British Terrorism Preemption: Subjectivity And Disjuncture In Channel “de-radicalization” Interventions

AbstractThis article examines Channel “de-radicalization” interventions, which take place on individuals suspected of having the potential to commit terrorist crimes. Situated within critical security studies, the article explores the British Prevent programme by utilizing primary interviews with hard-to-reach Channel mentors and senior Prevent officials. Following the work of anticipatory risk-governance scholarship, this research illuminates the three processes of risk-visibilization (how an individual becomes sufficiently “seen” as harbouring risk that they are offered Channel mentorship), risk-calculation (how practitioners negotiate supposed riskiness), and risk-knowing (how practitioners “know” risks they observe). It demonstrates how the practice of preemptive counter-terrorism is subsumed inherently by—even relies upon—subjectivity and human prejudice, and fundamental disagreements between practitioners. Through substantial empirical contribution on the phenomenon of Channel interventions, the discussion highlights ultimately that the algorithmic rationale of preemptive risk-spotting normalizes the suspicion of banal and everyday behaviors, precisely because such interventions are ultimately deployed through worst-case imaginations.

1 INTRODUCTIONThe British preemptive counter-terrorism program Prevent—and its intervention arm Channel—have been consistently lauded for producing “real results in helping divert people away from terrorism and violence” (Security Minister Ben Wallace, in HM Government, 2017). Looked to by other states as a paragon of preemptive risk management (Thornton Bouhana, 2017), it requires that teachers, nurses and social workers look for signs of radicalization in their clients, reporting them to Prevent so that interventions can occur to minimize the chance of any terrorism risk actually materializing. Yet it has been roundly criticized for operating as a national security apparatus on those who have not engaged in criminality. This article uses primary interviews with Prevent officials and hard-to-reach Channel mentors to trace the process of risk management of Channel cases, cases which are posited as the most risky “pre-crime” counter-terrorism referrals. In deploying a discourse analysis of 17 interviews with those overseeing the process (senior Prevent officials) and those managing “risk” (often Channel mentors), the article demonstrates the types of behaviors that are being associated with national security risk through terrorism preemption interventions. It seeks to connect for the first time empirical investigation of Channel interventions with the literature on anticipatory risk-governance. Prevent is of particular interest to those studying anticipatory risk-governance, as it has been critiqued for drawing together pre-crime and post-criminality (Ragazzi, 2016), and because its logic has been quickly and effectively exported abroad (Skleparis Knudsen, 2020). Ultimately, in alleviating the relative void of empirical data in this field, this paper documents the low threshold for terrorism-related interventions through the outworking of Channel, showing how an impersonal and subjective big-data rationality is operationalized, which brings an intractable security-consciousness into the everyday. This discussion matters because associating banal behaviors with “the terrorist Other” reduces the space within which political contestation can be outworked (Jackson, 2012).

The Channel processes are little researched, with the program closely guarding its practitioners and working practices. Officials often do not want to be identified because of the potential for backlash from within their communities, or for personal attacks on social media by particularly impassioned critics (Interviewee 6, Prevent Coordinator). This makes accessing interviewees extremely difficult, especially Channel mentors—who are often imams or who have some other responsibility over their communities and do not want to appear as government “stooges” (Interviewee 10, Channel mentor). This piece builds on the many critiques of the hegemonic “radicalization” discourse, which highlight a lack of scientific rigor underpinning key concepts, and the discursive construction of pre-crime risk (Ahmad Monaghan, 2019 Baker-Beall, Heath-Kelly, Jarvis, 2014 Githens-Mazer Lambert, 2010 Heath-Kelly, 2013 Kundnani, 2009 Martin, 2014, 2018 Silva, 2018). While the definitional ambiguity surrounding terrorism preemption has been widely documented (Elshimi, 2017 Elshimi, 2017 Lowe, 2017 Richards, 2011 Sedgwick, 2010), little empirical work has been undertaken to explore how contested terms like “radicalization,” “extremism,” and “de-radicalization” are navigated in practice through Prevent`s operation. Some research on the practitioning of the Channel program does exist, including that of Elshimi (2017), Thornton and Bouhana (2017), Spalek and Davies (2012), and Weeks (2017, 2018). This research mostly illuminates the mechanical process of Channel interventions, occasionally suggesting where it could be improved, and emphasizes the lack of clarity of definitions, aims, and outcomes central to preemptive risk management. However, most of this research does not seek to understand how supposed risk is being constructed, navigated, and managed, nor the knowledge the preemptive assemblage of Channel rests upon. Mythen and Walklate (2010, p. 58) call for a development of work around the impact of how “unknown” security risks are imagined and navigated: this investigation of the Channel “counter-radicalization” process illuminates, through significant empirical research, how practitioners negotiate this “unknown” and imaginary space, making it “known” and “producing security” by mobilizing pessimistic speculation in the absence of any tangible signs of danger. It shows how they subsequently engage in fundamental disputes about what constitutes riskiness and safety while managing “terrorism risk,” precisely because the operative logic is speculation. While the program is upheld for bolstering the security of society, this study shows how far prejudice, bias, and intuition—which produce variable results and which institutionalize stereotypes—are integral to the Channel mentoring scheme`s existence and functionality, bringing into question any claims of counter-radicalization success.

