In-secure identities: On the securitization of abnormality

This paper offers a local analysis of one commonly-used airport security technology: full-body scanners. To follow on Salter’s (2008b: xi) claim that ‘airports have long been laboratories for new strategies of both technological and social control’, the scanner provides a small theoretical experiment in thinking about state security – or perhaps a quite concrete case study thereof. In Foucauldian terminology, the scanner serves here as an anecdote ¹ demonstrating a certain change, a tendency, in the configuration of sorting bodies and the violence this configuration entails.

The change we want to identify does not pertain to any and all forms of state security. The scanner is an example of scenarios wherein security forces are configured to deliberately relinquish discriminatory presuppositions and establish objective measurements for identifying threats. We ask: what happens when security schemes are configured to operate objectively? Who emerges as the potential enemy, as the carrier of threat, when security is configured to treat all equally? Based on reports by government agencies, accounts by civil rights organizations, media reporting and interviews, as well as a critical examination of the technology itself, our analysis aims to substantiate the following observations: (i) in their attempt to rely on objective measures, security protocols have integrated the notion of a statistically calculated normality into the realm of security sorting. (ii) While ‘normal’ in this context supposedly represents the mere prevalence of a given phenomenon, these systems ultimately reproduce categories which are very much aligned with social norms. (iii) As a result, people whose bodies and behaviours deviate from measured standards of normality and who fall into categories of social abnormality (such as the mentally and physically disabled or gender non-conforming individuals), re-emerge in such settings as suspected terrorists.

Prior research has identified such tendencies and has provided compelling analyses of their effects on those caught up in new security technologies (for such analyses, see: Currah and Mulqueen, 2011; Samuels, 2016; Shepherd and Sjoberg, 2012). Our research diverges from this line of enquiry by shifting its departure point from identity categories to the technological apparatus itself. It provides an examination of the logic subtending these security systems, reviews its development within arrays of conflicting motivations and constraints, and examines its conditions of possibility. In so doing, we aim to explain, first, why social categories of abnormality have become prone to being reconfigured as potential risks and second, how this configuration differs from a long history of identifying the abnormal with the suspicious.

To give a sense of what is at stake, let us begin by alluding to the broader circumstances through a deliberately eclectic set of examples. In Canada, the Identity Screening Regulations Act states that ‘An air carrier shall not transport a passenger if (c) the passenger does not appear to be of the gender indicated on the identification he or she presents’ (Justice Laws Website, 2011). In an attempt to issue clear guidelines, this regulation does not recognize that for transgender individuals, such discrepancies result from the incompatibility of bureaucratic systems to people’s lived realities (Spade, 2011), but sees them as attesting to efforts to forge documents, to conceal identities and to lie (Beauchamp, 2009). Consequently, gender non-conformity, a category that was traditionally pathologized and appeared as an abnormality to be corrected within a binary system of sex/gender, re-emerges as an indicator of a potential security hazard. Or, to take an altogether different geo-political context and another type of ‘abnormality’: searching for suspicious behaviours, visual surveillance systems in train stations are not preprogrammed to look for particular triggers. Instead, they track types of movements of passengers on platforms, to establish what ‘normal’ patterns of behaviour consist of (e.g. time spent on platform, speed and direction of movement). Having established the normal baseline, these systems then trigger alerts when identifying people who deviate from it. The rationale is that people who do not behave ‘normally’ on the platform may not be regular passengers and may pose a security threat. According to security experts who have developed several such systems (AG, 2013), these algorithms frequently identify people who are considering committing suicide on the rail, homeless people and beggars as potential threats. Hence, a security system which works to minimize security risks by relying on non-discriminatory algorithms identifies as a threat persons who do not comply with assumptions concerning normative (that is, productive) ways of life. Or yet another context and another form of deviating from presubscribed norms: in the Palestinian West Bank, mentally disabled Palestinians are repeatedly shot by Israeli soldiers, despite posing no threat. ² To be clear, the violence which the Israeli security forces exercise in the occupied Palestinian territory is anything but ‘non-discriminatory’ as it rests on a clear differentiation between Palestinians (‘legitimate’ targets) and Jews or other non-Arabs (who are the people this violence seeks to protect). Nevertheless, non-discrimination transpires here since the Palestinian population is generally and uniformly treated as potential terrorists (Ghanim, 2008). In this context, mentally disabled Palestinians are disproportionately shot because, when encountering Israeli soldiers, they may shout unexpectedly, run away when ordered to stop, not obey soldiers’ commands and move erratically when expected to stand still. They are, therefore wrongly identified as posing a risk to the soldiers and are treated accordingly. That is, the mentally disabled in the West Bank are shot because of their mental disability, because they cannot abide by expected norms of behaviour.

