Theorizing the Performative Effects of Penal Risk Technologies: (Re)producing the Subject Who Must Be Dangerous

Abstract

This article explores the ways in which practices of risk assessment exert material and semiotic effects that structure how penal subjects are constituted, imagined, and governed. In so doing, it proposes conceptual shifts in how we understand risk logics and practices. It contends that techniques of assessment and classification within parole operate performatively; that is, they do not so much describe reality as they constitute, structure and alter what they appear to report on. While this occurs through shaping the beliefs of penal actors – that is, through ideological mechanisms – this article focuses on the ways in which assessments exert institutional, bureaucratic and automatic effects independent of beliefs. I argue that, through exerting these effects, assessments make the risk of paroled subjects an institutional and practical certainty. While the dangerousness of individuals on parole is historically and ideologically contingent, contemporary practices of risk operate in a way that precludes the possibility of a non-dangerous individual.

Keywords

Criminal justice parole performativity punishment risk assessment

Introduction

Understanding penal subjects as dangerous, as individuals who represent a threat to public safety and hence require penal intervention, is a complex socio-legal achievement that involves multiple elements (Foucault, 1977, 2007; Melossi, 2000; see also Becker, 1963). It entails – and assembles – law, theories of criminology and penology, lay theories and understandings, and media representations. It also includes techniques and devices of risk assessment, which have expanded exponentially in the last two decades and become nearly ubiquitous elements of penality. This article focuses specifically on these practices of risk assessment and classification and traces the ways in which they produce material and semiotic effects that structure how penal subjects are constituted, imagined, and governed.

In so doing, this article proposes a theoretical reframing of penal risk assessment with the objective of complicating and expanding our understanding of risk practices. Existing scholarship has traced the ways in which the assessment and classification of risk can, and does, influence both beliefs and behavior in the penal realm (e.g. Grattet and Lin, 2016; Harcourt, 2007; Petersilia and Turner, 1993, Werth, 2017; see Kemshall, 2011; Hannah-Moffat, 2013 for reviews). Yet, this article goes a step further as I argue that risk assessments not only influence penal practices, they constitute and structure the realm in which they operate and the targets (penal subjects) upon which they report. That is, assessments not only describe the social world, they help create it. Focusing specifically on parole, this article contends that classifications and assessments of risk participate in producing an institutional and practical reality where individuals on parole are inherently and always dangerous subjects.

While focusing on risk practices within parole, I will suggest that the arguments that follow are relevant to risk practices throughout the penal realm. While risk is a multivocal and somewhat contested term (Douglas, 1992; O’Malley, 2004), in the penal realm it typically refers to the risk of criminal reoffending (Sparks, 2001), but it can also refer to the risk of violating institutional rules or conditions that are not necessarily criminal (i.e. technical violations).¹ By risk assessment, I refer to acts of classification, evaluation, and/or prediction of the risk of penal subjects. This includes, but is not limited to, actuarial techniques. Parole personnel utilize varied techniques to assess risk, including actuarial risk assessment instruments, standardized (but non-actuarial) classifications, and more ‘subjective’ judgments (Hannah-Moffat et al., 2010; Werth, 2017). In exploring and theorizing these tools of risk assessment, this article draws both from previous scholarship and from empirical examples taken from archival and ethnographic research conducted by the author with parole personnel in California.²

The central arguments are threefold. First, I contend that acts of risk assessment do not only describe reality, they constitute, structure, and alter the reality that they purport to describe. That is, penal risk assessments operate performatively. The concept of performativity, developed by the philosopher Austin (1962), highlights that although we typically think of language as describing facts or a state of affairs, it can make things ‘true’ by saying them. That is, discourses do not always report on an already existing and objective world, they can create and/or impact the world that they ostensibly depict. Specifically, I contend that risk assessments within parole automatically and universally constitute individuals on parole as dangerous subjects – as individuals who possess risk as a condition. Further, they exert effects that perpetuate and sometimes intensify this risk throughout the period of parole.

Second, while risk assessments impact parole through shaping the beliefs of parole personnel, I contend that they also exert ontological effects – through institutional, bureaucratic, and automatic mechanisms – that structure penality independent of beliefs. As will be detailed, the ways in which risk assessments can and do shape beliefs has received considerable scholarly attention (Bullock, 2013; Hannah-Moffat, 2013; Hannah-Moffat et al., 2010; Kemshall, 2011; Petersilia and Turner, 1993), yet their ontological effects remain under-explored and under-theorized (although see Harcourt, 2007). I turn to performativity as it alerts us to consider the ways in which discourses – and the knowledges, techniques, and devices that inform them – exert effects that constitute and arrange the social world through mechanisms other than beliefs. I also draw from work emanating from science and technology studies and actor-network theory that attends to the ways in which objects, devices, and technologies operate as actors or ‘actants’ that generate effects (Latour, 2005; see also Ballestero, 2015). In particular, I contend that practices of risk assessment and classification produce effects separate from, and sometimes contrary to, the beliefs of parole personnel. Even in cases where parole personnel doubt or reject available risk assessments, these techniques constitute and structure the reality they purport to describe.

Third, through exerting these performative effects, risk assessments constitute all paroled subjects as risky/dangerous. That is, they establish an ontological order that precludes the possibility of a parolee who is not risky. While risk assessment is often understood as a predictive and probabilistic technology that embraces uncertainty (Beck, 1992; Ewald, 2002; Hannah-Moffat, 2013; Rose, 1998; Simon, 2005; cf. Douglas, 1992), this article contends that in the penal realm it operates in a way that makes risk a certainty. Acts of assessment disperse risk to everyone on parole; they produce all paroled subjects as risky of reoffending to some degree. In this way, it could be said that parole evaluation is somewhat of a false act of evaluation, or at least a predetermined and delimited one. Rather than querying whether or not someone is risky, assessments ask how risky is this person. While this may appear as a minor difference, I argue it is a largely overlooked and immensely consequential one, as it determines the limits of the possible: a non-dangerous parolee becomes an institutional and practical impossibility. Within classification, evaluation, and prediction, there is no outside to risk, no possibility of an absence of risk.³

In this way, assessments contribute to the production of the always already risky paroled subject; that is, the parolee who is constituted as definitely risky and who remains so throughout the period of parole. Importantly, this was not always the case. The dangerousness and redeemability of paroled subjects has displayed considerable historical and ideological contingency. For instance, in the early 20th century, penological discourses talked about parolees as already rehabilitated, and parole was framed as a way to help these reformed (i.e. not dangerous) individuals resettle into the community (Cavender, 1982; Simon, 1993). By the middle of the century, however, parolees were no longer framed as already reformed, they were now seen as individuals who still posed a danger to society and were in need of additional rehabilitation in the community (Simon, 1993). By the 1970s, individuals on parole – and, in fact, penal subjects in general – were not only dangerous, they were increasingly seen as individuals incapable of reformation or, at least, unlikely to change (Simon, 1993). Today, while rehabilitation has returned as a possibility, individuals on parole are seen as precarious, dangerous beings, and parole agencies are viewed as responsible for helping protect the public from these risky offenders (Dagan and Segev, 2014; Lynch, 1998; Simon, 1993; Werth, 2013). Highlighting this, the California parole agency publicly proclaims its first mandate as responsibility ‘for protecting the community’ from the potential threat of parolee recidivism (California Department of Corrections and Rehabilitation, 2016). Further, among the parole personnel that I conducted ethnographic fieldwork with, individuals on parole were seen as dangerous and deficient persons in need of state regulation. Constituting a group of individuals as dangerous and deficient – not just at the moment of being placed on parole, but throughout the period of parole – requires maintenance. Subjects are not formed in a singular moment; a multiplicity of acts are necessary to constitute, affirm, and reaffirm them as particular kinds of subjects (Butler, 1997b; Foucault, 1977).