François Ewald remarked that the incidence of danger has not necessarily changed, but that society is now obsessed with understanding, quantifying, and attempting to manage risks (1991 also see Lianos Douglas, 2000). Whereas “prevention,” the historically dominant modus operandi, acts on historical data to interrupt a course of probabilistic and calculable future events, “pre-emption” is the act of attempting to stop a course of possibilistic future events, without any reliance on past data (Amoore, 2013). Following the work of critical security scholars such as Claudia Aradau and Rens van Munster, this research takes “risk as a dispositif for governing social problems” (2005, p. 91 emphasis added), where a dispositif refers to a structure of hegemonic knowledge informing the conduct and relations within society. For anticipatory risk-governance scholarship (Amoore, 2013 Amoore de Goede, 2008 Aradau van Munster, 2005 Ewald, 1991 Mythen Walklate, 2008, 2010 O’Malley, 1996, 2004), the modality of anticipating and preempting possible security threats presents an “open set of endless possibilities” (Anderson, 2010). The logic of governing the future through limitlessly imagining the worst possible catastrophes becomes central (Amoore, 2013). Emphasizing the governance of possibilistic risk imaginaries rather than probabilistic risk trajectories relies on “the inductive incorporation of suspicion, imagination, and pre-emption” (Amoore, 2013, p. 10) rather than attempting to comprehend physical threats.

The contemporary move towards the preemption of security risks is undertaken through big-data calculative logics, whereby masses of information are gathered and filtered in order for categorization (Amoore Raley, 2017 Heath-Kelly, 2017a Roberts Elbe, 2017 Wilcox, 2017). Anticipation is the new game in town, as securing the future becomes the focal point for ever-more social research and policy-making (Zedner, 2007). Preparedness is the necessary response to the anticipated catastrophe, where the worst possible event is expected to occur (Lakoff, 2007, p. 247), and so the state of preparation is a requirement of the risk dispositif, which sees everything through “an all-risks approach” (Omand, 2010, p. 11). This modality strives to assess every possible permutation, seeking to observe the whole of society en masse for the purpose of “profiling and classification [to] determine how individuals and groups are shaped and managed” (Mittelstadt, Allo, Taddeo, Wachter, Floridi, 2016, p. 1 also see Heath-Kelly, 2017a). The potential of radicalization is one to which every character trait is ultimately susceptible under this rationality, and so “big data logics of inductive profiling” and not probability judgments are the operative logic (Heath-Kelly, 2017b, p. 314). Prevent works through a big-data, algorithmic rationality to sort and categorize this imagined risk (Heath-Kelly, 2017a, 2017b). Every character trait or behavior is therefore sorted into more-risky and less-risky categories and acted upon accordingly: the subject of observation moves from the person to abstracted risk factors, “From a targeted [person-centric] practice to a whole-of-society [risk factor-centric] practice” (Heath-Kelly, 2017a, p. 34 also see Heath-Kelly, 2017b). Because traits and behaviors extricated from individuals are the fulcrum of politics under this modality, all sectors of society are legitimate targets of surveillance, because all could demonstrate these traits and behaviors, and so all pose some hypothetical risk (Ericson Haggerty, 1997). Big-data calculative logics attempt the “visualizing, calculating [and] knowing or perceiving future events such that they may be acted upon in the present” (Amoore Raley, 2017, p. 5). This article will therefore take these three functionalities—visualizing, calculating, and knowing—to show how big data logics are operative within Channel`s procedures, how the person becomes supplanted by considerations of abstract risk factors, and how petty sovereigns with Prevent`s operation have significant room to make decisions and interpret their briefs as they see fit (Butler, 2004 Heath-Kelly, 2012), sometimes appearing to have fundamentally different aims.

Between September 2017 and September 2018, I undertook 17 semi-structured interviews with Prevent-related actors, for roughly an hour each, on average, but ranging from 40 to 90 min. The interviews were comprised of six Channel mentors (one right-wing, one with a mental health specialism, and four Islamic experts or imams), three national Prevent leads, the architect of Prevent Professor Sir David Omand, two Prevent officers, a former Prevent police officer, two directors of organizations that received Prevent funding to train schools, one of the original Prevent Violent Extremism Pathfinder consultees (who had also received Prevent funding), and a local authority worker with responsibility for Prevent. All but one of the interviewees (and all Channel mentors) were male, and of note the Islamic mentors were males of color, and the right-wing and mental health specialist mentors were white males. I also attended official Workshop to Raise Awareness of Prevent (WRAP) training, run by one of the two Prevent officers I interviewed. I recruited these interviewees in various ways, including approaching them at public events on Prevent, sending them emails and handwritten letters, or using existing contacts to put me in touch with others (snowball sampling). The focus of interviews depended upon the role of the official, but they would generally revolve around understanding Prevent`s operation and remit. Questions that were central to most interviews included: What would a typical Channel intervention look like? What would someone have to say or do to get themselves brought to the attention of Channel (or a Channel panel)? How many people who receive Channel interventions would pose a risk either to themselves or other people? After broad questions like these were posed, interviews followed a conversational trajectory through related follow-up questions, and so the content of discussion depended on the answers given by interviewees.

