Reconceptualising ‘risk’: Towards a humanistic paradigm of sexual offending

Within Western criminal justice traditions, the ‘risk’ paradigm has become the defining logic of contemporary laws and policies on sex offender management. This article critically examines the limitations of current technocratic and algorithmic approaches to risk in relation to sexual offending and how they might be addressed. Drawing on nearly two decades of theoretical and empirical research conducted by the author, it applies the learning on sex offender reintegration and desistance to advance a ‘humanistic’ paradigm of sexual offending. The paper attempts to counter some of the dangers of algorithmic justice and shift risk-based discourse away from its predominantly ‘scientific’ origins. It argues that such a move towards a more expansive and progressive version of risk within criminal justice discourses would better capture the realities of sexual offending behaviour and its real-world governance.


Introduction
'Risk' has become a dominant framework for conceptualising the social problems of advanced post-industrial societies. Scholars have noted the proliferation of the 'risk society' (Beck, 1992) thesis across a diverse range of fields, including science, economics, politics, health, technology, law and the environment. Despite its political and cultural currency, critics have questioned the ubiquitous nature of 'risk' and the utility of many of its foundational precepts. For example, it has been described as a rather nebulous concept which is 'slippery' (Kemshall, 2006: 60), 'opaque and disputed' (Mythen and Walklate, 2006: 3) and open to multiple interpretations. Scholars have similarly emphasised the 'negative turn' underpinning modernist discourses on criminal justice (O'Malley, 2004: 325) and the increasingly volatile and uncertain nature of risk regulation. In this vein, as Mythen and Walklate (2006: 3) have cautioned, it is prudent to be alert to the threat of 'risk imperialism' and the potency of risk as a potential 'blackout blind' (Mythen, 2014: 8) in obscuring the realities of crime or other social problems. This article furthers these debates by elucidating some of the blind spots of contemporary criminal justice policies on sex offender risk governance within Western societies.
Risk is said to be a 'mixed discourse' (see e.g. Sparks, 2001: 169) encompassing calculative and political as well as moral or emotive dimensions. This complex interplay is well illustrated within the domain of sexual offending where the evolution of technocratic expert-led approaches to risk has taken place alongside, and often in spite of, volatile and emotive community responses. 1 As Mythen (2014: 20-26) has highlighted, academic discourses on risk may be broadly sub-divided into realist and constructionist paradigms. The former is characterised by objective risk assessments and statistical modelling by technical experts as part of 'actuarial' or 'algorithmic approaches'. The latter recognises that institutional policies and social processes shape and are shaped by subjective knowledge about risk. This article is located at the intersection of these perspectives. It examines the salient gulf between meta-narratives on risk as a 'problem' as conceptualised within criminal justice frameworks; and risk as experienced by individualsincluding victims, communities and sex offenders themselvesin the course of everyday life (see generally Arnoldi, 2009). The term 'sexual offending' encompasses a wide variety of behaviours involving adults and children as both 'victims' and 'perpetrators', examples of which are drawn upon throughout the article.
While there is emerging literature around 'algorithmic' risk governance, discussed below, there is a dearth of critical work around the operation of this paradigm within the specific field of sexual offending. This article engages with this gap in the literature. The core argument is that the development and operationalisation of risk-based justice for sex offenders has been at the expense of the human dimensions of risk management both in terms of the nature of risk and the type of offender it presupposes. As explored further, the principal outworking of the algorithmic approach is two-foldit effectively precludes the notion of self-redemption for sex offenders and simultaneously diminishes the complexity of sexual offending behaviour and the importance of the social contexts of reintegration. The article ultimately makes the case for a more human and complex understanding of risk in this context as a much more protean and reflexive concept than current legal, political and academic framings of risk tend to suggest. In order to make these arguments, the article draws on key learning from academic discourses on sex offender reintegration and desistance (see e.g. Laws and Ward, 2011;McAlinden, 2007) including nearly two decades of theoretical and empirical research conducted by the author. 2 The article is structured as follows: Part I briefly outlines the theoretical underpinnings of contemporary risk-based policy on sexual offending within Western criminal justice traditions. Part II examines four of the core generic critiques suggested by risk scholars (see e.g. Mythen, 2014;O'Malley, 2004) as applied to sex offending policies: (a) the assumed homogeneity of risk, (b) zero-sum thinking in terms of the polarised identities of victims and perpetrators, (c) the focus on offender agency rather than the social aspects of risk management and (d) elitist modes of risk governance and decision making. In addressing these arguments, the third part of the article presents four key correlates as being central to a humanistic understanding of risk: (a) the diversity of risk, (b) risk as a continuum, (c) the importance of structure as well as human agency in managing sexual offending and (d) participatory approaches to risk governance. Finally, the conclusion identifies some policy and pragmatic challenges of this reconfigured risk narrative on sexual offending. Thus, this paper makes a scholarly contribution on two principal fronts. It furthers the dialogue between desistance and public protection practices and, more broadly, provides challenges to the technocratic and algorithmic epistemology of risk which is beginning to pervade risk-based scholarship and decision-making.