This is not to claim that the risky paroled subject is created and sustained solely by techniques of risk assessment and classification. As noted at the opening, this subject has a much longer history that traverses law, the mandates and practices of correctional institutions, and cultural imaginaries and representations. Yet, as will be detailed, risk assessments draw from these histories and (re)enact them. They pull from, for example, past criminal convictions, prison-based reports of institutional conduct, and scholarly accounts of aggregate recidivism rates and inscribe these histories in the present – and into the bodies of penal subjects. In the process, assessments participate in producing a paroled subject who is inherently risky in the here and now.

Importantly, this is not to gloss over the fact that both risk instruments and parole personnel recognize variations in risk across individuals. Further, it must be noted that criminological theories, media representations, and cultural understandings (including those of penal policy makers and parole field personnel) sometimes acknowledge the possibility of a penal subject who is not dangerous. Think, for instance, of discussions about wrongfully convicted individuals. Yet, attending to the performative effects of risk assessment within parole highlights that these devices preclude the possibility of a non-dangerous subject. While they allow for differences in risk, they constitute everyone as risky – as a threat to public safety – to one degree or another.

In what follows, I first briefly trace the proliferation of risk assessment technologies within the penal realm and then review the concept of performativity. The next section, drawing from previous scholarship and from data collected by the author, details the specific ways in which risk assessments operate performatively. While noting that this happens through shaping the beliefs of parole authorities, that is, through ideological channels, I focus attention on the ways in which assessments produce ontological effects that do not depend on beliefs but instead operate through institutional, bureaucratic, and automatic mechanisms. The conclusion reviews the conceptual shifts in understandings of risk that this article proposes. I then close by suggesting that the current arguments are relevant to risk practices throughout the legal and penal realm which have, to this point, contributed to organizing and legitimizing a vast penal network.

The Proliferation of Risk Governance

A wide range of scholarship engages with risk and its proliferation in our contemporary moment. ‘Risk society’ scholarship, for instance, traces how concerns about risk, efforts to profile risk, and calls for reducing, avoiding, or otherwise managing risk are fundamental in late modern governance and subjectivities (Beck, 1992; Giddens, 1991). The increasing prevalence and importance of risk technologies in the governance of crime has been well-documented (Cohen, 1985; Carlen, 2008; Castel, 1991; Ericson and Haggerty, 1997; Feeley and Simon, 1992, 1994; Garland, 2001; Hannah-Moffat, 1999, 2013; Harcourt, 2007; Kemshall, 2011; Kemshall and McGuire, 2001; O’Malley, 1998, 2004; Simon, 1993; Rose, 1998, 2002; Robinson, 2002).

Focusing on the penal realm, the last several decades have witnessed a turn toward technocratic, ‘evidence-based’ risk governance. Contemporary risk assessments rely less on the ‘clinical’, ‘subjective’, or ‘informal’ judgments of legal and criminal justice actors and more on standardized, algorithmic, validated, and actuarial methods of assessment (Castel, 1991; Feeley and Simon, 1992, 1994; Hannah-Moffat, 1999, 2013; O’Malley, 1998, 2004; Rose, 1998; Simon, 2005). The California parole agency was among the vanguard of this movement as it transitioned, in the early 1990s, to a risk management approach that embraced actuarial assessments (Simon, 1993). Since this time, standardized and often actuarial procedures for evaluating and classifying individuals’ risk have become ubiquitous and hallowed components of penality. The US Justice Department’s National Institute of Corrections now encourages the use of risk and needs assessments at every stage of the criminal justice process (Angwin et al., 2016). Further, an increasing number of political and academic voices contend that such instruments represent one of the most promising avenues for reducing mass incarceration or avoiding it in the first place (e.g. Bonta and Andrews, 2007; Harris, 2006; Milgram, 2014).

A considerable amount of attention has been devoted to the effects of this turn toward ‘risk management’, including critical scrutiny. In particular, various authors have challenged the accuracy and objectivity of risk assessment (Harcourt, 2007; Hannah-Moffat, 2013; Hannah-Moffat et al., 2010; Kemshall, 2011). For instance, authors contend that, rather than objectively measuring risk, assessments (re)produce racial biases, especially against African Americans, thereby intensifying already existing racial imbalances in correctional populations (Angwin et al. 2016; Gonzalez van Cleve and Mayes, 2015; Goddard and Myers 2016; Harcourt, 2007, 2010). Others note that actuarial assessments rely on aggregate data and averages and, hence, engage with an abstracted and decontextualized subject (Hannah-Moffat, 2013; see also Castel, 1991; Foucault, 1977; Harcourt, 2007; Hannah-Moffat et al., 2010). As ‘risk variables cannot be easily abstracted from the socio-political, economic and cultural specificity of individuals’ (Hannah-Moffat, 2013: 144), this calls into question the validity of risk assessment. While drawing from this work, I do not focus on how objectively or accurately assessments measure risk. Rather, I call attention to the ways in which they produce objective social effects in the penal realm. In particular, risk assessments produce a subject who must be risky and contribute to perpetuating the risk of this subject throughout the period of parole.

We know that assessments influence penal practices and the decisions of penal personnel. For instance, scholarship has documented that individuals assessed as higher risk have higher rates of parole revocation (Grattet et al., 2011; Petersilia and Turner, 1993). While this could reflect that high-risk individuals violate rules and commit crimes at higher rates and, thus, that assessments are accurate tools for identifying risk; scholars have contended this may be reflective of a ‘distortive effect’ produced by these tools (Harcourt, 2007: 12; see also Goddard and Myers, 2016; Grattet and Lin, 2016; Hannah-Moffat, 2013; Kemshall, 2011; Petersilia and Turner, 1993). That is, risk assessments can influence the environment they attempt to inform.⁴

Scholarship hypothesizing this effect has largely focused on how this occurs through shaping the beliefs of penal actors. For instance, when someone is assessed as ‘high risk’, parole authorities may believe this assessment to be accurate, leading them to expect problematic behavior and, as a result, impose additional rules and supervision. Doing so creates additional opportunities for discovering noncompliance which, of course, then serves to confirm the risk that has already been assessed (Harcourt, 2007; Petersilia and Turner, 1993; see also Grattet and Lin, 2016). Put another way, risk assessments may create a recursive circuit – a self-fulfilling prophecy (see Merton, 1948) – where paroled subjects’ previously assessed risk tends to be searched for, confirmed, and sustained over time.

Yet, establishing a self-fulfilling prophecy is contingent upon shaping beliefs (Callon, 2006). It occurs if penal personnel believe, and act as if, risk assessments are true, thereby recursively linking beliefs and practices. However, assessments are sometimes found suspect or unconvincing by penal authorities (Bullock, 2013; Hannah-Moffat, 2013; Ibarra et al., 2014; Kemshall and McGuire, 2001; Robinson, 2002). In particular, within the California parole agency, despite promotion of a risk management approach, many field personnel viewed actuarial tools as suspect or inaccurate and found ways to carve out a niche for their subjective knowledge and intuition (Lynch, 1998; Werth, 2017). In situations such as these, risk technologies fail to shape beliefs, and the creation of a self-fulfilling prophecy is forestalled. Yet, this raises a question: In these cases are risk assessments unimportant presences that exert no effects? Or, do they impact penality in other ways?