The substantive discussion takes three parts. First, it explores how a person moves from being an ordinary citizen to being “made visible” as a potential terrorism risk by safeguarding practitioners, demonstrating the low threshold required to become a national security concern. It then shows how calculations about what constitutes risk is negotiated, indicating that risk management is infused with informality, and disjuncture between officials’ conceptions of risk. Finally, the paper establishes how officials “know” risk, showing the importance of intuition and national security narratives to justify intervening in the non-crime space. The article points towards banality in every stage of the process, from initial referral to Channel mentorship, allowing vast contradictions between policy objectives of preventing terrorism and on-the-ground “social work” interventions. It illuminates how worst-case “what if” logics are mobilized to justify preemptive security interventions, in the same moment that practitioners describe their work with no apparent regard to risk. The prioritization of imagination in risk management operates through personal subjectivity, which produces fundamental differences in how risk is understood.

2 MAKING “RISK” VISIBLECentral to the dispositif of risk is the sorting of “unknown, uncertain, and unknowable future into… ‘thinkable’ [or quantifiable] materialities” (Aradau van Munster, 2012, p. 102), and so the process of risk-visibilization in Prevent (or the quantification of possible risk sites) is central in order for corrective interventions to take place (Aradau van Munster, 2005 Martin, 2018). This section explores how visibilization of risk occurs, by setting out how referrals are brought to the attention of risk managers within Prevent`s structures, and exploring the types of behaviors that result in counter-radicalization interventions. It concludes that because the pre-crime space is devoid of physical threat, and because worst-case imaginations are prioritized, banal behaviors become associated with terrorism concerns.

In a Prevent context, visibilization of spaces where risk may emerge is operationalized through the designation of certain behaviors as “hazardous” through the 22 Extreme Risk Guidance (ERG22) risk factors. In order that its “experts” can filter and sort and categorize, referrals to Prevent are encouraged on an exceptionally broad basis, and so these 22 factors that are alleged to demonstrate possible risk include “excitement, comradeship or adventure,” “transitional periods,” and “them and us thinking” (reported in CAGE, 2016. Originally in NOMS, 2014, restricted). Despite the science behind these signs being contested by practitioners (Royal College of Psychiatrists, 2016) and by over 140 academics (Guardian 2016 also see Knudsen, 2018 Scarcella, Page, Furtado, 2016), these 22 factors are how referrals are accepted or rejected. Given that all incidences of “excitement” or “transitional periods” are not considered equally as possessing terrorism risk, what is acted upon by risk managers?

The procedure of highlighting where risk may exist within the Prevent apparatus involves initial referrals from public sector workers (from doctors and teachers to parents and social workers), up to the relevant safeguarding official within an institution (such as a hospital or school). These individuals are then visited by the local Prevent police team if the referral is deemed risky enough. If the police are concerned by their visit, the referral then makes its way up to a local “Channel panel” (a board of safeguarding leads), which examines the referrals, discussing the individual`s “risk factors” using data provided by the original referee and the police team. The most risky-sounding cases are ultimately offered one-to-one interventions with a Channel mentor. The supposed benchmark for intervention is when an individual demonstrates intent, capability, and engagement with an ideology (Interviewee 1, Channel mentor), correlating with the official three ERG themes (which are then broken down into the subthemes of “excitement” etc.). However, a large number of cases receiving mentorship were discussed explicitly as not demonstrating all three traits (discussed peripherally in Thornton Bouhana, 2017), as will be explored shortly. Where mentoring is approved by the Channel panel, diaries of the mentor and mentee are swapped, and the mentor then travels to meet the mentee at an agreed date for the first session—generally 2–3 months after the sanctioning of the sessions, during which a Prevent Police Officer may continue meeting the individual to monitor them (Interviewee 9, former Prevent Police). On average 6–8 sessions are sanctioned, which take place over the course of 6–12 months, though with significant variations (Interviewee 6, Prevent Coordinator).

The threshold was narrated consistently on a very low-intensity basis. Individuals might have uttered racist, juvenile words with a threatening tone (Interviewees 3 and 8: Channel mentors). One mentor (Interviewee 1) mentioned that “very offensive, and very dehumanizing” language would be enough to validate an intervention. Another mentor (Interviewee 8) said that his typical Channel cases demonstrate:Support for an extremist organization or extremist ideas. Those can include things like hatred for non-Muslims purely because they’re not Muslim. Or politicized rhetoric and one that verges on support for violence in the pursuit of political goals. And it can be you know making misogynistic remarks but with a religious justification for that. (Interviewee 8, Channel mentor)