Despite the radical differences between them, taken together, these cases demonstrate our claim: In civic spaces in which security apparatuses are densely deployed, certain categories that have traditionally been addressed through the disciplinary discourse of abnormality are found to reappear within a matrix of risk and threats. This should not simply be understood as anecdotal or incidental. It demarcates, we claim, a re-shaping of the manifestations and rationalizations of particular forms of state violence. This process goes beyond a change in the level or type of violence to which people associated with such categories are subjected (that is: we do not merely argue that people who are in various ways marked as abnormal are subjected to more violence). Rather, the entire social attitude towards abnormality changes here: no longer primarily the target of examination and assessment, correction and improvement, abnormality becomes the object of isolation, delay and in some cases also elimination. We label this process the securitization of abnormalities. ³

Our analysis will develop this argument in more detail through looking at the case of the full-body scanner. Yet, this particular test-case leads us to another argument, a more specific addition to the literature concerning new security technologies. The accelerated pace of introducing technological innovations into the field of security leads research to focus, many times, on how technology itself reshapes modes of governmentality, the configuring of subjects, the flow and distribution of bodies, and the execution of violence (Amoore, 2009; Aradau et al., 2008; Leese, 2014). For us, however, the technology we examine for the most part serves as a test case, representing a broader phenomenon. Placing our argument within a larger framework, which includes policing, military operations, and border control regulations, we aim to demonstrate that the securitization of abnormality must be considered beyond technological implementation. At stake, therefore, is a wider claim, relating to a rationale of state violence that rests on objective non-discriminatory measures to justify itself.

We begin with clarifications relating to the scope of our argument and its position in relation to existing literature in the field of critical security studies. We continue, in the following section, with a brief review of the historical circumstances in which the tension between prejudice and objectivity has shaped security sorting and, specifically, the evolving introduction of the full-body scanner into airport security. In the third section, entitled ‘Securitizing Abnormality', we focus on the classification, by the full body scanner, of gender non-conforming bodies as a threat. We argue that in this case, this securitization of abnormalities is a paradoxical and unexpected, yet indicative and perhaps also logical by-product of the attempt to base security schemes on objective and regularized procedures that sift out prejudices and biases from the practice of such systems. The section ‘Demarcating the Terms II: Abnormality', further explores the notion of securitized abnormality by situating it in relation to other historical contexts in which abnormality appears as a threat or risk. The final section of this paper provides a more systematic conceptual and theoretical analysis of the notion of normality via an engagement with the work of Foucault.

Demarcating the Terms I: Security

Examining the screening methods used by security agencies, security professionals make a distinction between risk-based security screening and screening which assumes uniform risk (Elias, 2014). The first, risk-based security, is centred on the assumption that risk is not evenly distributed throughout society and that some categories of persons carry more risks than others. ⁴ Risk-based security thus seeks to economize efforts of security providers by identifying indicators that isolate these categories in a process which is often called ‘profiling’. In its cruder forms, profiling is organized along social classifications and categorizations, such as class, race, or citizenship status (also known as ‘racial profiling’); other, more intricate, forms of profiling rely on identifying patterns of behaviour through surveillance, data-mining or on-site observation (DHS, 2014). Both types of profiling are notoriously controversial: While profiling along social classifications is highly prone to discriminatory practices (if not discriminatory as such), the reliance of behavioural profiling on personal data is primarily dependent on intrusions into people’s privacy. The profound social implications of these various profiling practices have led most critical analysis of security – in both public and scholarly arenas – to centre on the implementations of risk-based types of securitized sorting methods.

Drawing much less attention is the second type, the uniform-risk security, which we encounter most often in highly restricted spaces. This type of security rationale assumes that everyone is potentially dangerous until proven otherwise (this is why everyone has to go through a metal detector before entering a government building or boarding a flight) (Elias, 2014). Importantly for us here, the fact that less attention is given by scholarly critiques to this type of security suggests these critiques latently accept the claim that this mode of security is less discriminatory (Aradau and Van Munster, 2007; Bigo, 2006, 2007; Wilcox, 2015). The full-body scanner at the focus of this article is predominantly a uniform-risk-based technology, and our argument should be read as pertaining solely to this particular field of security.

This brings us to a second point. Critical examinations that do work on uniform-risk technologies do not see them as operating at the level of subject-formation, identity categories, or normalization processes. To put this into context: When examining risk-based security, particularly policing, we see a long history that ties together the targeting, quarantining, and restriction of the poor, migrants and others who have been marked as morally and/or socially inappropriate, with contemporary security technologies that seek to identify, isolate, or eliminate those who are deemed a threat (Walters, 2010). As others have shown, this history corresponds to an intrinsic rationale of security that is always already entangled in social distinctions and classifications (Bigo, 2002; Lyon, 2003; Neocleous, 2008). Accordingly, critical analyses of risk-based security often understand them as a form of governmentality (drawing on a Foucauldian framework), and claim that they entail at their core strong normalizing tendencies (cf. Adey, 2009; Amoore, 2009; Salter, 2007). However, uniform-risk security is often analyzed outside of this history and the sorting practices uniform-risk security employs are perceived, by contrast, as disassociated from the management and regulation of collective categorizations.

In what follows, we seek to question this assumption, claiming that even these uniform, nondiscriminatory procedures which fragment the body and seemingly take it out of its social context are reliant on normalization processes. These procedures therefore also recreate processes of identity-formation which are always already entangled in normalization. The following section looks at the operation of the full-body scanner in order to better demonstrate these claims.