Several authors have suggested that risk assessments produce effects through mechanisms other than beliefs. For instance, in a review of scholarship on risk, Hannah-Moffat (2013) contends that risk assessments impact institutional practices and that potential risk is often translated into ‘administrative certainty’ (p. 139). Further, Kemshall (2011) suggests that ‘systems and processes for risk assessment and management can also produce risk’ (p. 223). While these works acknowledge the potential for risk to exert effects independent of beliefs, the mechanisms at work remain unspecified, under-explored, or tied to the realm of beliefs. For instance, the examples provided by Kemshall focus on errors in calculation or doubts harbored about the accuracy of risk instruments; in both examples, the effects of assessment are dependent on the (either mistaken or skeptical) beliefs of penal actors.

Harcourt (2007), by contrast, specifies an institutional pathway through which actuarial risk assessments may influence penal practices independent of beliefs: They can exert a ‘racheting effect’ through triggering additional supervision even when actors do not necessarily trust them. While drawing from Harcourt’s arguments, I contend that risk assessments do more than influence penal practices, they play a role in constituting the penal realm. Further, I argue that this applies to both actuarial and non-actuarial assessments, both of which inscribe a certain amount of risk/dangerousness into all paroled subjects, thereby establishing an ontological order that fundamentally structures penal practices. Even in situations where parole personnel find risk instruments unconvincing, a subject who must be risky to some degree has been constituted, classified, and placed on parole. Thus, in a surprising turn, these risk assessments become objective knowledge; not objective in the sense of accurate or true, rather they become objective by objectifying – institutionally materializing – their effects in the penal realm.

Discourse Matters: Considering the Performative

In Doing Things with Words, Austin (1962) troubles the idea that the function of language is primarily representational, that it merely describes an outside reality. Instead, Austin distinguishes between statements that describe a world (constantive utterances) and statements that act on the world and contribute to constituting it (performative utterances). He identifies two forms of performative statements: illocutionary and perlocutionary. Performative statements that are illocutionary produce immediate effects in the world; that is, they constitute reality through their announcement. Here, saying or signifying something is enacting what is spoken or signified. This enactment happens by way of convention and is immediate; the saying is simultaneously the enactment of effects in the world. By convention, Austin is referring to an intersubjective dimension undergirded by ritual, custom, ceremonies, or legality. For instance, when a judge marries two people, the statement ‘I marry you’ is not descriptive of an outside or prior reality, rather, the speaking of this statement inaugurates and brings to the world the reality that two persons are now married (Austin 1962: 8). Similarly, when a judge or jury issues a verdict in a criminal trial, the enunciation ‘I convict you’ enacts what has been spoken: it creates a legally convicted offender.⁵ By contrast, perlocutionary statements also impact the world, but by triggering a set of consequences. Here, the saying and consequences are temporally distinct. Perlocutionary statements do not rely on convention, instead they operate via consequences, and the ‘consequences are not the same as the act of speech, but are, rather, what we bring about or achieve by saying something’ (Butler, 1997a: 109).

More recent work – drawing from semiotics and post-structuralism – has focused attention on the ability of language to bring particular kinds of subjects into being and structure their social existence. For instance, Butler (1997a) focuses on the performativity ability of hate speech to constitute subordinated subjects. Foucault (1977), although not directly utilizing the concept of performativity, calls attention to the constitutive power of language by tracing how discourses – which encompass both pronouncements and practices for Foucault – produce the deviant subject as a pathological individual in need of correction. Pasquino (1991) notes that this subject, homo criminalis, represented a new ‘object personage’ (p. 236) that was constructed as different from, and more dangerous than, homo economicus. As such, Pasquino calls attention to the ways in which discourses shape reality by producing their own objects (see also Rose, 1999; Smandych, 2000), and through structuring how these objects are governed. For instance, Rose (1998: 185) observes that psychiatric diagnoses operate performatively through producing specific regimes of treatment.

As noted, several of Austin’s examples of performative speech focus on the ability of the law to produce – rather than describe – reality. The ‘constitutive approach’ within law and society scholarship resonates with this insight; it argues that law not only regulates social interaction, it participates in constituting social reality. Legal terminology and categories are not merely descriptive of reality, they shape reality and how actors engage with it (Geertz, 2000 [1983]; Conley and O’Barr, 1998; Pavlich, 2014; Swennen and Croce, 2016). Bourdieu (1987) contends that the law operates performatively through its power to name and enact reality. For him, legal decisions represent the quintessential form of performative utterance, as they are decisions formulated by authorized agents acting on behalf of society; these decisions ‘are magical acts which succeed because they have the power to make themselves universally recognized’ (p. 838). Of course, this does not mean that all legal acts are universally accepted or embraced, rather it highlights that legal acts (e.g. passage of a law or a court decision) are socially recognized as establishing what is legal and illegal.

This article contends that, like law, risk assessments and classifications operate performatively. They not only describe and predict reality, they constitute reality. Through drawing upon convention – including legal rituals (e.g. criminal convictions) and preexisting cultural tropes – they produce all individuals on parole as risky subjects who pose a threat to public safety.⁶ Further, like law, they have the power to make this reality universally recognized. This does not mean that risk assessments are always trusted by penal authorities or by the public. Rather, it means that acts of assessments produce a reality that must be and is recognized by penal authorities: A dangerous paroled subject becomes an institutional ‘fact’ and certainty. In this way, risk practices operate in what Austin terms an illocutionary performative manner: They immediately construct reality. But they also operate performatively in a perlocutionary way through triggering consequences that, over time, perpetuate or even intensify the ‘reality’ of this risky paroled subject.

It bears noting that the arguments in this article draw from and resonate with influential scholarship on the sociology of deviance (e.g. Becker, 1963). At the same time, bringing performativity into the conceptual picture, as well as insights from science and technology studies and actor-network theory, deepens our ability to understand the operation of risk techniques. First, scholarship in this vein attends to the ways in which objects, devices, and ‘technicalities’ intersect with and impact ‘social’ phenomena, such as labels, stigma, beliefs, and practices (see Ballestero, 2015; Latour, 2005; Riles, 2005). Second, the concept of performativity alerts us to consider how discourses and technologies (of risk) may operate through mechanisms other than beliefs.