This sentiment, on the issue of banality, was narrated in every interview with Channel mentors. When I asked one mentor who had mentored Channel cases for 5 years (Interviewee 3) whether anybody he had mentored was likely to pose a risk, he replied, “Erm… Possibly one… One person I thought ‘well yes he looks like he`s capable of doing it’ but then again you never know.” Another interviewee (Interviewee 5, Prevent Coordinator) remarked that whenever they deliver Prevent training they speak about “the badger cull that we had last year [and] HS2 [infrastructure development]… We would work with anybody with an extreme ideology.” So a consistently chronic lack of apparent risk, characterized by rhetoric verging on support of violence, religiously infused misogyny, and offensive language provides justification for Channel mentorship. Through Channel`s operational banality, potential terrorist intentionality is bestowed on those who would otherwise not have been considered security risks (Amoore Hall, 2009). As a program established through the counter-terrorism strategy CONTEST, and a program specifically aiming to “prevent people from being drawn into terrorism” (HM Government, 2015, p. 3), Channel performatively situates counter-terrorism as an everyday problem. Or another way, it situates the everyday as infused with terrorism risk. Under this rationality, “Anything can be a risk it all depends on how one analyses the danger” (Ewald, 1991, p. 199). Critical researchers illuminate that as the trend away from probability-based prevention and towards governing through possibility—or governing possibilities as likelihoods—takes place, security-consciousness becomes banalized and enmeshed within the everyday (Amoore, 2013 Amoore de Goede, 2008 Ewald, 1991 Heath-Kelly Strausz, 2018 Jackson, 2015 McCullough Pickering, 2009). So despite the history of badger cull- and HS2-related violence being markedly absent, these “ideologies” are deemed ripe for counter-terrorism intervention. Without substantial and rigorous checks, this logic of securing the banal becomes accepted and normalized in practice, which in turn produces anxiety about the risks posed by banality (Mythen Walklate, 2010 Zedner, 2008).

One of the recurring patterns from the interviews (especially with Channel mentors) is that individuals receiving mentorship have little in common with those who have engaged in actual violence. The sentiment narrated by one mentor (Interviewee 8), that “most Channel cases have been very low-level,” was consistent across interviews with mentors. This particular mentor, who also supervises Terrorism Act (TACT) offenders, mentioned that “usually the vast majority of Channel cases it`s a young person or a fairly young person who just needs a bit of mentoring and support rather than any very entrenched and extreme views.” Juxtaposed with the national security narrative around Channel, the lack of overt interest in violence that Channel cases appear to have, was startling. One mentor, talking about a young boy on Channel, commented that the childis not trying to be a terrorist … he doesn’t want to harm people, but he’s coming across material [like the conflict in Syria] which is attractive because [of] the way it’s presented or the issues they’re talking about. (Interviewee 10, Channel mentor)

Discussion with another mentor corroborated this: he remarked that he rarely discussed the legitimacy of violence during interventions: “You don`t want to exacerbate, introduce or create a problem which wasn`t there when you started” (Interviewee 8, emphasis added). Several officials noted that to help these individuals was better than not to intervene, because “if there`s [an] opportunity to help a young person, whether or not it has any violent extremist implications … racism in itself still deserves to be tackled” (Interviewee 8). While tackling racism is undoubtedly socially beneficial, bringing it into a space explicitly associated with “terrorism” appears problematic. This is an especially problematic trend, when we take into consideration that these individuals (who are, by the accounts of mentors, often just receiving anti-racism or critical-thinking mentoring with an inspirational role model) are receiving expedited social support like housing, mental health assistance, or aid into employment (Interviewee 6, Prevent Coordinator), at the expense of other safeguarding cases where serious harm has already been established.

Counter-terrorism interventions for racist children therefore make visible violent subjectivities of individuals who—by the accounts of mentors themselves—have never considered violence or who actively don`t want to harm others. But intervention occurs because practitioners imagine the individuals could at some possible future moment develop support for violence or engage in violence themselves. An explicitly risk-averse field, practitioners do not want to be held accountable if something goes wrong (Interviewee 6, Prevent Coordinator). Indeed, over- rather than under-reporting is encouraged: official Prevent training emphasizes “it might be nothing but it could be something” (“WRAP” training, provided by Interviewee 5). This ensures the amount of potential risks processed by risk managers is the highest it can possibly be, yet this methodology produces an embedded security-consciousness (Aradau van Munster, 2005). The imagination of every banal space as potentially associated with terrorism risk fulfils the circularity, and the management of such imagination to preempt risk from materializing “positively contributes to producing the conditions for its own exercise” (Massumi, 2010 also Lally, 2017). Imagining the actualization of “crime and criminal identities” (McCulloch Wilson, 2016, p. 24) and their logical conclusions (social harm) ensures the perpetuation of preemptive reasoning. Elshimi (2017, p. 169) writes that these rationalities of intervention on “vulnerable” people who demonstrate little apparent risk to anyone, suppresses the agency of individuals “to freely and rationally negotiate and contextualise their wider political, social and religious terrain.” Space for political engagement collapses as personal relations stop being personal relations and become risk-knowledge (Aradau van Munster, 2012), and everything signifies an indication of riskiness, in what Ericson and Haggerty (2006, p. 63) call a “disarticulation of subjectivity.”