Full-body scanners: A brief history

Flying through any of the major international airports in the United States or Europe, one has grown accustomed to seeing full-body scanners in operation and has probably had to go through them repeatedly herself. With the increasingly invasive airport security rituals travellers have to endure, the controversy surrounding the introduction of these scanners was soon forgotten, but their integration into the airport security theatre was anything but trouble-free. This controversy did not only delay the integration of the scanners into airports, but also reshaped the technology itself.

Searching for non-metallic objects hidden under people’s clothes, full-body scanners create a negative nude image of the body (Figure 1). Intensifying what Hall (2007) called ‘the esthetics of transparency’ – the equating of visibility with security and of concealment with threat – these scanners are designed to expose what we are socialized to keep private: the body under our clothes. Their introduction into airports, therefore, stirred fierce public objection, including claims that this scanning is ‘equivalent to a “virtual strip search”’, and that the machines show ‘extraordinary disregard for the privacy rights of air travelers’ (Electronic Privacy Information Center, 2013). This public outcry managed to push-back security considerations at the time (a rare example of such success indeed), and the plan to put these scanners into use was suspended (Hunter, 2009). It was not until Umar Farouk Abdulmutallab tried to blow up Northwest Airlines Flight 253 on Christmas Day of 2009, using plastic explosives he had hidden in his underwear, that the American Transportation Security Administration (TSA) implemented this plan. Other agencies internationally soon followed suit (Magnet and Rodgers, 2012). (This was despite security agencies’ officials admitting that operational scanners would not have aided in detecting Abdulmutallab’s hidden explosives, McCullagh, 2010.) OPEN IN VIEWER

Figure 1.

Image from an active millimeter wave body scanner.

To ease public objections to the use of these scanners, various measures aimed at increasing privacy were introduced. The first generation of these scanners would display the nude image to the security officer operating the machine. Thus, the officer who was also handling the passenger could literally see through the clothes of the person standing in front of her. As this procedure was quickly found to be too invasive, the next generation of machines had the image projected onto a screen in a remote booth, where it was examined by a different operator. This operator would then only inform the officers whether the image had raised any suspicions (Elias, 2012). While this solution was perceived as an improvement, it still raised considerable concerns and objections, as people were not only troubled by the fact that a nude image of their body was produced and seen by someone else, but that these images could be leaked and circulated (as they indeed have been in some cases, see Bosker, 2010). Finally, in 2011, new scanners were introduced which seemed to solve these problems, as they included an image-analyzing software. Called Automated Target Recognition (ATR), this software performs the analysis of the image automatically, and the nude image is neither seen by anyone, nor is it permanently stored anywhere (DHS, 2011; Elias, 2012). The scanners’ only output is an abstracted human figure, on which the location of suspected regions on the passenger’s body are marked (Figure 2). Passengers who are flagged by the scanner then go through a pat-down and additional questioning by security personnel. OPEN IN VIEWER

Figure 2.

The output generated by L3 ProVision® which includes an ATR system.

This narration of the evolution of the full-body scanner shows how the public’s concerns regarding privacy have shaped technology. From this perspective, the scanners are seen as a success story in which state-of-the-art technology offers solutions to human concerns and limitations (cf. Sutton, 2014). However, there was an additional constraint which was central to the shaping of this technology. Added at this junction, at which security and privacy meet, was the requirement to eliminate discriminatory practices.

To understand our claims, a brief history of the evolution of airport security as it has unfolded since the early 2000s would be useful. At its inception in December of 2001, the TSA searched for tried and tested verious aviation security solutions. As part of this process, it turned to Israeli security specialists. Israeli aviation security, which was perceived as an unprecedented success in battling terror threats, exposed that its mode of operation was very much reliant on racial profiling (targeting all Muslim and Arab passengers) (Hasisi et al., 2012; Pfeffer, 2016). While the TSA was happy to adopt this approach (Baker, 2002), it soon learned that liberal democracies posed particular legal challenges to overtly basing an officially inscribed policies on racial profiling (Gross and Livingston, 2002). After repeatedly being challenged on this issue in court and public domains, the TSA had to publically denounce its use of such methods. Pistole (2013), the head TSA Administrator at the time, renounced racial profiling in front of the House Committee on Homeland Security in 2013, called it ‘unlawful’, and insisted that such practices were ‘not tolerated by TSA’.

While not completely abandoning profiling methods, the public pressure has pushed the TSA to increasingly rely on uniform-risk security practices. Within this framework, uniform-risk security procedures were argued to be, beyond merely ‘nondiscriminatory’, a solution for discrimination. Abiding by what Daston and Galison (1992) called ‘mechanical objectivity’, reverting to full body scanners thus also rests on the presumption that their operation is necessarily impartial. Indeed, Michael Chertoff, who was the US Secretary for Homeland Security between 2005 and 2008, declared that the implementation of technologies such as full-body scanners ‘allows us to move beyond crude profiling based on prejudice’ (quoted in Amoore and Hall, 2009: 453). In line with such claims, the scanners are advertised by their manufacturers as eliminating discrimination, since they do not register pigment density or racial characteristics and are thus ‘color blind’ (Saletan, 2007).