Assessment as Performative: Constituting, Confirming and Sustaining Risk

This section details the specific ways in which practices of classification and evaluation within parole operate performatively. Before doing so, however, I first briefly review the methods of assessing risk utilized within the California parole agency, the Division of Adult Parole Operations (DAPO). Evaluating, predicting, and classifying the risk of individuals on parole displays variety and contingency; parole personnel draw from multiple sources of information and this process varies across actors and temporal moments (Werth, 2017). At the same time, there are standardized procedures and instruments for assessing risk within DAPO that are routinely utilized. First, all individuals on parole are assessed by the California Static Risk Assessment (CSRA), a ‘second-generation’ actuarial risk assessment tool that utilizes a variety of static risk factors, such as age at time of release, gender, and the number of prior felony convictions, to predict individuals’ risk of reoffending.⁷ It rates individuals as low, medium, or high risk in three categories: property reoffending, drug reoffending, and violent reoffending. Second, as of 2013, another actuarial instrument was available to DAPO personnel: Correctional Offender Management and Profiling Alternative Sanctions (COMPAS), a risk and needs assessment tool. Yet, I focus on COMPAS only minimally, as fieldwork showed parole personnel used it to examine needs only, and not risk, and because it was utilized much less frequently than the CSRA. Third, DAPO assigns everyone into one of six possible parole supervision levels.⁸ While institutionally standardized, this is not an actuarial assessment. This classification is completed by parole personnel and is based on a review of a person’s official file (the archive of state-produced documents), an in-person interview with the person prior to release from prison, their CSRA risk score, and their ‘criminogenic needs’ as identified in COMPAS and the interview. While supervision levels are related to the practicalities of supervision – they mandate a minimum number of meetings between agent and parolee – they operate as a de facto form of risk classification, as the logic undergirding them is that individuals with more risk and needs will be assigned to more intensive supervision levels.

In addition to, and actually in conjunction with, these institutionally mandated forms of assessment, parole personnel form opinions and judgments about individuals’ character, morality and, ultimately, risk (Ibarra et al., 2014; Lynch, 1998; McNeill et al., 2009; Werth, 2017). Through, for instance, reviewing the correctional file and interacting with individuals, parole personnel develop what we have traditionally termed subjective or clinical judgments. These impressions and judgments are highly consequential. In fact, they can and often do operate performatively in a perlocutionary manner: They unleash consequences that establish a self-fulfilling prophecy (see Petersilia and Turner, 1993; Harcourt, 2007). Coming to believe that someone is ‘high risk’ can lead to the imposition of additional supervision and rules and shape how parole personnel interpret individuals’ demeanor and character; it can render individuals’ conduct legible through the lens of their previously evaluated risk. Yet rather than proceed in this direction, this article focuses instead on the ways in which risk practices operate through institutional and automatic mechanisms independent of beliefs.

Beyond Beliefs: Classifying and Inaugurating the Risky Paroled Subject

I begin with the initial moment, and technique, of classification within parole: being placed on parole. Placing someone on parole – that is, classifying and announcing someone as a parolee – inaugurates a paroled subject. While noncontroversial, it is worth pausing and unpacking the forces at play in the moment of being classified as a parolee. I contend that this act of classification does not merely describe its subject, rather it creates a particular subject (for the parole agency and society): A subject who is dangerous, who represents a threat to society.

Further, this act of classification simultaneously places this dangerous subject into a specific socio-material position. That is, parole linguistically, and through immediately enacting convention, performatively constitutes the risky individual subject to parole supervision. While additional techniques, such as the CSRA, will be deployed to evaluate individuals, there is an automatic, de facto ‘evaluation’ taking place in announcing someone as a paroled subject: The moment when a formerly incarcerated individual becomes a parolee is simultaneously a moment of recognizing them as a threat to public safety in need of parole supervision.

This highlights the interconnections between parole and other legal and penal practices, in particular those occurring in the courtroom and the prison. The inauguration of the risky paroled subject is not an ‘originary moment of subject formation’ (Butler, 1997b: 32), rather it is one of the moments of producing and reproducing penal subjects as risky. In fact, being announced as a parolee is a rearticulation of what is already ‘known’ about legally convicted and imprisoned subjects: They are dangerous individuals requiring state intervention. In fact, parole is successful in performatively constituting all parolees as dangerous precisely because it reiterates, in a different institutional setting, what already occurred in the courtroom and prison: The announcement of an offender, ‘a convict’, who is a threat to the public and thereby requires penal intervention.

This rearticulation is the realm of convention that Austin (1962) refers to. The convention being enacted here entails previous legal and institutional ceremonies and practices (e.g. presentence reports, legal conviction of a felony, reports of in-custody conduct) as well as cultural imaginaries of ‘offenders’. Authors have traced ways in which penal subjects are represented, imagined, and produced as risky others (Foucault, 1977; Melossi, 2000; Pratt, 2000). In particular, Foucault (2007) has traced a path from legal ceremonies to cultural imaginaries of dangerousness and difference. He documents how persons convicted of a criminal offense are doubled as abnormal subjects through legal, penological, and psychological discourses about the accused. These cultural imaginaries, and the legal–penal rituals, mandates, and practices from which they spring, are the realm of convention that parole draws from and (re)activates in performatively producing the risky paroled subject. Through drawing upon this ritualized and ‘condensed historicity’ (Butler, 1997a: 3), parole announces the same risky individual but in a new context. That is, parole classification does not invent a risky person, rather it creates and enacts a subject who must still be risky.

Further, performativity calls attention to the ways in which language, discourse, and representation are intertwined with material effects (Bourdieu, 1987, 1992; Butler, 1997a). As Butler (1997a: 17–18) notes, performative speech acts do not just act upon listeners, they contribute to the social constitution of individuals and the structural relations in which they operate. In the case at hand, classifying someone as a parolee automatically places that person in a particular socio-material arrangement: one that entails stigma and subordination. Parole entails a sovereign relation between state and paroled subjects, who are required, for instance, to follow all instructions from agents and who face the continual specter of reimprisonment. Further, enlivened by both reformist and incapacitative penal logics, individuals on parole are conceived of as defective subjects; as persons currently incapable of ethical self-governance and hence in need of practical and moral regulation from the parole agency (Turnbull and Hannah-Moffat, 2009; see also Rose, 2000; Werth, 2013).

Thus, announcing someone as a parolee discursively constitutes a subject, and this discursive constitution is ‘inextricably bound to the social constitution of the subject’. (Butler, 1997a: 154). In particular, being classified as a parolee operates performatively by constituting a subject who is risky, defective, and in need of state oversight and, simultaneously, by subjecting that person to parole supervision. Importantly, like law – and in conjunction with law – this classification is universally recognized. It is semiotically and materially successful in institutionally inaugurating a risky paroled subject. Even if parole personnel think that someone should not be on parole – because, for instance, they believe the person was falsely convicted or they think that they no longer represent a threat to public safety – a presumptively dangerous subject has been constituted and a regime of parole governance enacted.

Specifying, Ranking, and Reproducing Risk

While the moment of being classified as a parolee entails the socio-material constitution of a definitely risky subject, the risk that this subject represents remains somewhat abstract; the specific content and degree of risk remains unknown. Additional assessments – including the CSRA risk score, parole supervision level, and more impressionistic judgments – are utilized by parole personnel to specify or elaborate on the type and amount of risk posed by individuals.

These risk techniques unquestionably operate in descriptive ways: They attempt to identify and narrate the risk that individuals pose. At the same time, I contend they also operate performatively. In particular, both the CSRA and parole supervision level performatively (re)produce the parolee as an inherently risky subject. As previously noted, the CSRA both quantifies and qualifies risk: It designates the amount (low, medium, or high) and type of risk (property, drug, or violence reoffending). Importantly, specifying a person’s risk – either the kind or the quantity – reinforces their risk to one degree or another. Although perhaps unsurprising, it is noteworthy that there does not exist a category of ‘no risk’. Of course, from an actuarial or probabilistic perspective, it is hard to imagine someone who poses zero risk of criminal offending. Yet, this highlights that the CSRA marks everyone on parole as risky to some degree; it makes a non-dangerous parolee an institutional impossibility.