Risk-based policy on sexual offending
Within Western societies, contemporary criminal justice policy on sexual offending has emerged from three principal permutations of risk, culminating in a predominantly realist or scientific paradigm: First, the ascendancy of 'risk' was given impetus by the spread of 'neo-liberal' policies premised on the governance of 'security' (Loader and Walker, 2007) and the precautionary control of 'pre-crime' (Zedner, 2007). In the USA and the UK in particular, this has manifested as a 'community protection' model of sex offender risk management (Kemshall and Wood, 2007) based on selective incapacitation and postrelease surveillance. 3 Since at least the 1990s, a proliferation of measures to monitor and control the behaviour and whereabouts of sex offenders in the community, such as residence restrictions and registration and notification schemes, have been enacted within an overall rubric of promoting public safety (Jones and Newburn, 2013). 4 Second, contemporary risk technologies related to sexual offending have also evolved from what Feeley and Simon (1992) term the 'new penology'. These authors describe the emergence of penal trends across both Europe and North America based on the language of 'probability' and 'risk', where system objectives are tied to the efficiency of internal processes and targeted at certain groups of offenders on an aggregate rather than individual basis. As such, the core priority within much criminal justice risk-focussed scholarship and policy, including that related to sexual offending, remains one of 'managerialism'the overriding concerns are with efficiency and evidence-based decision-making, rather than rehabilitation per se (Silver and Miller, 2002). In this sense, the new penology is said to assume 'a kind of waste management function' (Feeley and Simon, 1992: 470) as the focus on regulation rather than transformation, means that sex offenders as a population are dehumanised as a toxic risk in need of control.
Third, the latter approach is also bolstered by a symbiosis with 'actuarial justice' (Feeley and Simon, 1994) and most recently 'algorithmic justice/risk governance' (see e.g. Hannah-Moffatt, 2018;Završnik, 2019). Predicted rates of recidivism for individual sex offenders, and subsequent risk management strategies, are now almost universally based on checklists of actuarial risk factors or 'structured professional judgment'. 5 Algorithmic justice utilises predictive algorithmswith an emphasis on 'big data' and correlations through forms of machine-based learningto produce a new paradigm in knowledge production regarding risk and crime (Christin et al., 2015). One of the supposed plaudits of the epistemological turn to algorithmic risk assessment is that it is scientifically objective or 'technical neutral' (O'Malley, 2004: 326) by virtue of '(semi) automated decision-making' (Morison and Harkens, 2019). As part of a broader trend within legal decision-making by police and courts originating in the USA, algorithmic tools are used to inform judgements about 'risky' groups, and to underpin resource allocation and the extension of surveillance and control. 6 Across the UK, this technocratic orientation of risk as the preserve of experts has become the hegemonic discourse of multi-agency public protection arrangements (MAPPA). The police, probation and prison services work collaboratively as the 'responsible authorities' to risk assess, manage and treat adjudicated sexual and violent offenders (Kemshall and Maguire, 2001). 7 More widely, algorithms are used not only to inform risk predictions and control strategies for sex offenders but also to detect and mitigate the risk of harm to child victims from suspected 'on-line sexual predators' (Cardei and Rebedea, 2017). Consequently, therefore, the gaze of risk regulation has more recently extended beyond the statistical probability of individual (re)offending to encompass broader concerns about risk of harm to classes of victims. As discussed further, however, the overreliance on or 'trust in numbers' (Porter, 1996) obscures both the cultural contexts of sexual offending and the cultural assumptions that often underlie regulatory policies.
Indeed, the dangers and limitations of the current algorithmic episteme of risk are beginning to be recognised within academic and legal discourses. 8 Commentators point to the imagined 'certainty of science and statistics' (Morison, 2020: 96) and 'a reduced human agency relegated to machines' (Završnik, 2019: 16). This prevents consideration of offenders as individuals and human agency as well as the contextual nature of sexual offending. As Mehozay and Fisher (2019: 523, 537) argue, algorithmic risk assessment results in a 'performative conception of humanness' which is 'a-theoretical' and 'non-reflexive'. Put simply, the reductionist and deterministic focus of algorithmic justice on recognising and controlling patterns of 'risky' behaviour obscures a fuller understanding of the causes and complexities of offending behaviour. Extending this analysis to the realm of sexual offending, within current technocratic approaches to risk, the humanistic element is eradicated not only from official modes of configuring risk but also from a more nuanced socio-political understanding of the aetiology of sex offending behaviour.
Discourses on sex offender reintegration and desistance (the process of slowing down or ceasing offending), based on 'strengths-based' philosophies (Burnett and Maruna, 2006), have developed in parallel to 'risk-based' approaches. Whereas risk-based approaches are centred on imposing external controls on the whereabouts and behaviour of known sexual offenders, 'strengths-based' approaches are based on developing intrinsic motivations for personal change. Some scholars have argued for a 'blended approach' (Kemshall, 2008: 126-127) to sex offender risk management, synthesising reintegrative/ rehabilitative and risk management strategies (see also McAlinden, 2007). For the most part, however, the rhetoric and philosophy of desistance and reintegration have tended to operate on the margins of therapeutic and regulatory intervention for sex offenders without challenging the status quo or altering the systemic philosophy of risk management. In the spirit of continuing what Weaver (2014) terms a 'dialogue' between desistance research and public protection practices, the analysis below furthers this discussion by charting some of the key tensions within this dialogue and how they might be resolved.