Yet some social relations become imbricated by security-consciousness more than others: risk generally is positioned onto or “made visible” on subaltern bodies, on specific suspect communities (Hillyard, 1993 Pantazis Pemberton, 2009). What is seen, or what is sought out to be seen, as Butler writes (1993, p. 16), is “not neutral to the question of race it is itself a racial formation.” Despite an appearance of objectivity or distance from human subjectivity, calculative rationalities that rely on enormous data sets like that of Prevent merely outwork the racialized judgments of humans (Wilcox, 2017): indeed, two Channel mentors separately remarked that Prevent was immersed in hyper-visibilization of threat from Muslim bodies. Interviewee 8 complained, “But because [a boy] happened to be Muslim [the logic goes that] ‘we also need to look at this in some broader terrorism context’. That [mindset] for me has been very worrying.” Another mentor (Interviewee 13) lamented the disproportionate way in which Muslim terrorism offenders are treated, connecting it with the work of Prevent: “We are being disproportionate in how we`re treating Muslims. This again fuels the negative anti-Prevent that is a toxic agenda environment.” More personal and explicit micro-practices of the racism inherent within Prevent will be explored later, but for now it is incumbent to remember that while risks are everywhere under this preemptive modality, the inscri ption or visibilization of certain spaces as very risky is not a neutral but a highly political phenomenon (Aradau van Munster, 2008 Jackson, Jarvis, Gunning, Smyth, 2011 Martin, 2018). The association of certain behaviors, and the indirect association of people or groups demonstrating those behaviors, into more risky and less risky, is a process of visualization that “colonizes the future” by privileging certain outcomes as (less) desirable (Aradau van Munster, 2008).

As though risk was quantifiable and patently detectable, one Prevent Coordinator (Interviewee 5) said that the function of Channel panels was “to work out the level of risk” that individuals posed. The process of “determining” risk through Channel panel discussions was relayed to me as quite dull and bureaucratic by numerous interviewees. Interviewee 6 (Prevent Coordinator) made the point that individuals who are discussed at Channel panels are “getting first class support from a wide range of professionals who are all jointly invested and involved in the decision-making process.” These professional risk managers sat in the Channel panel make decisions that produce—in the words of Interviewee 8 (Channel mentor)—the “securitization” of children and young people, by being brought into the counter-terrorism space. The economy of (in)visibilization in the field of preemptive counter-terrorism functions not as an impersonal, random process, but through the actions of those able to make privileged claims about what constitutes risk, or what does ¬ constitute risk (Jackson, 2012).

3 CALCULATING (CHANGES IN) “RISK”So the sites of risk are made visible, made to appear, as a result of the bureaucratic negotiations of risk managers. But where “transitional periods” and “identity, meaning, and belonging” are such open risk signifiers for Prevent officials to negotiate, how are calculations performed—not just about risk itself but its management? What strategies are utilized to bring possibly dangerous divergents back into safety? Upon what basis are judgments about risk made so that possibly dangerous subjectivities become normalized? The following discussion will illuminate the informality of procedures and how that informality produces disjuncture throughout Channel`s operation.

The recruitment of Channel mentors is decidedly non-standardized. Particularly empathetic, respected, or established community leaders are approached to be recruited—sometimes many times until they accept the job—by those working within Prevent or by the security services (Interviewee 10, Channel mentor). The problematic nature of this informal approach to recruiting counter-terrorism practitioners was pointed out by one mentor (Interviewee 13):Have you ever seen a competency framework for a violence prevention practitioner? Have you seen it? [Me:] “I haven`t, no.” Exactly, there should be one! So therefore how do we know the people we’re employing are competent?

This practitioner had trained other Prevent practitioners from a position of being recruited without a job descri ption, and having received no comprehensive formal training. Indeed, just before this article was submitted, a fellow academic told me that during a research interview she had been conducting of a Channel mentor, he encouraged her to become a mentor herself. However, this is unsurprising given that the very foundations of Prevent were explicitly experimentative: David Omand (Interviewee 12), the architect of Prevent, remarked that the program was founded on the perceived need to do something, and that it might work, but if it didn`t work out then Prevent would be scrapped. This approach constructs the platform for disparate working practices, and demonstrates that risk calculations—being speculative imaginations about the future—are imbued with informality experimentation and ultimately guesswork. Without formalized procedures even in the recruitment of frontline terrorism preemption officials, it is inevitable not only that different interpretations of risk will result, but that the imagined risk will be managed significantly differently.

There are no set rules for how the intervention should function, and so the mentor decides what sessions will look like. Interventions take the form of very directed social work: Interviewee 6, a Prevent Coordinator, remarked that Channel, as “the jewel in the crown of Prevent, [is] silver service social work.” With sessions taking place once or twice per month (the most intensive lasting a whole afternoon once per fortnight), much time is spent just chatting with the individual, trying to understanding their thought patterns, so that mentors can gently probe to help the person become more reflective and critical. But Channel cases are also offered professional leadership courses and training with Premier League football teams (Interviewee 6, Prevent Coordinator), while others in need of employment are supported into a job (Interviewee 1, Channel mentor). Critical thinking and various forms of social, economic, and psychological support were framed as the “cure” to move Channel cases away from terrorism. One former Prevent Police officer (Interviewee 9) who had provided some Channel-style interventions mentioned that the “first thing I used to do was [figure out what] we have in common,” which was a typical response from actual mentors. The sessions would then progress from there, with mentors making decisions on how to construct the interventions, feeding back to the Channel panel to let them know how the case was progressing, and suggesting when a case should be concluded or whether they needed more sessions to be sanctioned (their input is valued above all when Channel panels make this decision). There is no protocol with how cases “should” be dealt with: after all, as was repeatedly emphasized in interviews, each case is unique.