But once the ATR algorithms were integrated into the full-body scanners, it very quickly became clear that particular groups of people were being singled out. Gender non-conforming individuals such as butch women and trans persons, black women with particular hair styles (such as ‘sisterlocks’), people wearing certain religiuos garb (such as dastaar (Sikh turbans)) or passengers carrying medical appliances on their bodies (such as people with a stoma), found that they regularly became subjected to invasive pat down and questioning, and exposed to extended security probing. ⁵ Paradoxically, the attempts to ensure objectivity are precisely what have produced such modes of discrimination; not because the claims of objectivity are merely a lie (that is, yet another way to bring in racial profiling and other forms of discrimination through the back door), but because discrimination is engrained into the way in which objectivity is produced. The case of gender non-conforming persons is indicative here, and will thus stand at the focus of the next section.

Securitizing abnormality

To understand why and how transgender persons find themselves targeted by security procedures such as the full-body scanner, let us look at its operation. Like all scanners of its kind, the full-body scanner is designed to identify concealed weaponry and contraband objects under people’s clothing. Using millimetre waves, the full-body scanner produces an image of the scanned entity, which is then analyzed by the ATR algorithm. As an executive of L-3, the company producing these scanners, has stated, the algorithm is designed to detect ‘objects that are not supposed to be there’ (Grabell and Salewski, 2011). Yet, unlike other scanners, such as those using X-rays, which are designed to detect differences in the types of materials being scanned based on their relative penetrability, millimetre waves penetrate through the thin layer of clothing but cannot see through objects and body tissue. Consequently, most objects and body parts may appear indistinguishable in the output image. The analysis of the image cannot rely on the attributes of the different examined elements (such as colour, brightness or sharpness of contours) in order to tell body and object apart. To make the distinction between what is and is not ‘supposed to be there’ the scanner is therefore dependent on an examination of the shape of the reproduced figure. Thus, the ATR algorithm needs to rely on an understanding of what the body looks like and what its ‘parts’ are, in other words, on certain predetermined definitions of the human body, as a base-line. Since human bodies come in different shapes and forms, these definitions cannot be predefined by the designers of the algorithm and are, therefore, measured extrapolations that are determined in accordance with aggregated data from which statistically calculated configurations (i.e., ‘norms’) are defined (Valkenburg and van der Ploeg, 2015: 328). Passengers’ bodies are then compared to these normalized configurations of the human body. Indeed, when the ATR algorithm was first integrated into full-body scanner use by the TSA, the agency stated that it was designed to detect ‘anomalies’ on the bodies of passengers (DHS, 2011).

Importantly, when aiming to identify ‘objects’ that are separate from the human figure (that is, trying to distinguish between what ‘should’ or ‘should not’ be ‘there’) the algorithm uses one of two possible extrapolated figures of a body, as it analyzes ‘the anatomy of men and women differently’ (TSA, 2017). To feed this information into the algorithm, the security agent operating the scanner has to push a button indicating the sex of the scanned person (stereotypically marked in blue and pink), and misrecognizing one’s gender results in an alert: the breasts of a female who was wrongly identified as male, or the testicles of a trans-woman, for instance, trigger an alert as if they were concealing suspicious ‘objects’ under their clothes (Currah and Mulqueen, 2011). Additionally, binding items, breast forms or other prosthetics regularly used by trans people often appear as suspicious objects on such systems. Therefore, having to go through full-body scanners has become a major concern for trans and gender-queer persons. Evidently, despite not being designed for this purpose, these scanners do, in fact, reproduce the distinction between ‘properly’ gendered bodies and those that do not comply with gender norms, and should thus be understood as a form of policing non-normative gendered individuals. (Currah and Mulqueen, 2011; Magnet and Mason, 2014). As prior research has already demonstrated, these procedures significantly affect the distribution of trans bodies in space, limit people’s mobility or change their gender expression in such settings. Many avoid air travel as much as possible; others report being compelled to amend their gender performance to pass as compliant with their assigned gender in order to circumvent harassment (avoiding binders, prosthetics or makeup, changing hairstyle and accommodating body language) (Ring, 2016).

Yet, there is more than an unintended policing of gender norms here. The mounting evidence that people whose gender performance and bodies do not comply with cisnormative (Bauer et al., 2009) assumptions have increasingly been targeted has caught the attention of civil rights organizations and advocacy groups. They thus have called on the TSA to revise its scanning procedures. Such groups have highlighted the fact that this singling out has not only impeded the rights of already compromised groups of people, and exposed them to harassment and abuse (Grant et al., 2011), but also ‘erode[d] the dignity and privacy of vulnerable travelers’ (HRC Staff, 2016). Sensitive to public criticism, the TSA responded in December of 2015, by stating that in line with new directives ‘TSA officers may no longer use the term “anomaly”. The actual scanning procedures did not change and there were no revisions to the algorithm itself, but the term ‘anomaly’ was replaced with ‘alarm’ (Ennis, 2015). While this response was criticized by many, claiming that a semantic replacement does nothing to address the actual harm caused by security procedures (Margolin, 2015), the TSA insisted the difference was crucial: once a certain anatomy ‘warrant[s] further inspection’, according to their statement, ‘TSA officers are trained to clear the alarm, not the individual’ (TSA, 2017).