Deconstructing the 'risk' of sexual offending
Despite the supposed scientific certainty and objectivity of technocratic formulations of risk, contemporary criminal justice policies on sexual offending are in reality 'messy' and infused with a range of skewed, moralistic cultural assumptions about risk (see generally O'Malley, 2006). In order to advance the overall argument of advancing a new humanistic paradigm of sexual offending which better captures the realities of sexual offending behaviour, and extending the broad critiques of the risk thesis presented above, this section presents four of the principal deficits of the contemporary policy construction of risk related to sexual offending. While the first two relate to the extant vernacular and syntactic framing of risk, the third relates to criminal justice system objectives and the fourth to risk-based processes.

Homogeneity
A key criticism of the 'risk' hypothesis is the assumption that 'diverse configurations of risk can be collapsed into a single category (O'Malley, 2004: 323). Within public discourses, the homogenous framing of sexual offending around the 'predatory paedophile' and 'stranger danger' has been used to spearhead child protection campaigns in the USA, the UK and Australia (Mythen, 2014: 54). This cultural conceptualisation of risk is highly gendered, reflecting a 'hegemonic masculinity' (Cowburn and Dominelli, 2001: 399) wherein abuse at the hands of the random, adult, male 'stranger' is problematised over abuse within intra-familial contexts (see Kitzinger, 1999). Terms such as 'paedophile' and 'sexually violent predator' have become a site of convergence for popular conceptions of and policy responses to 'risk'. Most recently, for example, this is evidenced in the assemblage of public and private modes of policing in the USA, UK and Europe via 'paedophile hunting' (Campbell, 2016). This narrow cultural framing of risk has knock-on consequences for strategic criminal justice responses to sexual offending, where risk management technologies are reactive in nature and directed towards the small number of known 'high risk' sex offenders (or 'the critical few'). In fact, even highrisk offenders are not a homogeneous entity but have a diverse range of motivations and offending patterns (Weaver, 2014).
Feminist scholars have highlighted how this minimalist and one-dimensional interpretation of risk ignores the realities of domestic contexts (see, e.g. Cowburn and Dominelli, 2001;Newburn and Stanko, 1994). This skewed view of risk also subjugates a broader range of harms, including those posed by strangers as well as children and adolescents. While female sex offenders account for ∼5%-12% of sexual abuse (Cortoni et al., 2017), some commentators have suggested that peer-to-peer abuse comprises between one-third to two-thirds of overall levels of child sexual abuse (McAlinden, 2018;Radford et al., 2011). Although adult men continue to account for the majority of sexual offending against children, the growing statistical significance of a wider range of risks are largely obscured within public consciousness. In this respect, the clinical provision of specialist therapeutic programmes for such groups is ad hoc and highly inconsistent across the UK. Thus, there are clear territorial dimensions to the conceptualisation of 'risk' where the public-private divide is pivotal to both the cultural hypothesising and institutional ordering of risk (see Ashenden, 2003). In essence, the notions of 'risk' and 'vulnerability' are firmly allied with the public sphere and, consequently, 'safety' with the private domain of the family and home.

Zero sum thinking
The semantic framing of risk is also based on 'zero sum thinking' (O'Malley, 2004: 323) where the identities and interests of 'victims' and 'perpetrators' are seen as dichotomous and mutually exclusive and where one must be championed at the expense of the other. Within the process of policy formulation, there is evidence of what Tilly (2008: 33) has termed 'a double game of credit and blame' which bifurcates those deemed 'innocent' and 'culpable'. Child victims are perceived as the dialectical opposite of sexual perpetrators where the young, 'innocent' victim is juxtaposed as the 'antithesis' of the offender who is epitomised as 'wicked' or 'pure evil' (McAlinden, 2014). 'Memorial laws' (Valier, 2005), such as 'Megan's Law' in the USA, 'Sarah's Law' in the UK or 'Natalie's Law' in Germany, which call for community notification/public disclosure or harsher sentences for sex offenders, are named after deceased child victims. Despite the possible consequences of such measures for offenders if their identities become knownin the form of stigma or even vigilante attack -'the burden of risk' (Mythen, 2014: 55) is effectively shifted from victims to offenders. This argument also speaks to what Douglas (1992: 27) refers to as the 'forensic' function of risk in allocating blame and responsibility.