Unlike the CSRA, parole supervision levels do not specify the type of risk, rather they operate via the hierarchization of risk and needs.⁹ More intensive supervision levels, which mandate more frequent meetings and additional parole conditions, institutionally signal a riskier subject. Importantly, while ostensibly descriptive – in that this level reflects existing risk and needs – this classification is performative in an illocutionary way as it immediately calls groups into existence (see Bourdieu, 1992). Individuals classified as ‘intensive reentry’ (one of the most intensive supervision levels) are immediately marked – that is, constituted – as individuals overflowing with risk/needs and therefore requiring close supervision. Further, analogous to the CSRA, each of these groups denotes a certain level of risk. Someone classified as ‘personal and subsistence care’ (the least intensive level) is presumed to represent less risk/needs than someone classified as intensive reentry. Yet, both levels (re)produce a subject who is risky to some degree and requires some level of parole supervision.

That is, the differences ranked by the CSRA and supervision levels are differences in degree, not kind. Regardless of the specific type or amount of risk, these evaluative acts report on, and therefore produce, a subject who must be risky to some degree. In the process, these evaluative acts function as corroboration of the condition of inherent risk that being announced as a parolee inaugurates. It bears noting that in many cases, parole authorities think CSRA risk scores and supervision levels are accurate reflections of individuals’ risk; that they are ‘on the money’ or ‘pretty right on’. At other times, they think these devices ‘miss the mark’ and do not depict the actual risk posed. Yet, in either case, these devices of risk assessment and classification act in a way that makes a non-risky parolee bureaucratically inconceivable.

A discussion I had with Xavier, a parole agent, highlights this point. We were discussing a particular parolee on his caseload, Ted. Xavier told me that the CSRA risk score rated Ted as low risk in all three categories of reoffending. And Ted’s supervision level was ‘case management supervision’, a level that reflects low or moderate risk and needs. Xavier further explained that he views Ted as ‘very low risk’ and as someone who ‘…is doing good. He’s got a job, keeping his nose clean. I don’t think he’s going to mess up’. Yet, Xavier went on to note, ‘but you know, regardless, I gotta meet specs. Show that I’m doing my job. And at the end of the day, you never know, right?’ Specs, shorthand for specifications, refer to a series of reports that agents must submit to their supervisor at the end of each month detailing their supervision activities for each person on their caseload. Xavier’s comments draw attention to a tension between Ted’s ‘very low’ risk and the institutional requirement that, regardless of this, Xavier supervise him regularly and diligently. Further, his comment that ‘at the end of the day, you never know’ highlights that although Xavier views Ted as less risky than many other parolees on his caseload, a certain amount of risk still adheres to/resides within Ted. Comments such as this one, where agents would note that even ‘low-risk’ individuals represent a potential threat of criminal offending, were common among the agents I conducted fieldwork with and serve to highlight that although some individuals are assessed, classified, and viewed as low(er) risk, even these individuals do not – and cannot – fully escape the specter of criminal dangerousness.

Unleashing Institutional Consequences

The previous two sections outline practices – announcing/classifying someone as a parolee and ranking/specifying risk via the CSRA and supervision levels – that operate performatively in what Austin terms an illocutionary way. These practices discursively construct parolees, automatically and immediately, as subjects who must be risky. This section, however, turns to the ways in which risk techniques operate performatively in a perlocutionary manner. Here, rather than immediately constituting reality, risk practices unleash consequences, later in time, that structure the material realities of parole supervision. These consequences entail the activation of automatic, institutional, and bureaucratically ritualized practices that can, and often do, operate independently of the beliefs of parole personnel.

First, even in cases where agents were skeptical or dismissive of the CSRA or supervision level, data show that they often acted as if these tools were trustworthy in order to protect themselves professionally (see also Hannah-Moffat et al., 2010; Werth, 2017). Thus, even when perceived by parole agents as inaccurate, these assessment tools shaped supervisory practices. This was especially common in cases where parole agents felt the CSRA risk score overestimated someone’s risk. In such instances, despite contentions that it might not be necessary, agents often imposed additional conditions or supervision in order to preempt questioning from supervisors or possible institutional censure if the parolee eventually reoffended.

As an example, one afternoon during fieldwork I was traveling with Luis, a parole agent, as he made ‘home visits’ to some of the individuals on his caseload. Most of Luis’ clients had been informed in advance that he was coming, as this avoided ‘wasted trips’ and having to go back multiple times to find someone at home. Yet our next stop, at the residence of Nick, was going to be a ‘surprise visit’. As became clear during our conversation, this stemmed directly from the CSRA risk score, despite Luis’ skepticism about this risk instrument. The CSRA had assessed Nick as high risk in both violent and drug reoffending. Luis explained that, in general, he did not ‘put a lot of stock’ in the CSRA. He saw it as a ‘blunt, generic tool’ that was less useful than his professional experience and intuition. Further, in this particular instance, Luis felt the CSRA risk score overstated Nick’s risk. While noting that Nick’s criminal history included instances of both drug and ‘minor’ violent offending, Luis felt that Nick had ‘kind of turned a corner’ and was ‘getting his life back on track’. Yet, even though he did not trust the CSRA, Luis explained that ‘if it [the CSRA] says someone is high risk, you have to show that you’re aware of that, that you take that seriously’. In this particular case, this resulted in Luis sometimes conducting a surprise visit or an extra visit (beyond the minimum required visits) and, importantly, documenting that he had done so even though he felt this was ‘kind of a waste of my time’. In situations such as these, even if parole agents did not believe that available assessments were accurate reflections of risk, these devices impacted how they supervised clients.

Second, setting aside the question of whether agents believe a particular assessment to be accurate or not, risk devices sometimes trigger automatic institutional practices and/or common bureaucratic rituals. In particular, being classified into a more ‘intensive’ supervision level automatically triggers heightened surveillance. As previously noted, more intensive supervision levels institutionally mandate more frequent meetings with parole agents. And those placed in the ‘specialized caseload’ category are subjected to electronic GPS monitoring and additional rules (e.g. geographic restrictions). The specialized caseload category is utilized for individuals classified as ‘sex offenders’ as well as for some ‘high-risk gang members’. Unsurprisingly, in many instances, parole agents believed that individuals in this category represented a considerable risk and should be subject to intensive supervision, including electronic monitoring. Yet, there were a number of individuals in this category who were not perceived as especially risky, and in these cases agents contended that electronic monitoring was unnecessary or even wasteful of their time and resources. Nonetheless, everyone in this supervision level – both those who parole agents felt ‘needed it’ and those who did not – were subjected to heightened surveillance.

Further, while not automatic, CSRA risk scores often lead to ritualized, highly common (although not institutionally standardized) practices. For instance, when a new person was added to an agent’s caseload, if the CSRA denoted a high risk of drug offending, this almost invariably led parole agents to immediately impose drug testing as a condition of parole – prior to meeting the person or even reviewing their file. As Amanda, an agent, noted, ‘that’s just standard operating practice for me. If they’re rated as high, or even medium, risk of drugs, they’re gonna have to pee in a cup’. In this way, the CSRA triggered a pro forma, ritualized response even though agents had not yet decided if they trusted this evaluation. These examples highlight that independent of the beliefs of parole personnel, risk techniques within parole trigger institutional, automatic, and/or ritualized consequences that result in individuals assessed as higher risk being supervised in more intensive ways, which then makes confirmation of their risk more likely.