This lack of formality underlying the operation of Channel was consistent throughout all the interviews. The non-standardization of interventions is explicitly acknowledged in official documentation: “At present OSCT-funded [Channel mentors] do not have a ‘standardized’ risk assessment tool each project has developed and deployed its own risk assessment” (HM Government, 2011, p. 61). Indeed, many officials pointed to the totally different assessments in each area, noting that what would be referred and accepted by the process in one area would not in another. So the process is designed to localize and silo the sorting of risk, which produces variable results from different practitioners. Interviewee 10, a Channel mentor, challenged another mentor at a joint meeting because they were suggesting a particular Channel case should be concluded. But the interviewee said:I assessed that [the other mentor] hasn’t provided sufficient tackling of [the person’s ideas]. So these are cosmetic changes. So in my estimation there’s a problem here, because I would not say “this person is safe”.

So while one mentor determines that an individual has “become safe,” another judges this not to be the case. This problem is intractable, because disparate views of risk (and its management) exist from practitioner to practitioner—differences were repeated across interviews (Pettinger, 2019). The illusion of the existence of non-situated knowledge (Haraway, 1988, p. 581) is central to justifying interventions: it was claimed many times that the solution to differences between practitioners’ views is just more training, and working together with a more joined-up approach. Yet Prevent is an apparatus inherently infused with subjectivity, being “as good as the people inputting into it” (Interviewee 1: Channel mentor). No two practitioners will ever have the same views on what constitutes risk. Given their own life experience, how they`ve been trained (or not), and their reading of body language (see below), risk will continue to be defined differently by each practitioner. Wilcox (2017) points out that despite the appearance of impartiality, the inputs of any algorithmic sorting rationality are imbued with bias and human partiality, which ultimately produces not only a racialized methodology but one which functions through disagreements and disjuncture, hence the fundamental disagreement described above.

Occasionally, where interventions are sanctioned, mentors are told by the Channel panel that “we don`t think he`s got any capability at all, however they say ‘we`d still like you to meet them because you know to put them on the right path’” (Interviewee 3). In one case, a Channel panel asked a mentor to provide an intervention, but after spending 5 min with the individual, the mentor (Interviewee 8) stopped the session:This is a year 7 kid … and he`s also someone with special needs, learning difficulties. [I asked the Channel panel] “So what are the ideological concerns around this person, what signs of extremism have they exhibited?” [They replied,] “Oh he said something about wanting to go to Kurdistan where there are soldiers and guns.” But he’s half Kurdish, and the family went to attend a family holiday there in the summer holidays, where people celebrated by firing guns in the air, and he was talking about it… This is the sort of thing that can give Channel and Prevent a terrible bad press, the securitization of a kid purely because they`re a Muslim. And had he not been a Muslim, that would [have] been quite quickly realized [it was] a kid talking about his summer holidays.

This example symbolizes the problematic nature of preemptive interventions. To function without discrimination, it would require a universal understanding of risk indicators throughout the process, and across society. British newspaper The Guardian carried statistics (Versi, 2017) following the publication of Channel data in 2017, showing that while arrests due to so-called “Islamist terrorism” were 5 times higher than “extreme right-wing terrorism,” Muslims are 40 times more likely to be referred to Prevent and their referral be accepted. This represents an 800% differential between crimes and perceived risk. Yet where cultural, familial, and even personal customs are so varied, how could it be expected that “risk” be understood uniformly—especially where it has not begun to materialize into physical danger, and where the supposed signifiers of risk are so banal? Despite countless claims that neutrality and multiple safeguards exist in the process, race is central to all predictive sorting programs (Wilcox, 2017). Moreover, this racially and gendered bias is even less surprising given the explicit context of the initial direction of Prevent, which constructed Islam as problematic by providing “counter-extremism” or “risk-management” funding to areas “based on the proportion of the Muslim population in that region” (DCLG, 2008, Section 5 also read Spalek, 2012). Because practitioners are dealing with unknown and unidentifiable possibilities, with arbitrary and imagined distinctions like “extremism” and “radical’, and with behaviors that could be interpreted various ways, practitioners operate with different objectives and metrics of success.

4 KNOWLEDGE OF THE SUBJECT“Knowing” risk is central to the performance of risk management (Amoore de Goede, 2008). As one mentor (Interviewee 13) said, “I’m in the business of assessing dangerousness.” But how is risk known to the extent that it becomes actionable knowledge (Amoore Raley, 2017)? How are future events perceived so clearly that they merit action, when no apparent intent has developed (according to practitioners themselves)? How do practitioners understand the possible futures of individuals without demonstrable signs of preparation? This final section explores how practitioners rely upon both intuition and worst-case logics to justify acting on these individuals.