This change in the use of language attempted to address the concerns raised by transgender people and the advocacy organizations through a double endeavour. First, it prohibited the use of a term that carries a judgement value (anomaly) and then – and thereby – it aims to distance the suspected findings from the identity of the passenger. Unlike ‘anomaly’, which may reflect on the person being checked, ‘alarm’ supposedly distinguishes between the ‘object’ which is the cause of the concern and the passenger’s actual ‘body’. Haraway (1991) has already pointed out the fantasy of separating bodies from objects as a fabricated ideology by demonstrating that all bodies are already integrated into, dependent on, and gain their illegibility through various objects. As this separation between body and object is doomed to fail, the (queer) body becomes fully and thoroughly securitized: the composition of the non-normative body itself becomes an ‘alarm’. This is the ultimate meaning of securitizing abnormality.

The securitization of the gender queer we identified is therefore a particular outcome of an algorithm that concurrently rests on these two assumptions: that male and female bodies are neatly distinguishable (TSA, 2017), and that bodies and objects are mutually exclusive. Yet The queer body here stands for bodies as such. Theoretically, the securitization of gender non-conforming bodies could be addressed, by changing, at least, the first of the two above assumptions. However, the combination of the reliance of the threat-identifying algorithm on the apparatus of (a statistically calculated) normality, alongside the principle impossibility of dissociating bodies from objects, means that these algorithms are deemed to identify and/or produce abnormal corporealities, even if alongside different grids, since in such systems without ‘abnormalities’ the concept of ‘threat’ loses its meaning.

Demarcating the Terms II: Abnormality

For those of us who are versed in the convoluted history of how queers were treated by the state, this may sound as all too familiar. Indeed, it echoes a long history of demarcating sexual and gender minorities as posing security threats. In the United States, during the 1950s, for instance, when being gay was heavily stigmatized, several thousand gay men and lesbians were fired en masse from jobs which required having federal security clearance. The deviation of gays and lesbians from the dictations of the heteronormative society is what rendered them presumed security threats. Being perceived as queer, they were seen as belonging to a dangerous subculture, a shady habitus which lured the young, the feeble-minded, and the psychologically unstable to weaken American society by undermining family formation and nurturing moral decay (Johnson, 2004). Similar such histories can be found in most liberal countries, including Canada (Robinson and Kimmel, 1994), the United Kingdom (Davis, 1971), Western Europe (Psonak, 2000), Australia and New Zealand (Knapp, 2008).

At some level, then, the fact that trans persons are stopped at the airport by security officers for additional screening and questioning can be seen as a technologically advanced manifestation of this history. Once again, being queer denotes you as suspicious. However, if we examine the particular logic and grammar of securitized deviation in the case of the scanner, we will find a change, if not a complete reversal. In these past examples, the intersection of security and assumptions regarding normality rested on a dense social grid within which deviation appears as such. The determination that gays are unfit for public service due to security considerations was precedented by, and dependent upon, their being marked as a distinct category of social deviance. While the security categorization did reshape the social signification of these identities, it was still a secondary process. Differently put, for homosexuals during the Cold War, the security consideration was but a derivative of the frameworks of sexual normalization that pushed homosexuals to the margins of society, rendering them presumably more vulnerable to being blackmailed, for example. Thus, whilst it undoubtedly reproduced prejudice within a different field (considerations concerning espionage rather than concerning the propriety of one’s character and behaviour), the language and concepts remained the same, merely momentarily borrowed by another field (security rather than sexuality), without losing their foothold – even their root – in the discourse of sexuality.

In the case of the scanner, however, the social grid that is the hinge for the securitization of gayness above is sidestepped, overridden perhaps, and the screening procedure changes both the meaning of ‘norm’ and the apparatuses from which the norm draws its meaning. The sexual orientation or gender identity is no longer explicitly seen as a problem, and the answer (‘treatment’) no longer takes place in the domain of gender and sexuality. Instead of social vulnerability or assumptions regarding emotional instability that is supposedly tied to sexual orientation (as in the case during the Cold War), transgender persons are singled out by security systems, since they present a very different kind of abnormality. They are marked as a threat since their bodies do not adhere to the expectations of their perceived gender (such as in the case of the full-body scanners) or by the mere fact that there is a discrepancy in their documents (as in the case of the Canadian Border Crossing regulations).

Seemingly, these measures have nothing to do with conservative preconceptions regarding the stability and permanence of gender identities, but are aimed at identifying potential terrorists and undocumented migrants (categories which are increasingly fused, Bigo, 2002). All judgments concerning gender abnormality are irrelevant in such cases, as they are replaced by the seemingly objective questions of the compatibility of documents to appearance or of bodies to gendered models. However, the reliance on such seemingly objective criteria means that gender non-conformity has become a cause for suspicion. Importantly, then, what may seem to be a marginal effect of a technology that seeks to isolate security threats concealed under people’s clothing, yet accidently captures also ‘other’ people (all those who do not conform to the sex/gender binary order, or whose bodies depend on ‘objects’ to function) is an essential and direct outcome of these systems’ very logic of operation.