Lastly, it bears noting that unleashing these consequences holds the potential to not only confirm already established risk but to intensify it. Instituting additional rules or more intensive supervision creates conditions that favor ‘discovery’ of rule violations. And detecting such violations not only confirms risk, it can increase individuals’ perceived risk. Moreover, risk assessments can impact what happens upon discovery of a parole violation; they may influence whether someone receives a verbal warning, an intermediate sanction (e.g. subjection to a curfew), or is returned to prison. Data from fieldwork suggest that being evaluated as ‘high risk’ by the CSRA, especially for violent or drug reoffending, increased the likelihood that agents would recommend parole revocation and reincaceration following a violation. Notably, if a person’s parole is revoked, this will be factored into future risk assessments and, almost certainly, elevate their risk score when they are re-released onto parole. In this way, assessments not only tend to confirm and reproduce individuals’ risk, they hold the potential to intensify it.

Conclusion: (Re)conceptualizing Risk

Among proponents, the case for the usage of risk assessments is seemingly straightforward: These ‘evidence-based’ tools objectively identify and report on individuals’ risk. And the differentiation of risk (and needs) allows for more effectively organizing penal governance. For instance, assessments can identify ‘low-risk’ offenders who may be good candidates for alternative sentencing or early release from prison or, conversely, identify ‘high-risk’ individuals being released onto parole who are in need of enhanced surveillance or programming (see Taxman and Caudy, 2015). Yet, this article troubles this perspective – which views assessment as an objective technology of description and differentiation – and proposes some reconceptualizations in how we think about risk.

I argue that risk practices not only report on risk, they operate performatively through contributing to the constitution of a paroled subject who cannot be other than dangerous and through reproducing, or even intensifying, this subject’s risk throughout the period of parole. That is, the assessment and classification of risk is not merely a descriptive activity, it is a world making one. One of the ways risk practices make the world is through ideological channels, that is, through shaping the beliefs of penal authorities. But, as this article has shown, they also exert performative effects through automatic, institutional, and bureaucractic mechanisms independent of, and sometimes in opposition to, beliefs. We know that penal authorities engage agentively with risk tools and sometimes doubt or reject them (e.g. Kemshall and McGuire, 2001; Lynch, 1998; McNeill et al, 2009; Werth, 2017). Yet even in these cases, these devices institutionally and bureaucratially materialize effects in the penal realm.

In tracing these peformative effects, this article suggests the need to revisit, and possibly amend, the idea that contemporary risk practices are undergirded by a probablistic logic that embraces uncertainty (Castel, 1991; Rose, 1998; Simon, 2005; cf. Douglas, 1992). To be sure, clinical, subjective, and binary diagnoses of dangerousness – where someone is either dangerous or not – have been de-privileged by more standardized risk assessments oriented toward continuum-oriented thinking. Yet, rather than diminish the certainty of penal evaluation, I make the perhaps counterintuitive claim that current risk practices have contributed to making risk a technical and practical certainty.¹⁰ Contemporary continuum-oriented assessments certainly acknowledge and allow for imprecision, as a precise calculation of risk is difficult, but individuals must be placed somewhere on the spectrum. That is, reliance on continuum-oriented thinking links individuals to the constitutive element of the continuum: in this case, risk/dangerousness. Thus, in moving from a binary judgment (dangerous or not) to the logic of a continuum, risk becomes a condition of not just some, but all penal subjects (see also Pratt, 2000). While risk is sometimes delineated as distinct from dangerousness – dangerousness represents a quality of individuals, while risk is conceived as a probabilistic forecast about future hazard – assessments actually serve to blur this distinction.¹¹ The very process of attempting to forecast future risk necessitates evaluating the dangerousness of penal subjects as a quality that exists in the here and now. Thus, risk is more than a probabilistic logic, it is a condition – a quality – that makes up the paroled subject. Risk techniques not only forecast a possible future, they constitute and naturalize individuals as dangerous in the present moment.¹²

Further, taking up the lens of performativity highlights that although risk practices entail differentiation, they simultaneously homologize and homogenize paroled subjects. To be sure, assessments quantify, relativize, and hierarchize risk. Yet, I suggest that they simultaneously operate in a way that dilutes the very distinctions they produce (cf. Seeds, 2017). In the effort to measure risk, assessments homologize, as they link everyone evaluated to a common trait: criminality. That is, past criminal offending and ‘risk factors’ are converted into criminality, into a likelihood for future reoffending. And linking individuals to criminality in this way inexorably homogenizes them, as it makes them all risky to some degree. This is not to say they are all made equally risky, yet it highlights how the project of classifying, evaluating, and predicting differences in risk serves to mark – and essentialize – all assessed individuals as possessing some amount of risk/dangerousness.

Importantly, this is not to claim that risk assessments remove agency from penal authorities and turn them into mere ‘executants’ of penal policy and risk knowledges (Castel, 1991). Rather than remove agency, I contend that risk assessments set limits – help determine the institutional conditions of possibility – in which parole personnel encounter and agentively engage with clients. Risk techniques do not preclude the implementation of varied interventions, supervision styles, or decisions. But they do constitute, announce, and present parolees as subjects who are definitely risky and in need of penal intervention.

To be sure, the risky penal subject is not originated in parole assessment. Rather, recognizing the performative effects of assessment highlights the deep entanglements between parole and other legal and penal realms. The risky penal subject has a long, accretive history traversing legal and penal discourses, ceremonies, and practices. Yet, it is precisely this ritualized and sedimented history that assessments (re)enact as they constitute the paroled subject as an inherently risky person. Put another way, risk assessments certainly reflect already existing cultural assumptions and penal logics (that criminalize character and view convicted offenders as risky/dangerous), but they do more than reflect these logics, they produce and distribute them. Preexisting assumptions and understandings are the convention that assessments draw upon, but assessments play a role in creating and reifying conceptions of the parolee.

Of course, the idea that individuals on parole represent a potential risk to public safety is neither scandalous nor inaccurate. In fact, all citizens represent a potential risk of offending. Yet, contemporary practices of risk assessment do more than acknowledge the possibility of offending, they produce a paroled subject who must be risky; who is an always already dangerous being. And this represents a historical change, as state-based discourses in the early 20th-century framed parolees as already reformed individuals who did not represent a threat to public safety. It could be that today’s inherently risky paroled subject is a product of changed cultural ideologies, criminological theories, and/or media representations. Yet, while these cultural imaginaries play a role, they remain multivocal. Some make room for the idea of a penal subject who is not dangerous; for instance, for someone who was wrongly convicted or for someone who has ‘gone straight’, ‘found religion’, or otherwise reformed. Risk assessments, on the other hand, do not and cannot make room for this. In their everyday operations and administrative effects, they constitute everyone as risky; they preclude a non-dangerous paroled subject.

Further, risk practices operate in a way that ensure that risk adheres to individuals throughout the period of parole (see also Hannah-Moffat, 2013). Foucault (2007: 19) has observed how expert psychiatric evaluations produced for criminal trials often resulted in making individuals resemble their crime long before they committed it. In the current setting, it seems that assessments produced within parole can lead to individuals resembling their past crime – and the presumed criminality that accompanies this – long after they have committed an offense and been released from prison. And it seems probable that this makes it difficult for parole authorities, as well as the public, to imagine individuals on parole as anything other than risk-bearing bodies.

