloc, The Servile State, cit., ch. IV-V, where Belloc reconstructs how the English Protestant Reformation — with the confiscation of ecclesiastical property — set in motion the process of concentration of property that the enclosures would later complete.

[23]For a systematic treatment of vagrancy legislation, cf. Douglas Hay, Peter Linebaugh et al., Albion’s Fatal Tree: Crime and Society in Eighteenth-Century England, Allen Lane, London, 1975. E.P. Thompson, Whigs and Hunters: The Origin of the Black Act, Allen Lane, London, 1975 analyzes the Black Act of 1723, which turned the traditional practices of the poor into capital crimes. Cf. also A.L. Beier, Masterless Men: The Vagrancy Problem in England 1560-1640, Methuen, London, 1985.

[24]The Vagrancy Act 1824 (5 Geo. 4 c. 83): section 81 of the Police, Crime, Sentencing and Courts Act 2022 provides for its repeal, conditional on the approval of replacement legislation not yet fully operative. On the history of transportation, cf. Robert Hughes, The Fatal Shore: A History of the Transportation of Convicts to Australia, Collins Harvill, London, 1987.

[25]Michel Foucault, Surveiller et punir: Naissance de la prison, Gallimard, Paris, 1975 (English ed. Discipline and Punish, Pantheon Books, New York, 1977). Foucault analyzes how the birth of the modern prison corresponds to a transformation in the object of punitive power: no longer the body of the condemned (the spectacle of punishment), but his identity and his character.

[26]Thomas Aquinas, Summa Theologiae, I-II, q. 74 (De peccatis interioribus): Aquinas distinguishes the sin of thought from the sin of deed; both are moral sins, but human law cannot punish the former. Cf. also I-II, q. 91, a. 4 (De lege humana): “human law cannot forbid everything that natural law forbids… it limits itself to repressing the most serious vices”.

[27]Ulpian, Digest, 48.19.18: “Cogitationis poenam nemo patitur.” The principle is taken up and systematized in modern criminal-law theory in: Claus Roxin, Strafrecht. Allgemeiner Teil, vol. I, C.H. Beck, Munich, 4th ed. 2006, §8 (the principle of the act, Tatstrafrecht, as the foundation of liberal criminal law). Cf. also Luigi Ferrajoli, Diritto e ragione. Teoria del garantismo penale, Laterza, Rome-Bari, 1989, ch. II (the principle of materiality).

[28]Claus Roxin, Strafrecht. Allgemeiner Teil, cit., §6: systematic distinction between Tatstrafrecht (criminal law of the act) and Täterstrafrecht (criminal law of the offender). Cf. Luigi Ferrajoli, Diritto e ragione, cit., ch. IV, on the “criminal law of the enemy” as an authoritarian drift of contemporary criminal systems.

[29]Loomis v. Wisconsin, 881 N.W.2d 749 (Wis. 2016), cert. denied, 137 S. Ct. 2290 (2017). Eric Loomis was sentenced to six years’ imprisonment; the judge relied on the COMPAS score while stating that it was not the exclusive basis of the sentence. The Wisconsin Supreme Court permitted the use of the tool but laid down certain cautions, acknowledging the risks of a system whose methodological foundations could not be subjected to adversarial challenge.

[30]Cathy O’Neil, Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy, Crown Publishers, New York, 2016. Frank Pasquale, The Black Box Society: The Secret Algorithms That Control Money and Information, Harvard University Press, Cambridge (MA), 2015.

[31]Cesare Lombroso, L’uomo delinquente studiato in rapporto all’antropologia, alla medicina legale e alle discipline carcerarie, Hoepli, Milan, 1876 (first edition). The work reached its fifth edition in 1897 (Vincenzo Bona, Turin), substantially expanded. Lombroso estimated that 70% of the worst criminals fell into the category of the “born criminal” — individuals in whom criminality is inherent to their biological nature, irredeemable subjects to be confined or eliminated in the name of social defence.

[32]Enrico Ferri, Sociologia criminale, 5th ed., Utet, Turin, 1929 (first edition I nuovi orizzonti del diritto e della procedura penale, Zanichelli, Bologna, 1881). Raffaele Garofalo, Criminologia, Bocca, Turin, 1885. The Positivist School explicitly opposed its method to that of the Classical School: cf. Francesco Carrara, Programma del corso di diritto criminale, Fratelli Cammelli, Florence, 1859, where the offence is defined as “a violation of the law of the State, promulgated to protect the security of citizens, resulting from an external act of man, positive or negative, morally imputable and politically harmful” — a definition that presupposes freedom, culpability, and an accomplished fact.