A very similar claim can be made when examining the category of race and how profiling-like methods reappear in the workings of the scanner. Black women have repeatedly complained that full-body scanners identify them as concealing objects in their hair, and that they are consequently subjected to thorough examinations by security personnel (Peterson, 2015). While again, it is black (women) rather than white (men or women) who are identified as a threat, the crucial difference between this and ‘classic’ racial profiling is that here ‘blackness’ appears as a potential security issue not due to preconceived racist notions of dangerousness or explicit racial biases; racial biases appear ‘automatically’ through configurations of normality embedded into these machines. That is, racial biases are reconstructed through these technologies since in a society in which race appears as an aberration from ‘whiteness’ (Bhabha, 1998), they are inseparable from what is programmed into these machines as a normal body or the normal composition of a hairstyle. What we witness here is therefore, at one and the same time, a continuation (of securitizing abnormality; of racial ‘profiling’) and a new phenomenon. In accordance with the paradigm of uniform-risk security, black women are singled out not because they are considered dangerous in any way, but by an algorithm which associates aberrations from a particular (white) body composition and thus finds their hair to be ‘abnormal’. The fact remains, however, that they are flagged as a security threat.

Thus, various forms of prejudice are nevertheless central to the operations of these systems, yet the nature of prejudice has changed in the process. This prejudice has both a different content (it primarily discerns other groups) and a different mode of operation (it is not defined a priori, nor does it exist independently of the operation of these scanners, but appears to be no more than a byproduct of the scanners’ reliance on statistically-based algorithms.)

In this regard, we depart from the two most prevalent lines of argument concerning the re-inscription of race and other social categories into seemingly objective security measures. The first sees the objective, statistical, scientific language of security as a lie, an attempt to deceive in order to obtain legitimacy. This critique claims that security systems simply enact old racisms, while employing pseudo-scientific language to legitimize their actual reliance on racial profiling or other forms of prejudice (cf. Handeyside, 2015; Lord, 2013). Yet, such modes of critique miss the productive elements of the legal and social constraints set on security, failing to see the concrete role they play in altering the security practices themselves as well as the social operation of identity categories. The second approach regards the resurfacing of identity categories within such automated calculations of security threats as no more than a technical malfunction that can be fixed given the right adjustments (cf. Lianos and Douglas, 2000; McCartney, 2014; Weinberger, 2010). Contrary to both approaches, we claim that these attempts to purge prejudice from mechanisms of security-sorting are neither insincere nor mere technical glitches, but an essential component within this logic of security that was nonetheless deemed to fail from the onset.

What is normal?

To an extent, at least, our analysis thus far hinges on an examination of the concept of ‘normal’. Foucault distinguishes between two types of normal (even if this distinction shifts and blurs at times). The first is the normal as it appears within disciplinary apparatuses (such as mental disability or gender non-conformity). This ‘normal’ functions in relation to a model, a pre-given standard of propriety, health, mental stability, identity, efficiency or productivity to which one should conform: ‘the normal being precisely that which can conform to this norm, and the abnormal that which is incapable of conforming to the norm’. (Foucault, 2007: 85). The processes of measuring against this module and adopting subjects to it he then calls normation. The second type of normal is that of biopolitics, which is, as Elden (2007: 573) observes, ‘the means by which the group of living beings understood as a population is measured in order to be governed’. This second meaning is devoid of judgement, and is extrapolated from the calculated measurement of particular characteristics: here ‘normal’ marks a certain frequency of a trait and its location on a Gaussian curve, presumably reflecting the natural order of things. Accordingly, ‘it is calculation (calcul)… which is the model for these rationalities’; (ibid) rationalities that, in turn, are connected both to liberalism and to security (and indeed the two often merge in the 1977–1978 lectures). Within this domain ‘normal’ is not defined by a pre-given social model – marking a ‘good’ or a ‘should’ to which one must conform – but is extrapolated from natural processes; it is derived from empirical reality rather than being imposed on it in order to shape it. This, in short, is the normalizing technology of security: a calculation of the frequency of a given phenomenon, which is inferred from the natural flow of things and living beings, their patterns of movement and modes of action.

The security technologies that are at the focus of this paper are guided by this latter type of normalization. Due to their highly localized nature, they do not amount to a fully-fledged biopolitical dispositive; ⁶ they are, nevertheless, integrated into such mechanisms. More importantly, these technologies operate after the fashion of biopolitical apparatuses: they operate on the level of population, with the same logic of statistical calculation and in a similar field of circulation management. At least as a matter of principle, but largely also as a matter of operation, all that the security techniques and technologies do is to single out the statistically infrequent as a security threat. Yet, when we examine these technologies and procedures, we see a conflation of the two meanings of ‘normal’ whereby these systems would tend to identify as a security threat precisely those who were already marked as abnormal in one shape or form by disciplinary apparatuses. Consequently, processes of sorting people which are based on the statistically calculated empirical ‘normal’ (which is deprived of value-judgment), collapse into what can easily be identified as a process of normation. Crudely put, this means that our analysis demonstrates how, in the settings we examine, the distinction between these two meanings of ‘normal’ cannot be sustained and they, in fact, fold into each other.