However, this assumed dichotomous distinction between those who pose a risk to others and those at risk from sexual offending is not always so logical or clear cut. In this vein, cultural and political discourses around 'risk' are dismissive of two important realities about sexual offending behaviour. One is that a significant proportion of child victims go on to become adolescent or adult sex offenders. 9 The other is that the dynamics of sexual offending are such that individuals may be both victims and offenders simultaneously. This is often the case, for example, in institutional settings, such as residential care homes or schools, where abuse or exploitation have become 'normalised' as part of the organisational culture. In peer-based settings in particular, the child may pose a risk to others as well as be at risk themselves (Firmin and Curtis, 2015;McAlinden, 2018). It is also evidenced in complex forms of sexual offending, such as 'child sexual exploitation', where children may be victimised at the hands of adults but also 'groomed' to recruit other potential victims (Ost and Mooney, 2013). In such circumstances, in becoming complicit in the victimisation of others, the child may be technically said to be both a 'victim' as well as a potential 'perpetrator' of child sexual abuse. This also cautions us to be mindful of the use of the often pejorative and partisan terminology of 'risk', such as 'victim', 'perpetrator' or indeed 'vulnerability', which is not always diametric, static, or even useful, in terms of identifying the 'dangerous' or those in need of protection.

Agency
The third problem with the contemporary risk paradigm in this specific context is that it accentuates 'agency' to the neglect of the 'human and experiential dimensions of risk' (see Mythen, 2014: 136). That is, the policy emphasis is on effecting control of individual sex offenders rather than the broader socio-structural contexts which impinge upon offending behaviour and personal motivations for change. In practice, the scientific calculation of risk tends to focus on individual offender proclivities, including static and dynamic risk factors (see generally, Beech et al., 2016), rather than the underlying relational or situational contexts of risk management. In fact, far from being 'predatory' or 'preferential', the majority of sexual offending is opportunistic or situational in nature (Farmer et al., 2016;McAlinden, 2012). As noted above, perhaps one of the most significant weaknesses of the 'risk-based' approach to sexual offending is that it is orientated towards 'control' rather than 'change' (Weaver, 2014) and in so doing, effectively disregards the possibility of offender redemption or the 'non-risky' individual (Werth, 2019). The upshot is that, in advocating regulation over rehabilitation, contemporary risk practices tend to cement the sex offender 'label' and thus confirm the identity of the individual as a 'sex offender'. This is often ultimately counterintuitive to risk management by effectively precluding the individual from 'moving on' from a sex offending past.
Findings from the small but growing body of studies on sex offender 'desistance' (see generally Laws and Ward, 2011;Harris, 2017) are beginning to filter into clinical criminal justice provision. By way of illustration, this has recently occurred within the context of a revised suite of sex offender treatment programmes delivered by HM Prisons and Probation Service in England and Wales. As part of a modification of the 'Risk-Needs-Responsivity' model (see Bonta and Andrews, 2007), these new accredited programmes have incorporated, inter alia, the importance of optimism and 'hope' within a future-focussed approach as a key component of the formation of a new non-offending identity (see generally, McAlinden et al., 2017). As a whole, however, desistance and reintegration focussed approaches to sexual offending have been confined to the periphery of criminal justice interventions and tend to be subsumed within the overall aims of risk management.

Elitist risk governance
Power differentials are fundamental in 'shaping communications about risk … and modes of risk governance' (Mythen, 2014: 138). The technocratic architecture of actuarial justice privileges elitist or expert forms of knowledge and decision-making over lay or participatory forms and the community are afforded only a minimal role in the processes of risk management. While the purported voice of the local community is instrumental to the formulation of 'populist' penal policy (Johnstone, 2000), in practice the community is most often distanced from routine decision-making. Within neo-liberal societies such as the USA and the UK, risk management is seen as the preserve of experts as the public is, for the most part, excluded as 'a potential site of risk' due to the threat of vigilantism (Kemshall and Wood, 2007: 207). However, as Miller (2013: 283, 293) observes, and as discussed further, while contemporary scholarship on crime and punishment generally treats democratic politics and crime policy as 'a dangerous mix'and the public as 'ill-informed, vengeful and easily manipulated'differently constituted democratic institutional arrangements might facilitate public participation in crime politics.
Within the context of public disclosure or community notification schemes, both information about and responsibility for sex offenders are 'shifted' to the general public (Simon, 1998: 463). Similarly, the introduction of circles of support and accountability (COSA), first in Canada and then elsewhere, involves the local community working in partnership with statutory and voluntary agencies to aid sex offender reintegration. However, despite their rhetorical commitment to building effective relationships with the community, international evaluations are predominantly focussed on recidivism rather than psycho-social outcomes (see e.g. Clarke et al., 2017). This is particularly the case in England and Wales where the overall direction of schemes is systemically managerial, rather than organically reintegrative (Thompson et al., 2017). 'Lay advisers' are included within MAPPA in England and Wales, where they are consulted in the review of MAPPA functions but afforded no role in the risk management of individual cases. 10 Thus, for the most part, such seemingly inclusionary forms of risk control are little more than gestures to 'nodal' (Shearing and Wood, 2003) or 'networked' modes of governance (Crawford, 2006), so that the regulatory state is, in reality, agnostic towards the involvement of the local citizenry. The community and collective risk management strategies, however, are much more pivotal to the broader processes of sex offender reintegration than officially acknowledged. The pertinent issue becomes, therefore, how 'smarter' (Gunningham and Grabosky, 1998) forms of sex offender risk regulation might be fostered and implemented.