[33]Ashwini K.P., United Nations Special Rapporteur on contemporary forms of racism, racial discrimination, xenophobia and related intolerance, Report on racism and artificial intelligence, OHCHR, 2024: “bias from the past leads to bias in the future.” Cf. also Rashida Richardson, Jason Schultz, Kate Crawford, “Dirty Data, Bad Predictions: How Civil Rights Violations Impact Police Data, Predictive Policing Systems, and Justice”, New York University Law Review Online, vol. 94, 2019, pp. 15-55: the study documents how numerous American predictive-policing systems were trained on data collected during periods of documented systemic racial discrimination by law enforcement.

[34]Julia Angwin, Jeff Larson, Surya Mattu, Lauren Kirchner, “Machine Bias”, ProPublica, 2016, cit. The mechanism of the racial feedback loop in predictive-policing algorithms is analyzed systematically in Bernard Harcourt, Against Prediction, cit., ch. V, and in Cathy O’Neil, Weapons of Math Destruction, cit., ch. 5.

[35]Francesco Carrara, Programma del corso di diritto criminale, cit., §30: “The right to punish arises solely from the offence committed, and is measured by the culpability of the offender and the gravity of the harm.” The critique of social dangerousness as a basis for punishment is developed in Luigi Ferrajoli, Diritto e ragione, cit., ch. II-IV, where Ferrajoli shows how the concept of social dangerousness, although formally surviving in the Italian system in security measures (arts. 202-212 of the Criminal Code), is compatible with a garantista framework only insofar as it presupposes an offence actually committed as a necessary condition.

[36]John Stuart Mill, On Liberty, John W. Parker and Son, London, 1859, ch. I. The harm principle is the classical liberal foundation of any legitimate restriction of individual liberty: the law intervenes on the harm-producing fact, not on the psychological disposition that might produce it. Cf. also Luigi Ferrajoli, Diritto e ragione, cit., for the garantista elaboration of the principle in continental criminal law.

[37]European Court of Human Rights, Handyside v. United Kingdom, 7 December 1976 (application no. 5493/72), §49: freedom of expression applies “not only for ‘information’ or ‘ideas’ that are favourably received or regarded as inoffensive or as a matter of indifference, but also for those that offend, shock or disturb.” Cf. also ECtHR, Oberschlick v. Austria, 23 May 1991.

[38]Lamont v. Postmaster General, 381 U.S. 301 (1965): the first explicit recognition of the chilling effect by the U.S. Supreme Court. Cf. Keyishian v. Board of Regents, 385 U.S. 589 (1967): the chilling effect as an autonomous harm to fundamental freedoms. The Court of Justice of the EU applied analogous reasoning in Digital Rights Ireland, C-293/12, 8 April 2014: mass surveillance as a violation of privacy with a deterrent effect on freedom of expression.

[39]Stefano Rodotà, Il mondo nella rete. Quali i diritti, quali i vincoli, Laterza, Rome-Bari, 2014. Rodotà, among the leading European scholars of law and technology, likens algorithmic surveillance to the Benthamite panopticon as revisited by Foucault: a control device that operates through the internalization, by the controlled subject, of his own visibility.

[40]George Orwell, Nineteen Eighty-Four, Secker & Warburg, London, 1949. The concepts of thoughtcrime and Newspeak as instruments of cognitive control have been analyzed from a politico-legal perspective by many later scholars. Cf. Philip K. Dick, The Minority Report, in Science Fiction Stories, 1956: pre-crime as a narrative dystopia that anticipates with precision the logic of contemporary predictive justice.

[41]Massimo Donini, Il volto attuale dell’illecito penale, Giuffrè, Milan, 2004, analyzes the crisis of the criminal law of the act in the contemporary Italian legal system and the drift toward forms of enemy criminal law. Cf. also Sergio Moccia, La perenne emergenza. Tendenze autoritarie nel sistema penale, ESI, Naples, 2nd ed. 1997.

[42]Antoinette Rouvroy, Thomas Berns, “Gouvernementalité algorithmique et perspectives d’émancipation: le disparate comme condition d’individuation par la relation?”, Réseaux, no. 177, 2013, pp. 163-196 (DOI: 10.3917/res.177.0163). Rouvroy and Berns show how algorithmic governmentality does not merely describe social reality, but produces and normalizes it, systematically eluding individual subjectivity.

1. Alessandro Fusillo is a libertarian attorney based in Italy, Spain, and Germany. [↩]