Intuition is one of the most functional components that practitioners deploy to calculate risk (and its reduction or increase), precisely because their work operates through imagining or speculating about possible futures which have not yet materialized. The role of “gut feeling” was emphasized as highly valuable in interviews to figure out what risk was posed. Interviewee 9, a former Prevent Police officer, said, “It sounds naff but [determining risk is] a gut feeling.” While intuition and gut feeling was framed as central to police work and therefore helpful to calculating risks under Prevent (Interviewee 6, Prevent Coordinator), Prevent is not police work: it is social work-style engagement with non-criminal subjects (though some do make outrageous or viciously racist remarks), and who are often children. Narrating intuition as applicable in a pre-crime context because of its post-crime utility to spot “sites of risk” situates the line between pre- and post-crime as fluid and open, and ultimately amenable to intervention: because there intuitively is something wrong, it must be acted upon. Rather than subduing stereotypes and personal prejudice, the encouragement to deploy intuition runs through the whole of the practice of Prevent, from initial referral to the conclusion of a Channel case, as we will now see.

The probability of success (represented broadly as engagement with the process) or failure (the individual leaves as they came in) of a Channel case can be expected early on through how “introspective” the subject is (Interviewee 1, Channel mentor), as well as analyzing characteristics like their body language, tone, and facial expressions. Alongside unidentifiable gut feelings being narrated as useful, one mentor spoke about intuitively interpreting body language to comprehend how much risk individuals posed, in order to deduce whether the counter-radicalization sessions were ready to be concluded:You can tell from the way they ask questions and you can tell from how they’re listening to you, are they attentive, are they looking at their watch, playing with their phone, so it’s all manner of things. Body language gives it away, it’s dead easy, body language and I’ve studied body language as well. Not professionally but I’ve looked into it you know. You can tell and I’m a people person myself.

Security-consciousness in this way draws together what would otherwise not be considered risks, and looks for anything that could constitute divergence—such as playing with a phone in the company of another—in order to “put that person back on the right path” (Interviewee 5, Prevent Coordinator). Ever-new forms of inductive reasoning are deployed to draw together otherwise disparate risk factors, in order that embodied risky spaces can be identified, or “visibilized” (Amoore, 2009). The encouragement of over-reporting through intuition produces a risk-averse consciousness, and so subjective perspectives remain unchecked and are even valued, because this prejudice is how the vast scale of referrals—a key priority of Prevent (Heath-Kelly, 2017a)—is achieved. Reliance of the sorting process upon the unquantifiable and unexplainable intuition of police officers and safeguarding officials is itself characteristic of the calculative rationality. Burrell (2016, p. 1) writes that when big-data inductive logics are deployed, “Rarely does one have any concrete sense of how or why a particular classification has been arrived at.” Intuition and gut feeling are by definition unexplainable, yet these characteristics were simultaneously lauded as useful tools to “spot risk” separate from any scientific methodology.

Finally, the centrality of the potential catastrophe underpinning the preemptive rationale was consistently posited as justifying interventions, despite the overwhelming banality of the cases described. While risk was narrated consistently as being largely absent, strategic leads often returned to national security to validate Prevent. While mentors—in describing the intricacies of their work—generally focused on the role of belief systems and divisive language (and an absence of violent desires of the individual), more strategic senior officials spoke of the role of Channel as intimately related to violence. Interviewee 6, a Prevent coordinator, declaring that the apparatus didn`t just operate to encourage people to change their beliefs, stated:We’re not the thought police … I would draw [the line at] the distinction of violence or the threat of violence, or the promotion of violence. Everybody can aspire to the political ideology they wish, in my view, but if they seek to achieve it through promoting violence, then that’s where we need to think about getting involved. (Emphasis added)

Similarly, a national Prevent lead (Interviewee 15) mentioned that Channel mentoring would be offered “if there is an individual who needs to be disengaged from terrorism or de-radicalized so to speak.” So despite the consistent testimony that not even “very entrenched and extreme views” are being tackled in most interventions, the concept of disengagement from “violent extremism” was in many moments also mobilized. This discursive positioning of Prevent around physical threat occurs consistently at the more strategic and policy levels. Yet these utterances sit awkwardly with the overall sentiment given by Channel mentors, who function as intermediaries between those allegedly posing risk, and strategic risk managers. The testimony of those “at the coalface” instead suggests that monitoring and amending ideas is central to their work, rather than tackling thoughts of violence which are often entirely absent. This disconnect between narratives of significant threat and cases themselves not being considered risky was an under-running theme. Particularly when remembering that mentors are most intimately exposed to the individuals once intervention begins, this disjuncture discredits the relative uniformity of public narrative on Channel as a “preventative” program related to diverting individuals from “terrorism.”