In highlighting these four deficits, the foregoing analysis has established that the contemporary technocratic formulation of risk related to sexual offending neither captures the diversity of risk; the identities of 'victims' and 'perpetrators', nor indeed the potential cross-over between them; the situational context of much of sexual offending; and, consequently, the central importance of engaging the community in longer term or 'secondary' (Maruna and Farrall, 2004) desistance. With each of these quandaries in mind, the next section endeavours to map out a more expansive, progressive version of risk which addresses the lived experiences of victims, offenders and communities.

Towards a humanistic paradigm of 'Risk'
Developing a new normative, humanistic paradigm of 'risk' entails reformulating the language, objectives and processes of contemporary risk governance. This reconceptualised paradigm, which addresses each of the core critiques set out above, would comprise four central pillars. The first two presented below relate to a different way of thinking and talking about risk governance, and are framed as embracing the diversity of risk and a continuum of risk; the third, relates to the overriding purpose of the criminal justice system, as being transposed to longer-term reintegration rather than short-term risk management; and the fourth relates to the augmentation of self-governing regulatory processes which are based on participation and which meaningfully include the voices of victims and the local community.

Diversity
Recognising the plurality and diversity of risk entails political and cultural understanding of three core facets of sexual offending behaviour: that it transverses age and gender boundaries and may stem from females as well as children; that it most often occurs in private intra-familial or quasi-intra-familial contexts rather than public settings 11 and that manifestations of 'risk' and 'harm' occur in much broader contexts than contact forms of sexual abuse and may also stem from non-contact behaviours such as 'sexting' (McAlinden, 2018) and pre-abuse behaviours such as 'grooming' (McAlinden, 2012). The significance of these particular features is that, due in part to the core cultural assumptions about risk presented above, all are more difficult to pinpoint as 'risky' prior to actual harm. More broadly, therefore, at the level of polity, a humanistic understanding of sexual crime also challenges us to focus on 'risky' behaviours rather than simply 'risky' individuals (O'Malley, 2004: 118-19) and to balance offencefocussed with offender-focussed strategies. This extension of protective knowledge also has the potential to increase public safety and improve risk governance by moving beyond the 'critical few'.
The principal outworking of this advocated approach to 'risk' is the need to think more constructively about the terminology which underpins public discourses on sexual offending as well correctional policy and forensic practice. In the main, the use of the labels 'paedophile' and even 'sex offender' are ultimately unhelpful as they tend to promote an atypical version of risk. Such labels have knock-on negative effects for how both the public (Imhoff, 2015) and professionals (Willis, 2018a) perceive those previously convicted of sexual offences by reinforcing the fallacious belief of 'once a sex offender, always a sex offender'. In fact, it is well documented that sexual recidivism rates are low in comparison to other types of offender and that the base rate of sexual re-offending generally declines with time spent offence free in the community (Hanson et al., 2018). Moreover, such labels may also serve to deflect attention and resources from other forms of sexual offending behaviour which do not fit neatly within the prevailing paradigm.
A reconceptualised phraseology of risk would acknowledge the possibility of offender redemption and community re-integration and ultimately allow individuals to forge new personal and social identities free of an offending past. As desistance scholars are beginning to recognise, morally neutral and non-judgmental terminology should include instead 'person-first language' (Willis, 2018b) such as 'individuals with a history of sexual offending'. Consistent with the notion of 'reintegrative shaming', this involves 'decertify[ing] the offender as deviant' (Braithwaite, 1989: 101) by separating individuals from their behaviour and recognising the 'evil' of the act rather than the 'evil' of the person (see also McAlinden, 2007). Furthermore, rethinking the dialectics of 'risk' also entails re-evaluating how we perceive 'victims' of sexual crime and where we deem 'vulnerability' and 'risk' to lie.

Continuum of risk
A move away from labelling individuals based on their offending past also means recognising that there is a continuum of riskthat risk itself is not a static concept but rather fluid and subject to change over time. For young people in particular who display harmful sexual behaviour (HSB), a level of risk may be assigned after initial assessment within the criminal justice system. However, without periodic review and tailored and appropriate levels of therapeutic intervention, such risk assignations may become permanent carrying life-long labels and negative consequences as a 'sex offender'.
Further, recognising the mutability of 'risk' also means embracing what I have previously termed the 'victim-offender continuum' (McAlinden, 2014) that approximately half of children who are abused may go on to become perpetrators themselves (see Craissati et al., 2002;Hackett et al., 2013) and that in complex forms of sexual abuse or exploitation, children may be both 'victim' and 'perpetrator' simultaneously or oscillate between these core categories. For the most part, however, polarising discourses around risk mean that the child or young person who displays HSB may be deemed a 'perpetrator' at the investigation stage with account of their victimisation being left to mitigation by the courts. Moreover, there is also empirical evidence to suggest that professional evaluations of 'risk' related to sexual abuse or exploitation are sometimes highly gendered, with femininity most often being linked to victimhood and masculinity to offending behaviour (McAlinden, 2018). A more holistic assessment of the needs and circumstances of young people who display HSB means exercising 'professional curiosity'. This involves recognising the possible dual vulnerability of the child as both 'victim' and 'perpetrator' as well as considering the contextual nature of any perceived wrongdoing. Renewed emphasis on the experiential and intuitive dimensions of risk assessment would provide an important check on the technical and 'calculative' dimensions (Sparks, 2001: 169) which have thus far dominated the policy and organisational landscape.