Seemingly, what we have here is yet another case in which bio-politics and disciplinary powers can be seen as working in tandem, much like in many other examples given by Foucault and Foucauldian literature: the results of tests administered to school children as a disciplinary measure are also gathered as statistics on a nation-wide measurement of population qualifications; the meticulous management of the patient’s file at the hospital becomes a component within population-wide statistics; the regulation of women’s sexuality and bodies is closely tied to the management and calculation of a population’s size, parameters and ethnic or class boundaries. All these obey a similar logic of disciplinary and bio-power converging into or enforcing each other. However, there are three crucial differences that render our case both different from such examples and indicative of a broader phenomenon: (i) the mode of convergence, (ii) the nature of the converging discourses and (iii) the effect of this convergence on the nature of subjectivity.

Firstly, unlike the examples above, as well as many other cases Foucault analyzes, the collapse of these two meanings into each other does not amount to a conversion between the disciplinary and the biopolitical in manners that mutually reinforce both. This conjunction between two different configurations of power and two distinct meanings of normalization occurs despite consorted efforts to maintain the distinction between them, and, more importantly, it occurs exactly in the apparatus of control whose justification, its raison d’etre, is its ability to maintain the distinction between these two meanings of ‘normal’. Thus, this conversion of meanings counteracts the fundamental rationale of the very same security apparatuses.

The second crucial difference relates to the nature, or the realm of the conflating discourses. Foucault identifies, not precisely the emergence of, but rather the ‘reactivation’ of ‘normalizing’ power with an intersection of psychiatric discourse and the legal system, occurring roughly in the 18th-century. With this intersection, a new form of ‘positive’ power emerges and with it appears what we may call ‘the modern subject’. The latter is also an outcome, among other historical coincidences, of a series of ‘doublings’ generated by the intersection of law and psychiatry (identity). Foucault explicitly refers to this doubling as the emergence of personality alongside offense (action), of norm alongside prohibition and of irregularity in relation to moral or psychological rules alongside transgression or violation. These doublings produce, Foucault argues following Nietzsche (1994), a depth which is the modern subject, as they shift the discourse ‘from action to conduct, from an offence to a way of being’ (Foucault, 2003a: 16). With this process, the law is transformed from restriction (manifesting sovereign power by its ability to punish and, ultimately, kill transgressors) to a productive element within a larger system, which sorts individuals into normal and deviates (categories that themselves become possible by the above doubling) and constantly works to conform the latter – but always also the former – to the model of ‘normal’ (Foucault, 1978). The subject, in turn, is no longer situated within the realm of legal responsibility, but rather within ‘a realm of mental abnormality’, as ‘the legally responsible individual is replaced by an element that is the correlate of a technique of normalization’ (Foucault, 2003a: 25). The new power that takes precedent at this moment would later be termed by Foucault ‘disciplinary’ and later still situated by him within a wider matrix of several normalizing powers (Foucault, 2003b).

Following many Foucauldian scholars, our analysis carries this paradigm to a contemporary setting, in which the modalities of power, the plateaus on which the different modalities meet and the effects of power (on both the subjects and the manifestations of power) change. While Foucault identifies an intersection of psychiatry and law as the moment that both symbolizes (if not encapsulates) and plays a role in the formation of the modern subject, we identify a different intersection: an intersection between the discourse of security and the law in which, we argue, the psychiatric discourse no longer plays a part. This intersection is key to understanding the politico-technological negotiations we have examined above. The technological changes in full-body scanners we have outlined in the second section follow the contours law dictates. Thus, legal concerns regarding discriminatory practices, as well as the safeguarding of personal privacy and the production and distribution of nude images, intervene in reshaping the fashion and means through which security is manifested (Elias, 2012). In particular, the demand to cleanse security practices of anything resembling discrimination is essentially the requisite that security will not operate after the fashion of psychiatric discourse, that is, the sorting into distinct groups. Put differently, it is law that moulds the principal inability of security, in these contexts of the uniform-risk paradigm, to resort to identity-based categorizations for the purpose of profiling, and that poses the need to come up with objective and universal criteria.

The exclusion of psychiatry and its modes of action from these particular sites of governmentality explains the changes in the securitization of queerness to which we point at the end of the previous section: the intersection of law and security does not produce the doublings that turn acts into subjects. Here transpires the third difference from more classic Foucauldian formulations: the effects of this conjuncture on the nature of subjectivity. This is not to argue that there are no subjectivation processes involved. If one follows the Foucauldian logic of power as productive rather than merely excluding and restricting, one should seriously consider the possibility that such new intersections would give rise to new forms of subjectivity. As Salter (2007) clearly demonstrates, these local contexts also yield themselves to complex, and much wider processes of subject formation. Furthermore, in highlighting the interaction between law and technology, we do not seek to argue that it has replaced the intersection of law and psychiatry or the disciplinary understanding of normalization. Despite the increasing prevalence of security-oriented encounters, and despite the radical transformations in disciplinary systems (Deleuze, 1997), disciplinary subjectivisation has yet to disappear and should be understood as working in tandem with the processes of securitization we identify here.

