Bernard E. Harcourt – Against Prediction: Profiling, Policing, and Punishing in an Actuarial Age (2007)

Against Prediction: Profiling, PolicingThe practice of maintaining public order and enforcing laws through authorized institutions., and Punishing in an Actuarial Age (2007) is one of the key works by Bernard E. Harcourt and a foundational analysis of contemporary actuarial justice. Harcourt critiques the growing reliance on statistical risk assessment in criminal justice—particularly through police profiling, predictive policing, and risk-based sentencing. The book is a sharp critique of predictive logic and a call for a political and ethical rethinking of criminal law and security regimes.

Social Context and Theoretical Framework

Harcourt situates his critique in a paradigm shift in criminal policy since the 1980s: away from the idea of individual culpability and rehabilitation, toward a technocratic logic of risk that governs people based on aggregated probabilities. He argues that this shift has fundamentally transformed not only criminal law, but also policing, social policy, and security architecture.

The book combines theoretical reflection (drawing on Foucault, Becker, Bentham, and Hayek) with empirical analysis and case studies—covering topics such as racial profiling, zero tolerance, and data mining in the U.S. justice system. Harcourt builds on his earlier work on neoliberal models of order, particularly his ideological critique of the “free market” as a state-constructed myth.

Key Points

Key Arguments

A central thesis of Harcourt’s work is that risk-based criminal justice undermines equality before the law. Rather than focusing on individual guilt, justice is based on statistical probabilities. Individuals are no longer seen as responsible agents, but as bearers of abstract risk values. This shift leads to people being targeted not because of actual wrongdoing, but because they belong to certain risk groups—such as youth, people of color, or low-income individuals. The result is selective law enforcement that reinforces existing social inequalities.

Harcourt also argues that actuarial justice is performative: it creates the very risk profiles it claims to measure neutrally. Those who are policed more often appear more often in statistics—and are therefore labeled as “high risk.” This ratchet effect creates a self-reinforcing feedback loop that keeps certain groups under permanent suspicion. The statistics do not reflect reality—they construct a new reality of social selection.

Harcourt pays special attention to the idea that racist stereotypes are not accidental byproducts of predictive models—they are systemic. He illustrates this with examples from traffic stops and drug laws in the United States, where Black Americans are disproportionately surveilled, stopped, and punished—despite similar behavioral patterns to white Americans. Predictive models amplify these disparities because they are based on biased, racialized data.

Lastly, Harcourt critiques the technocratization of criminal policy. Political debate over key constitutional principles—such as liberty, proportionality, rehabilitation, or the social causes of deviance—is increasingly replaced by an allegedly “neutral” data-driven logic of governance. Algorithms substitute deliberation, and risk scores displace normative arguments. This leads to a depoliticization of criminal justice and a shift toward an authoritarian security logic where efficiency trumps justice.

An illustrative example is the CompStat system in New York City. Initially introduced as a data-driven management tool, it became a central mechanism for guiding police work. The pressure to reduce crime numbers led to selective enforcement and a concentration of policing in disadvantaged neighborhoods. Harcourt shows how such systems do not merely reflect statistical bias—they amplify it.

Harcourt distinguishes his critique from other predictive methods based on concrete knowledge of offenders—such as geographic profiling in serial murder investigations. His argument is not against prediction per se, but against a generalized logic of risk that produces and legitimizes structural discrimination.

In this context, there are strong connections to Thomas Mathiesen – The Politics of Abolition (1974). Both authors criticize the expansion of repressive state apparatuses, the erosion of democratic control, and the symbolic nature of security politics. While Mathiesen focuses on institutional self-preservation, Harcourt examines the epistemic logic that legitimizes such systems.

A similarly critical perspective is offered by Virginia Eubanks – Automating Inequality (2018), which extends the analysis of actuarial logics beyond criminal justice into the realm of social welfare and public administration. Eubanks documents how predictive algorithms are increasingly used to manage poverty, allocate housing, and determine eligibility for state assistance—often to the detriment of the most vulnerable populations.

Both Harcourt and Eubanks reveal how data-driven systems, though presented as neutral and efficient, in fact amplify structural inequalities. Where Harcourt speaks of a ratchet effect that traps marginalized groups in cycles of surveillance and punishment, Eubanks describes the emergence of a “digital poorhouse” that automates exclusion and reinforces poverty. Their work jointly illustrates how predictive governance undermines democratic accountability and redistributes power away from citizens and toward opaque technological infrastructures.

Impact and Interdisciplinary Relevance

Against Prediction has significantly shaped discussions on algorithmic policing, risk assessment in criminal justice, and digital surveillance. It is considered essential reading in critical police studies, algorithmic ethics, and predictive justice.

Harcourt’s ideas have been widely cited in debates on predictive policing (e.g., with software like Precobs or Gotham), and they have influenced critiques of risk-based tools such as CompStat, RECAP, and Risk-Needs-Responsivity Models.

The book also deepens many of the arguments later expanded in The Illusion of Free Markets (2011), where Harcourt explores the convergence of neoliberal market ideology and punitive state power.

Against Prediction thus provides a foundational critique of algorithmic policing—especially in an era increasingly shaped by AI-powered prosecution, predictive analytics, and data-driven prevention.

Conclusion

Bernard E. Harcourt’s Against Prediction is a theoretically incisive and empirically grounded work that exposes the technocratic logic of modern security regimes. It demonstrates how deeply the idea of “predictability” has permeated criminal justice, policing, and governance—and how this shift endangers the rule of law, equality, and democracy.

The book remains a cornerstone for the critique of algorithmic policing—now more relevant than ever in an age of AI-driven justice, predictive policing, and digital control.

References

• Harcourt, B. E. (2007). Against Prediction: Profiling, Policing, and Punishing in an Actuarial Age. Chicago: University of Chicago Press.
• Harcourt, B. E. (2011). The Illusion of Free Markets: Punishment and the Myth of Natural Order. Cambridge, MA: Harvard University Press.
• Foucault, M. (1975). Discipline and Punish. New York: Pantheon Books.