In addition, a 'unifying' approach to risk (O'Malley, 2004: 326), and effecting a balance between 'victim' and 'offender' perspectives, also involves adopting a panoptic approach which encompasses the protection of victims as well as the management of offenders. Over the last two decades, both offender management (Kemshall and Maguire, 2001) and child protection discourses (Wigfall and Moss, 2001) have been reformulated on the basis of 'joined up' working via multi-agency approaches to management and protection. Throughout the UK, this is exemplified in MAPPA to risk assess and manage violent and sexual offenders. However, organisational discourse and practice, for the most part, still operate on the basis of siloed and compartmentalised thinking where victim-focussed and offender-focussed practices take place alongside each other rather than being fully assimilated. Shifting the parlance of risk to include the broader protection of children, beyond the rather narrow confines of individual offender management, also helps to reposition risk-based justice away from a reactive politics of risk towards what feminist scholars have termed a 'politics of safety' (Stanko, 1990).
In this sense, cultural and political discourses on risk would shift from a narrow focus on crime to a broader consideration of social harm including early intervention and prevention. Rather than expand the reach of formal risk-orientated processes to an everincreasing number of individuals, as Cohen (1985) and others have cautioned (see e.g. Garland, 2001), the focus is on incorporating more informal modes of social control as a means of offsetting the potentially criminogenic and self-fulfilling aspects of criminal justice risk management. 12 The structure agency nexus Effecting a delicate equilibrium between internal and external controls is fundamental to a humanistic paradigm of sex offender risk management. As desistance theorists recognise, charting the pathways from 'risk' to reintegration and desistance means balancing a focus on internal or attitudinal change with enhanced capacity for self-risk management and the provision of external pro-social opportunities to help realise this change (see e.g. Weaver, 2014). While most sex offenders will eventually mature or 'age out' of sexual offending (Harris, 2014;Lussier et al., 2010), there are perhaps three significant findings from the small but growing body of empirical research on sex offender desistance which would underpin a more humanistic approach to risk management.
First, is the complex interplay between the individual and structural correlates of desistance and the need to take account of the situational and relational contexts in which sexual offending typically occurs (Farmer et al., 2016;McAlinden, 2012). Socialisation influences, whether in the form of peer support within treatment programmes or relationships with significant others (intimate partners, families or religious communities) are known to be pivotal in shaping offending behaviour and influencing desistance (see Weaver, 2016). Second, is the combined effects of both formal controls (such as treatment; sex offender notification schemes) and informal social bonds (such as stable employment; positive social relationships) on rates of re-offending (see, e.g. Kewley et al., 2017;Kruttschnitt et al., 2000). Third, is the importance of optimism and hope for the future in underpinning and sustaining a personal desisting narrative or alternative non-offending identity (McAlinden et al., 2017). Within this broader context, pro-social concepts, such as work and relationships, while not necessarily operating as 'turning points' in desistance, in the way that they are known to do for non-sexual offenders (Sampson and Laub, 1993), are significant in giving individuals 'social capital' (Farrall, 2004) and ownership of their own rehabilitation.
At a conceptual level, the implications of research on desistance necessitate changing the strategic focus of criminal justice, from managerialism to transformation, by aspiring to effect genuine long-term change rather than short-term control. It also means effectively balancing internal and external loci of control (Cohen, 1985) on pro-offence behaviours and recognising the agentic capacity of sex offenders in working collaboratively 'with' them (McCulloch, 2005) rather than applying regulatory measures 'to' them. By way of example, 'strengths' and 'needs-based' principles (Burnett and Maruna, 2006) are premised on developing intrinsic motivations for change in order to not only reduce the risk for the community but increase the individual's chances of rehabilitation. These are encapsulated in the 'Good Lives Model' which recognises the need to safeguard the rights and well-being of putative sex offenders as well as those of victims (see e.g. Ward and Gannon, 2006). Moreover, empirical research with sex offenders also establishes that strengths-based approaches are more likely to promote desistance and internal cognitive and identity change (Hulley, 2017). Put simply, there is an a priori redistribution of the responsibility for risk management, where the emphasis shifts from elitist and external risk management strategies to individual self-regulation whereby sex offenders take responsibility for their own actions and are supported in doing so.
At the level of praxis, assimilating desistance research with public protection practices entails two core elements: First, is the need to foster the development of positive social contexts within which early trajectories of identity change can be nurtured and maintained (see McNeill and Weaver, 2010). This may involve, for example, the development of a greater range of work-based opportunities for those previously convicted of sexual offending as well as the expansion of programmes to engage offenders' families to support identity transition. Second, empirical learning about the situational and opportunistic nature of sexual offending, as well as the importance of optimism about the future, must be incorporated within sex offender treatment programmes. There has been a distinct move away from 'confessional' or backward-looking approaches towards more future-orientated 'biopsychosocial' approaches to sex offender treatment (Young et al., 2010). However, I would argue that a delicate equilibrium must be maintained between these perspectives, balancing the need to break with a history of sexual offending, with cognisance of the 'risk' factors and behaviours (such as 'grooming') that were an implicit or explicit pathway into individual offending. This approach would also offset the broader consequentialist tendencies of contemporary risk penality with nonconsequentialist thinking which focusses on the aetiology of offending and intrinsic pro-offence behaviours.