Yet such processes take a different form in this context. Unlike the systematic and comprehensive effort of disciplinary technologies, which are embedded throughout social institutions to shape the subject thoroughly, pervasively, operating on her deepest levels (indeed, technologies which are set to create the subject as having a deep core, Foucault, 1979), security technologies meet the subject sporadically and momentarily – when the scan takes place; when the border is crossed. These encounters are not systematized within coherent frames aimed at producing a uniformity within identity categories. In these contexts, the relevant identity categories are therefore flattened, to an extent, and are displaced onto a different social sphere. The outcome is dual: first, with the side-lining of the richer social context within which disciplinary technologies are embedded, there is also a shrinking of venues for resignification and re-subjectivization that are at the disposal of the bodily queer in other contexts. The earlier ‘transvestite’ and the later ‘gender identity disorder’ or ‘gender dysphoria’ were reclaimed through self-assigned identities such as transsexual, transgender or the gender-queer (to name just a few); such acts of reclaiming are not effective modes of resistance when categories of identity are sidelined by categories of risk. Secondly and relatedly, rather than a complex system of signification and resignification, bodily meanings are organized around the binary distinction between the threatening and the benign and corrective treatments are replaced with the violent suspicion reserved to the suspected terrorist and other enemies of the state.

This folding of the two meanings of normal into each other, this failure of the scanners to deliver on their stated goals to eradicate the discriminatory practices through producing an impartial gaze, could presumably be explained as a limitation of technological design. Critical research of technology has demonstrated that the design and assemblage of technological innovations cannot be seen as being void of social preconceptions (Valkenburg and van der Ploeg, 2015). Machines are often shaped under implicit assumptions that their users would share the same social categories of their designers, and, inter alia, their perception of the composition of human bodies. However, the example of the scanner goes beyond demonstrating the biases embedded into technologies that reflect the preconceptions and prejudices (implicit or otherwise) of their designers (in the example here, the assumption that men and women should be scanned differently). Such biases can presumably be minimized by more political awareness and social sensitivities. Yet, as the logic of operation of the algorithm is designed to identify threat with deviation (from the ‘normal’ body or ‘normal’ human behaviour), it is bound to reproduce the securitization of abnormalities in one form or another.

This occurs since the objectively calculated normalization would necessarily replicate the categories of normation. This assertion rests on the claim that processes of empirical (statistical) normalization of the body, measuring human behaviour and constitution, are irrefutably entangled in the ways in which the body has been disciplined and categorized, deciphered and signified. This entanglement, queer theory teaches us, is always already immersed in normation processes. Bodies can be sorted, measured, compared and averaged only after they have been normalized; only after they have been construed by the categories that render bodies intelligible and are, thus, the effects of prior disciplinary processes (Butler, 1993). At least when engaged in the particularities of bodies, then, the second type of normalization (that of biopolitics) unavoidably carries with it the first type (of discipline). What we have here is a technological manifestation of Butler’s structural claim that the liberal paradigm of inclusion can never achieve its promise: there will always be forms of exclusion. Even if such algorithms were designed under different sets of assumptions concerning the structure of gender categories, abnormalities of some kind would necessarily still be produced by these technologies and marked as a security problem (be it heart rate, body heat, size, mobility or functionality for instance). As we have argued, without such a production, there would be no meaning to “threat” within this paradigm.

Conclusions

In this paper, we have sought to show that the uniform-risk security logic we increasingly encounter rests on, and brings about, new meanings of ‘normality’, as well as new implications for the classifications of identity categories which are marked as ‘deviant’. This is not simply a process in which one set of (disciplinary) classifications is superimposed on, or infiltrates into, another field (that of security), but a productive process in which these distinctions are blurred and fuse to produce a new articulation of abnormality. As we have discussed, disciplinary power and biopower/security often indeed converge and work in tandem within Foucauldian paradigms, yet the two meanings of ‘normal’ obtained by these two configurations of power remain distinct. While one is a predefined and an ethically-loaded model that dictates judgement based on one’s ability to conform to it, the other is a purely empirical measurement, extrapolated from the order of things. The first creates distinct categories of abnormality in well-defined fields, while the second operates in diverse milieus. Our analysis, however, demonstrates a process by which the categories of abnormality from the disciplinary field emerge within the biopolitical logic.

As Warner (1999) shows us, a critique of power must maintain a distinction between disciplinary and bio-political modes of normalization, since a conflation of the two meanings often serves to render transparent the operation of social powers. Being normal in terms of one’s gendered and sexual identity, for instance, makes little sense from a statistical point of view, neither as an aspiration, nor as a lived reality. What we have shown in this paper is a case in which the two meanings of normal fold into each other, and disciplinary categories re-emerge as statistical aberrations within the biopolitical realm. Importantly, in their re-emergence these identity categories appear cleansed of the historical, social and cultural contexts in which they were construed and in which they gain their meaning. This decontextualization depoliticizes such identities, and masks them as value-free and merely numeric, yet within a field wherein risk is defined as abnormality. The result is that discrimination, state violence, detention and removal are reinstated as a treatment for ‘aberrant' identities. Not only does the depoliticizing effect of such security apparatuses have theoretical implications, but it also defines what would constitute effective resistance to such practices of exclusion. From our analysis herein, we could begin postulating that framing resistance in the language of rights or demands for accountability by the relevant state-agencies, or focusing on struggles for signification and representation fail to address the core issues that are at stake here. While articulating this claim is evidently beyond the scope of this paper, these are the exactly the questions that urgently need to be rethought.