While I have focused here on assessment within parole, I suggest that risk techniques throughout legal and crime control arenas operate performatively and tend to reproduce their own logics. As already noted, a number of voices claim that risk and needs assessments can help mitigate or even end the era of mass incarceration. While sympathetic to the contention that risk represents a malleable logic that holds the potential to be deployed for multiple purposes (O’Malley, 2004), I would contend that risk assessment has, to this point, helped organize the penal state and fortify its legitimacy. Producing offenders and penal subjects as risky beings and reproducing them as such over time, undergirds and provides ideological support for incarcerating, supervising, regulating, and criminalizing a massive number of people,¹³ as well as for imposing an array of restrictive post-penal measures to more and more individuals. And the rise of this historically unprecedented legal–penal complex has occurred alongside, and in interaction with, the proliferation of risk knowledges, discourses and technologies.

Highlighting recognition of the power of words, the US Justice Department’s Office of Justice Programs, near the end of the Obama administration, announced a change in its lexicon: It would no longer use ‘disparaging’ terms such as ‘felon’ or ‘convict’ on its website to refer to individuals who were previously incarcerated and would instead use terms such as ‘person who committed a crime’ (Noble, 2016). This represents a notable – and promising – departure from longstanding penal and cultural practice. Yet, it needs to be situated within a larger context (including a new presidential administration). Risk assessments are routine and expanding techniques within law and crime control. As such, I would contend that even if this change in nomenclature serves as a clarion call and other correctional actors follow suit and abandon these ‘disparaging’ and historically dense terms, risk technologies and discourses are already doing much of the same labor that these words have done. That is, even if convict or offender cease to be officially pronounced, risk assessments operate in a way that keeps a criminalized and risky subject present in the penal field. The effects of assessment sustain this subject through creating a recursive, self-sustaining circuit: Risk scores and classifications produce a subject who must be risky. They also shape how this subject is governed, and they do so in ways that ultimately contributes to reaffirming the very notion of an inherently risky subject – who serves as the focus and warrant of the contemporary penal state.

Footnotes

Acknowledgements

A number of people have contributed significantly to this work. In particular, I am indebted to the parole agents and supervisors who allowed me to conduct fieldwork and who shared their time, experiences and insights with me. And I owe a special thanks to the individuals subject to parole supervision who did the same and who allowed another person to intrude into their lives. Andrea Ballestero, Reuben Miller, Ashley Rubin, Alessandro De Giorgi and Travis Linnemann provided thoughtful feedback on this article and improved it considerably. Elliott Currie, Mona Lynch, Kelly Hannah-Moffat, Jonathan Simon, Joshua Page, Phil Goodman, Judah Schept and Randy Myers provided valuable comments, encouragement and support. And thank you to Jasmine Lin, Madison Blattel, Sophia Newman and Jeremiah Murrell for their research assistance. Previous versions of this paper were presented at the Penal Boundaries Conference at the Centre for Criminology and Sociolegal Studies at the University of Toronto in 2016 and at the annual Law and Society Association conference in Seattle in 2016.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Science Foundation (0819562), the University of California-Irvine, and Rice University.

Notes

Appendix 1

Data for this article came from a long-term, multi-sited ethnographic research project on parole in California. Participant observation, interview, and archival data collection took place between 2008 and 2014: 16 months of continuous and immersive fieldwork during 2008–2010 and intermittent, but ongoing, fieldwork spanning 2011–2014. Participant observation and in-depth interviews were conducted with over 40 parole personnel (field agents and their on-site supervisors). I observed them as they supervised and interacted with clients, both in parole offices and in the homes and workplaces of individuals on parole. Fieldwork took place at nine separate parole units throughout the State. See Werth (2017: 4–7) for a more thorough discussion of the methods and data.

References

Angwin

Larson

Mattu

. (2016) Machine bias. Propublica 2016. Available at: https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing (accessed 3 July 2016).

Austin

(1962) How to do Things with Words. Oxford: Clarendon Press.

Ballestero

(2015) The ethics of a formula: Calculating a financial-humanitarian price for water. American Ethnologist 42: 262–278.

Beck

(1992) Risk Society: Towards a New Modernism. London: Sage Publications.

Becker

(1963) Outsiders: Studies in the Sociology of Deviance. New York: The Free Press.

Bonta

Andrews

(2007) Risk-Needs Responsivity Model for Offender Assessment and Rehabilitation. Ottawa: Public Safety.

Bourdieu

(1987) The force of law: Towards a sociology of the juridical field. Hastings Law Journal 38: 805–853.

Bourdieu

(1992) Language and Symbolic Power. Cambridge: Harvard University Press.

Bullock

(2013) The construction and interpretation of risk management technologies in contemporary probation practice. British Journal of Criminology 51: 120–135.

10.

Butler

(1997a) Excitable Speech: A Politics of the Performative. London; New York: Routledge.

11.

Butler

(1997b) The Psychic Life of Power: Theories in subjection. Redwood City, CA: Stanford University Press.

12.

California Department of Corrections and Rehabilitation (2016) Division of Adult Parole Operations. Available at: http://www.cdcr.ca.gov/Parole/index.html (accessed 4 March 2016).

13.

Callon

(2006) What does it mean to say that economics is performative. In: Mackenzie

DFM

Siu

(eds) Do Economists Make Markets. Princeton: Princeton University Press, pp. 1–58.

14.

Carlen

(2008) Imaginary penalities and risk-crazed governance. In: Carlen

(ed) Imaginary Penalties. Cullompton, Devon: Willan, pp. 1–26.

15.

Castel

(1991) From dangerousness to risk. In: Burchell

Gordon

Miller

(eds) The Foucault Effect: Studies in Governmentality. Chicago: The University of Chicago Press, pp. 281–298.

16.

Cavender

(1982) Parole: A Critical Analysis. Port Washington, New York: Kennikat Press.

17.

Cohen

(1985) Visions of Social Control. Cambridge: Polity Press.

18.

Conley

O’Barr

(1998) Just Words: Law, Language and Power. Chicago: University of Chicago Press.

19.

Dagan

Segev

(2014) Retributive whisper: Communicative elements in parole. Law and Social Inquiry 40: 611–630.

20.

Douglas

(1992) Risk and Blame: Essays in Cultural Theory. New York: Routledge.

21.

Drake

(2011) The ‘dangerous other’ in maximum-security prisons. Criminology and Criminal Justice 11: 367–382.

22.

Ericson

Haggerty

(1997) Policing the Risk Society. Oxford: Oxford University Press.

23.

Ewald

(2002) Risk in contemporary society. Connecticut Insurance Law Journal 6: 365–379.

24.

Feeley

Simon

(1992) The new penology: Notes on the emerging strategy of corrections and its implications. Criminology 30: 449–474.

25.

Feeley

Simon

(1994) Actuarial justice: The emerging new criminal law. In: Nelkin

(ed) The Future of Criminology. Thousand Oaks, CA: Sage Publications.

26.

Foucault

(1977) Discipline and Punish: The Birth of the Prison. New York: Panetheon Books.

27.

Foucault

(2007) Abnormal: Lectures at the Collège de France, 1974–1975. New York: Macmillan.

28.

Fourcade

Healy

(2013) Classification situations: Life-chances in the neoliberal era. Accounting, Organizations, and Society 38: 559–572.

29.

Garland

(2001) The Culture of Control: Crime and Social Order in Contemporary Society. Chicago: University of Chicago Press.

30.

Geertz

2000 [1983] Local Knowledge: Further Essays in Interpretitive Anthropology, 3rd ed. New York: Basic Books.