Participation
Participatory approaches to risk management may neutralise some of the principal deficits of the 'new penology' and 'algorithmic justice' described above. First, it is axiomatic that participatory decision-making which engages the local community may counterbalance the otherwise technocratic and elitist processes which characterise contemporary risk assessment and management. This also resonates with broader established trends in regulatory governance such as 'responsibilization' (Garland, 2001), 'partnership' approaches to crime control (Crawford, 1999) and 'decentred regulation' (Black, 2001). In this sense, shifting responsibility for sex offender risk management away from state agencies and towards ordinary citizens may operate as a form of 'pragmatic managerialism' (Sparks, 2001: 166). For instance, given the limitations of the law and legal frameworks in targeting pro-offence behaviours such as 'grooming' (McAlinden, 2012), the community are much better placed than state agencies to mobilise 'protective factors' (De Vries et al., 2015) and stop such behaviours.
Second, socio-political discourses on sexual offending also encompass significant moral and emotive dimensions that are not formally acknowledged within bureaucratic and technocratic processes. Undoubtedly, sexual offending, particularly that concerning child victims, is one of the most emotive issues of modern times, often inciting extreme public reactions (see e.g. Ashenden, 2002). Criminological literature also tends to articulate a linear and one-dimensional relationship between emotions, crime and 'justice' (Karstedt, 2002) which is usually centred on the dynamics of 'penal populism' (Johnstone, 2000). Public emotions concerning sexual offending, however, are considerably more complex than existing narratives might suggest. Similarly, victims of sexual violence are not a homogenous group and often seek a broader range of outcomes than that provided by conventional justice processes (see e.g. Herman, 2005;Naylor, 2010). In this sense, the voices of communities and victims can also be utilised diagnostically and, when channelled effectively, may provide a legitimate means of addressing the community reintegration of sex offenders. Affording the wider community a greater role in sex offender risk management and reintegration means formally recognising the affective as well as the effective dimensions of risk management and reintegration.
As Miller (2016) argues, while the centrality of criminal justice actors to the policy process has often been at the expense of other groups such as community organisations, citizens and victims; different democratic frameworks might facilitate broader forms of public participation, ultimately promoting less punitive crime policies. Meaningful citizen participation could take several forms ranging from information/consultation to fuller 'partnership' approaches. Moving beyond 'the risk society', however, means thinking more imaginatively than measures that only engage a small number of nonrepresentative community volunteers such as COSA and so should be multi-layered.
For instance, participatory approaches to risk would engage children and young people in their own protection (McAlinden, 2018) via mainstream education programmes in all schools. Such programmes should focus on a broader range of themes such as 'harm', 'respect', 'privacy' and 'consent' in on-line and off-line contexts rather than risk per se. More broadly, they would recognise the role of children in child sexual abuse not just as 'victims' but also as possible 'perpetrators' or 'bystanders'. In a similar vein, programmes would also engage parents/carers and wider society in fostering risk consciousness so that child protection and risk management are recognised as 'everyone's responsibility' as well as the media around responsible, non-sensationalist reporting of sexual offending. At the other end of the spectrum, town hall meetings could be held surrounding the release of high profile offenders to listen to community concerns and lay members could be involved more fully in MAPPA in decision-making rather than a pure oversight capacity.
In contrast to tertiary criminal justice approaches, this emphasis on early identification and prevention of sexual offending is well encapsulated in a 'public health approach' (Laws, 2000) which is characterised as proactive and preventative in nature. It is acknowledged, as O'Malley (2004: 335) contends, that 'local knowledge' may 'locate problems of risk and security (and their solutions) in different ways' to experts. However, new participatory forms of citizenship, or what O'Malley (2008: 460) terms the 'democratizing of risk', concerns a dialogue with the local community not only about problems or what will work but which can also accommodate challenges to riskbased expertise within localised settings. Such an approach, however, is not without its challenges and poses far-reaching implications for criminal justice as well as public discourses around sex offender risk management.

Conclusion
Utilising some of the core critiques suggested by risk scholars, this article has argued that the technocratic and algorithmic episteme of risk, which is beginning to pervade criminal justice policy on sexual offending within Western traditions, has been at the expense of a more heuristic and humanistic understanding of the aetiology of sex offending behaviour. In particular, a narrow or exclusive focus on an algorithmic episteme is dismissive of the nuances of sexual offending behaviour, the broader structural determinants of offending and reintegration as well as the possibility of individual offender reformation. As McNeill (2012) contends, a focus on the mechanics of risk assessment and management has led to a neglect of the dynamics of change that the rubric of risk aims to support.