While the disconnection described was most visible when considering seniority within the Prevent hierarchy, disjunctures sometimes still occurred within the same interview. Mentors, for instance, were not allowed to tell me how many cases they had mentored on the grounds of national security. Moreover, while holding the views of a lack of threat or violence noted above, some also spoke of the threat posed by those on the program: when asked if Prevent should have an expiry date, one mentor became agitated, raising his voice and declaring “No!... Safeguarding against violent extremism in various different forms is safeguarding against extremely violent behaviour, having a project that safeguards against people in that area is essential.” Another (Interviewee 12) mentioned his job was a “violence prevention practitioner.” The observing gaze is fixated onto abstracted risk factors (the making of disturbingly racist remarks), rather than onto the person (often a child, who, making the same remarks 15 years ago would not have been considered a potential terrorism risk), and so this disconnection is able to take place. Those at strategic levels are most exposed to this abstracted gaze—after all, they are not spending several hours with the individual concerned, but oversee a number of cases—yet for it to be apparent in mentor interviews too is indicative that, “In the technologized age of bio-political war, bodies are absent” (Parpart Zalewski, 2008, p. 103). Instead of bodies being present, all cells within an organism pose hypothetical risk, all characteristics within a body could contribute to the person`s radicalization, bringing us back to the banality of interventions. The person is just a vessel possessing potential risk factors. If people (often children) rather than the mere abstracted risk factors were observed, most of these cases would not ultimately receive a counter-terrorism intervention through Channel, because it would be understood or “seen” that these individuals pose little-to-no threat—as per mentors’ own testimony. Yet because risk factors and not the person themselves are visibilized, interventions occur on people, or “bodies,” who are understood or “seen” ¬ to pose risk. But because social relations exist in a paradigm where potential risk is considered ubiquitous (Lianos Douglas, 2000), it follows that Channel as a counter-terrorism assemblage would intervene upon children who are understood not to harbor violent intentions. This form of knowledge—suspecting everything and everybody—privileges people’s possibly malicious future intentions over other knowledges like their presumed peaceability, and is at its root reliant upon suspicion and worst-case logics (Pettinger, 2019 Valverde Mopas, 2004 also see Amoore de Geode, 2008 Ericson, 2008). However, by definition, the spotting of and intervention upon risk is “not personal”—after all, the embodied person has all but disappeared.

5 CONCLUSIONSo Channel mentoring intervenes upon banality, functioning through subjective, intuition-infused, and worst-case logics, allowing children who “[don`t] want to harm people” to be offered counter-terrorism interventions. This seems absurd when the program is simultaneously described in interviews (and public discourse) in terms of diverting people from terrorism. Channel subjects are released from the program once they think more critically—hardly an established litmus test for the potentiality of engaging in violence. The level of distance from taking into account the whole person, and the fixation instead on abstracted risk factors, allows narratives of national security to be intertwined with interventions that largely do not appear to be violence-related. As a result, everyday behaviors become national security concerns—although in specific, racialized bodies. So while big-data calculative logics rely upon the veneer of objectivity and a distanced observer able to make neutral judgments (Wilcox, 2017), according to the testimony of Prevent`s own practitioners such logics mobilize prejudice, imposing risk-centric subjectivities onto non-violent citizens. Yet the management of risk never proves itself fulfilled, and risk managers are judged not just by their knowledge or lack of knowledge, but by what they should have suspected when a breach occurs (Aradau van Munster, 2005 Ewald, 2002). If risk cannot be seen, those tasked with our security (which is, more and more, citizens themselves) haven’t looked hard enough (Aradau van Munster, 2008). As much information as can be gathered on everybody and everything is therefore an essential component of governmentality in the dispositif of risk (Heath-Kelly, 2017a Miller Sabir, 2012). Thus, because risk must be out there somewhere, banality becomes entrenched within counter-terrorism, and counter-terrorism responsibilities become embedded in everyday life. As national security-consciousness becomes normalized, the desire to look for signs of catastrophic risk in every locality, and to separate “people, places, and products” on their risk capacity only becomes stronger (Mythen Walklate, 2010, p. 58). This is a dangerous turn, because, as Ericson (2008, p. 76) writes, “Insecurity proves itself.”

Elshimi (2017, pp. 156–185) writes that a complete lack of available data, measurement, and external oversight characterizes preemptive counter-terrorism in Britain. My research demonstrates that these three problems are as acute in the physical “counter-radicalization” intervention space as in the conceptual space—suggesting these problems exist precisely because practitioners work within a risk-averse environment that encourages the hyper-consciousness to risk, an obsession with worst-case logics in the face of banality. The weight of suspicion in the dispositif of risk generally overrides probability judgments, or the criminal justice methodology (de Goede de Graaf, 2013 McCullough Pickering, 2009 Pettinger, 2019), allowing the discursive positioning of banal behaviors as terrorism-related. Without significant challenge, the prevailing modality of governing through impersonal judgments will continue to become embedded (Tutt, 2016) regardless of success. As security moves from prevention to preemption, citizens’ identities are able to be reconstituted on a whim so that risks can be negotiated before they materialize (Walker, 2008). Citizens move from being seen as persons with political capacities to a collective and potential future risk to be managed (Amoore, 2013). An individual consequently moves from “ordinary citizen” to “potential national security risk” because of discussion of a panel of safeguarding leads which has never met them. This model is one of “universal suspicion [where] everyone is suspect” (Ericson, 2008). Once it has taken hold, the logic becomes immersed into normality with little space for resistance: after all, who opposes security? As one Channel mentor (Interviewee 13) remarked, “We know Prevent`s shit, we know it`s highly problematic, but it`s not gonna go away. And even if you got rid of it, it would still come back in another form.”