31.

Giddens

(1991) Modernity and Self-Identity. Cambridge: Polity Press.

32.

Goddard

Myers

(2016) Against evidence-based oppression: Marginalized youth and the politics of risk-based assessment and intervention. Theoretical Criminology 21: 151–167.

33.

Gonzalez Van Cleve

Mayes

(2015) Criminal justice through ‘colorblind’ lenses: A call to examine the mutual constitution of race and criminal justice. Law & Social Inquiry 40: 406–432.

34.

Grattet

Lin

(2016) Supervision intensity and parole outcomes: A competing risks approach to criminal and technical parole violations. Justice Quarterly 33: 565–583.

35.

Grattet

Lin

Petersilia

(2011) Supervision regimes, risk, and official reactions to parolee deviance. Criminology 49: 371–399.

36.

Hacking

(1986) Making up people. In: Heller

(ed) Reconstructing Individualism. Stanford, CA: Stanford University Press, pp. 222–236.

37.

Hannah-Moffat

(1999) Moral agent or actuarial subject: Risk and Canadian women’s imprisonment. Theoretical Criminology 3: 71–94.

38.

Hannah-Moffat

(2013) Punishment and risk. In: Sparks

(ed) The Sage Handbook of Punishment and Society. Los Angeles, London: SAGE, pp. 129–151.

39.

Hannah-Moffat

Maurutto

Turnbull

(2010) Negotiated risk: Actuarial assessment and discretion in probation. Canadian Journal of Law and Society 24: 391–403.

40.

Harcourt

(2007) Against prediction: Profiling, policing, and punishing in an actuarial age Chicago. University of Chicago Press, pp. 145–147.

41.

Harcourt

(2010) Risk as a proxy for race. In: Olin

(ed) Public Law and Legal Theory Working Paper Series. Chicago: The School of Law, University of Chicago.

42.

Harris

(2006) What community supervision officers need to know about actuarial risk assessment and clinical judgment. Federal Probation 70: 8–14.

43.

Ibarra

Gur

Erez

(2014) Surveillance as casework: Supervising domestic violence defendants with GPS technology. Crime Law and Social Change 62: 417–444.

44.

Kemshall

(2011) Crime and risk: Contested territory for risk theorising. International Journal of Law 39: 218–229.

45.

Kemshall

McGuire

(2001) Public protection, partnership and risk penality. Punishment and Society 3: 237–264.

46.

Latour

(2005) Reassembling the Social: An Introduction to Actor-Network Theory. Oxford: Oxford University Press.

47.

Lynch

(1998) Waste managers? The new penology, crime fighting, and parole agent identity. Law and Society Review 32: 839–869.

48.

Masco

(2015) Theater of Operations: National Security Affect from the Cold War to the War on Terror. Durham, NC: Duke University Press.

49.

Mcneill

Burns

Halliday

. (2009) Risk, responsibility and reconfiguration: Penal adaptation and misadaptation. Punishment and Society 11: 419–442.

50.

Melossi

(2000) Changing representations of the criminal. British Journal of Criminology 40: 296–320.

51.

Merton

(1948) The self-fulfilling prophecy. Antioch Review 8: 193–210.

52.

Milgram

(2014) Why Smart Statistics Are the Key to Fighting Crime. TED Talk. Available at: https://www.ted.com/talks/anne_milgram_why_smart_statistics_are_the_key_to_fighting_crime/discussion?utm_campaign=eNewsletter&utm_source=hs_email&utm_medium=email&_hsenc=p2ANqtz--vpSZPZVesvwK97FYJ0xiGWQZXfOWz7jit2tl6NBQOUbwDGRmjkbXkMayVGUZ-gk2DV3aN (accessed 4 March 2017).

53.

Noble

(2016) Justice Department program to no longer use ‘disparaging’ terms ‘felon’ and ‘convict’. The Washington Times, 4 May.

54.

O’Malley

(1998) Crime and the Risk Society. England: Ashgate Aldershot.

55.

O’Malley

(2004) Risk, Uncertainty and Government. London: Routledge.

56.

Pasquino

(1991) Criminology: The Birth of a Special Knowledge. In: Burchell

Gordon

Miller

(eds) The Foucault Effect: Studies in Governmentality. Chicago: University of Chicago Press, pp. 235–250.

57.

Pavlich

(2014) Criminal justice and Cape law’s persons. Social and Legal Studies 23: 55–72.

58.

Petersilia

Turner

(1993) Evaluating Intensive Supervision Probation/Parole: Results of a Nationwide Experiment. Washington DC: U.S. Justice Department.

59.

Phelps

(2016) Mass probation: Towards a more robust theory of state variation. Punishment and Society 19: 53–73.

60.

Pratt

(2000) Governmentality, neoliberalism and dangerousness. In: Smandych

(ed) Governable Places: Readings on Governmentality and Crime Control. Brookfield: Ashgate, Dartmouth, pp. 133–150.

61.

Riles

(2005) A new agenda for the cultural study of law: Taking on the technicalities. Buffalo Law Review 53(3): 973–1124.

62.

Robinson

(2002) Exploring risk management in probation practice: Contemporary developments in England and Wales. Punishment and Society 4: 5–25.

63.

Rose

(1998) Governing risky individuals: The role of psychiatry in new regimes of control. Psychiatry, Psychology, and Law 5: 177–195.

64.

Rose

(1999) Powers of Freedom: Reframing Political Thought. Cambridge: Cambridge University Press.

65.

Rose

(2000) Government and control. British Journal of Criminology 40: 321–339.

66.

Rose

(2002) At risk of madness. In: Simon

(ed) Embracing Risk: The Changing Culture of Insurance and Responsibility. Chicago: University of Chicago Press, pp. 209–237.

67.

Seeds

(2017) Bifurcation nation: Strategy in contemporary American punishment. Punishment and Society 19: 590–610.

68.

Simon

(1993) Poor Discipline: Parole and the Social Control of the Underclass, 1890–1990. Chicago: University of Chicago Press.

69.

Simon

(2005) Reversal of fortune: The resurgence of individual risk assessment in criminal justice. Annual Review of Law and Social Science 1: 397–421.

70.

Smandych

(2000) Introduction: The place of governance studies in law and criminology. In: Smandych

(ed) Governable Places: Readings on Governmentality and Crime Control. Aldershot: Ashgate, pp. 1–14.

71.

Sparks

(2001) Degrees of estrangement: The cultural theory of risk and comparative penalty. Theoretical Criminology 5: 159–176.

72.

Swennen

Croce

(2016) The symbolic power of legal kinship terminology: An analysis of ‘co-motherhood’ and ‘duo-motherhood’ in Belgium and the Netherlands. Social and Legal Studies 25: 181–203.

73.

Taxman

Caudy

(2015) Risk tells us who, but not how or what. Criminology and Public Policy 14: 71–103.

74.

Turnbull

Hannah-Moffat

(2009) Under these conditions: Gender, parole and the governance of reintegration. British Journal of Criminology 49: 532–551.

75.

Werth

(2013) The construction and stewardship of responsible yet precarious subjects: Punitive ideology, rehabilitation, and ‘tough love’ among parole personnel. Punishment and Society 15(3): 219–242.

76.

Werth

(2017) “Individualizing risk: Moral judgment, professional knowledge and affect in parole evaluations. British Journal of Criminology 57(4): 808–827.