As a counter to the dangers of algorithmic approaches which are predictive and non-evaluative, the article has outlined the parameters of a more contextual, humanistic lens on sex offender risk management wherein risk is understood as a broader social construct rather than 'a personality characteristic of individuals' (Maruna, 2020: 10). Under this philosophical re-positioning of 'risk', the core institutional and cultural imperative becomes one of understanding the complexity of sexual offending behaviour, recognising the humanity of 'offenders' and supporting the processes of change. In keeping with humanistic thinking, rather than focussing on dysfunction, the moral imperative of penal policy shifts from control to helping individuals fulfil their potential in a positive social context. Thus, the operative inquiry within criminal justice discourses shifts from why is it that sex offenders re-offend? to why is that they don't?
Advancing a new human paradigm of risk in relation to sexual offending poses a number of inherent challenges. At the policy level, this entails embracing the learning from research on sexual offending including the ameliorative possibilities of reintegration and desistance. As Hanson et al. (2018) contend, for example, penal policies designed to manage the risk of sexual re-offending need to have in-built review mechanisms to amend initial risk classifications or regulatory restrictions to allow individuals with a history of sexual offending to be released from the 'sex offender' label. This calls for policy retrenchment from some of the core features of the public protection model (such as public disclosure/notification) that involve the local community as mere passive consumers, rather than as potentially competent social actors, in risk management processes.
At the level of praxis, in moving away from a purely elitist approach to sex offender risk management, there is a need to effect the fine balance between 'populist' and 'participatory' approaches (see Johnstone, 2000). Although populist sentiments have clearly contributed to the enactment of punitive penal policies on sexual offending, participatory approaches could ultimately encourage a retrenchment from a politics of intolerance towards more rational, less technocratic forms of risk governance. Involving the local community and even victims more fully in the development of properly directed state-led risk management processes may foster both 'smarter' public attitudes to sexual offending and more sophisticated modes of risk governance. Here, in tandem with restorative thinking, the community, in particular, has a potential educative function in shifting wider public attitudes away from socially regressive versions of risk towards 'the common good' (Schweigert, 1999). As Kemshall and Wood (2007: 215) have argued, reconfiguring the risk of sexual offending requires 'a normalising discourse'. The first step, therefore, has to be the provision of accurate public information about the risks and realities of sexual offending behaviour via public awareness campaigns.
Within this context, there are broader societal imperatives underlying a 'civilizing' approach (Loader and Walker, 2007) to managing the risks of sexual offending. The first of these requires recognition of the inherent limitations of current technocratic and algorithmic approaches to risk governancethat no system can ever guarantee safety or eliminate risk entirely. There will always be false positive or false negative risks derived in human or machine bias (see e.g. Završnik, 2019). Therefore, the concentrated policy focus and cultural expectation become one of minimising risk and opportunities for abuse and maximising early detection. The second relates to societal acceptance of the intimate nature of riskthat 'sex offenders' come from families and communities and the potentially ubiquitous nature of sexual offending. More broadly, it also entails moving from an externalising blame culture towards acknowledging the integral possibilities of new modes of risk governance that support social reintegration and individual change.
of sex offender risk assessment, treatment and management, and with victims and perpetrators of sexual offending. 3. While some criminal justice policies are shared across all three jurisdictions within the UK (England and Wales, Scotland and Northern Ireland), there are also subtle policy variations with respect to, for example, risk classifications and the scope of multi-agency public protection arrangements (see Stafford et al., 2011). 4. Many other countries, including Canada, Australia, New Zealand and several Western European countries, also require those convicted of sexual offences to register with the authorities, although most do not make this information available to the public (Schulhofer, 2020). 5. Sex offender risk assessment is broadly comprised of actuarial approaches (e.g. Static-99; Sex Offender Risk Appraisal Guide (SORAG)) and structured professional judgement (e.g. Sexual Violence Risk-20; Risk for Sexual Violence Protocol (RSVP)), although many argue for a mixed approach (Singer et al., 2016). 6. See, e.g. the Harm Assessment Risk Tool (HART) used by Durham Constabulary (UK) following arrest to inform resource targeting for offenders based on predicted recidivism rates (Oswald et al., 2018). See also Correctional Offender Management Profiling for Alternative Sanctions (COMPAS), used within the American criminal justice system to assess offender risk and needs for diversion purposes (Brennan and Dieterich, 2018). 7. In the context of England and Wales, MAPPA is now governed by the Criminal Justice Act 2003, see especially ss 325-326. 8. Critics point to ethical considerations including bias, unfair outcomes and privacy (see e.g. Mittelstadt et al., 2016). See the decision of the English Court of Appeal in The Queen (on the application of Edward Bridges) v The Chief Constable of South Wales Police & others [2020] EWCA Civ 1058 where the use of automated facial recognition was declared unlawful on the grounds, inter alia, of non-compliance with the duties of public authorities to safeguard privacy rights under Article 8(2) ECHR. 9. Craissati et al. (2002), for example, reported that almost half of those convicted of sexual offending had been abused as children. Hackett et al.'s (2013) study of 700 young sexual abusers reports similar percentages. 10. See the Criminal Justice Act 2003, s 326. 11. Approximately 90% of sexual offending against children is committed by those known to them (Radford et al., 2011). 12. In this respect, the empirical evidence for concerns about net-widening are thought to be overstated: McMahon (1990).