When Biometrics Fail: Gender, Race, and the Technology of Identity
by Shoshana Amielle Magnet
Duke University Press, Durham and London, 2011
224 pp., illus. 18 b/w. Paper, £60.00/£14.99
Reviewed By Hannah Drayson
University of Plymouth
Biometric technologies used for the confirmation of individual human identity, are a persistent example of a technological vapourware, probably more familiar to most of us from science fiction than from their slow increase in use in personal travel documents, such as passports and fingerprint locks on cars and laptops. However, particularly in the post 9/11 United States, the biometrics industry is currently expanding, landing multi-billion dollar contracts in border control, adding to its existing footholds in law-enforcement, welfare, and prison management. Focusing on five main themes surrounding the technology, When Biometrics Fail provides an historical overview of the growth of the industry since the 1970s. Using evidence from a range of contemporary sources, journalism, media and recent research published on biometric technologies, Magnet demonstrates rather compellingly how these technologies fail to live up to Daston and Galison’s (2007) concept of ‘mechanical objectivity,’ the idea that machines are capable of making unbiased judgements owing to the privileged link to the physical world (and lack of subjectivity) with which they are associated by marketers, computer scientists, politicians, and the media.
To this end, When Biometrics Fail offers a damning analysis of the technical problems that dog biometric identification, many of which stem from the premises on which individual difference is defined. Magnet explores the implications of these limitations with reference to the range of sites in which biometric technology has been implemented, exploring the social problems that it is posited to be able to solve and showing that in many cases its effect is only to hide, or exacerbate, these problems. As Magnet argues, biometrics fail because they are discriminatory both in the methodology used to identify individuals and also in the social implications of the manner and sites in which it is implemented (prisons, welfare, border control). Further, there is a wider and deeper problem throughout that cannot be separated from the bias of the users of the technology. This is that the basis of the whole enterprise is on the false premise, the idea that the biological body can be fixed upon as an enduring, digitisable, and recoverable document, an idea that has been challenged in the scholarship of thinkers like Donna Haraway and Judith Butler and invoking the critical concept of ‘corporeal fetishism’. It is this model of biological identity as an essential and fixed quality that is both out of date in so far as the humanities have understood the body for considerable time (notions such as gender are no longer considered only biological rather than performed) but also positions the body as a commodity.
In four short chapters Magnet makes clear the practical impact of these assumptions and faulty models of biological identity, both dated and troublingly linked to discredited scientific ideas of biological race and practices of profiling and reading the body such as physiognomy. By discussing these assumptions as they manifest in the way in which biometrics are applied, Magnet identifies a range of cultural problems that stem from the acceptance of the technology as a cure-all. She is also not slow in pointing out that the research in this area is painfully unaware of how out of date its approach is; the concepts that she draws upon to question it are now basic to the canon of gender and technology studies. In addition to this, the models of objectivity that these technologies invoke in order to state their authority are long discredited and highly simplistic in scientific terms.
The opening chapter analyses what can be seen as broadly discriminatory aspects of how biometric recognition is achieved. One example, retinal scanning, is only effective on light coloured eyes because of light reflections in darker corneas that interfere with the camera’s view of the veins in the retina, making the technology as it has been developed in Western laboratories, biased in favour of a particular population. Magnet very effectively places a number of these problems that seem to constitute in-built biases in the technologies, in contrast with the arguments of industry professionals and law enforcement officers automated biometric systems are superior to humans because they are unable to discriminate. Technical limitations that are easily brushed under the carpet in a scientific lab or paper, take on a life of their own when the technologies are implemented, and the assumptions that underlie them are played out in the real world. In an analysis of the techniques of biometric identification, the apparent naïveté of current research and biometric identification techniques shows the extent to which cultural assumptions and essentialist paradigms underlie the methods by which researchers identify and train systems to make identifications. For example, the use of 'soft’ biometric differentiators, i.e. sex and race, invoke what appear to be extremely limited ideas about which detectable characteristics might offer evidence of male or femaleness, such as hair length, and signifiers related to dress and other markers of gender (such as an individual’s décolletage and the amount of skin showing or wearing a tie).
Magnet also discusses the industries’ exploitation of U.S. prisons as a testing site for the implementation of technologies like hand geometry scanning. In a chapter titled “Acres of Skin” (after Allan Hornblum’s 1999 text on the exploitation of American prisoners as medical research subjects from the 1940s to 70s) she discusses how this method of counting prisoners (an activity undoubtedly carried out for their own safety as well as keeping track of their whereabouts) interferes with prisoner’s rights to refuse to give biometric data but also results in prison spaces where the watcher is no longer human and no longer able to infer that prisoners are protesting, hunger striking, depressed, or sick, or even (as familiar from science fiction, and as Magnet tells us, in recent crimes) still attached to the body part being scanned. Another case study shows how the industry has targeted another state apparatus to which another large population is subject, the U.S. welfare system, with the implementation of mandatory fingerprint scanning in a number of states. Superficially the goal of this has been to reduce the number of individuals making multiple claims for assistance and have on the surface appeared successful. Enrolment in the benefit system has dropped. However, as Magnet points out, this change has been argued to be a result of the discrimination against immigrant populations who are unwilling to submit for fingerprinting because of fears that this will interfere with their applications for permanent US residency (p.88). Therefore, the apparently unintended side effects of implementing these technologies appear to have a far less ‘objective’ and non-discriminatory consequences than their supporters argue.
As Magnet describes it mechanical objectivity is “a visual trick to hide subjectivity from view” – a subjectivity that is associated with discriminatory bias (a bias which she points out appears with the introduction of new anti-terrorism laws post 9/11 to have become more and more clearly detectable in the attitudes and behaviour of both UK and U.S. police). The ideal of a machine system that can identify both individuals and specific traits without the involvement of a potentially biased human observer is one that, as Daston and Galison (2007) tell us, a notion was historically short lived in much of the sciences; she does not note, although it is an observation often made by those studying a range of imaging technologies focused on the body (Dumit, 2002; Saunders, 2008; Joyce, 2009), and that this ideal of objective machines is regularly invoked by developers, users and salespeople.
Some aspects of this critique could be explored further, and the text offers much evidence that would support further meta-analysis, particularly in terms of how the technological functions of biometrics are understood. While we might accept that subjectivity invading the process of identification is a failure of the technology, an anti essentialist position would question the legitimacy of any technological system of categorisation that makes a claim to truth. In the case of biometrics, identification technologies offer a trade-off between what is affordable, possible, and pragmatic, which can be enriched or limited by ideas of what can be mapped onto or discovered as manifest in the body. In particular, the concepts of power and institutional control that themselves impact on the way in which these systems are understood, interacted with, or resisted by their users that seem central to this discussion are not invoked as often as they might be.
The synthesis offered by How Biometrics Fail suggests a further reappraisal on philosophical terms of what these technologies actually are, how they factor into human action and experience as well as being themselves expressions of a particular way of seeing. The book, therefore, offers further compelling evidence of the entanglement between the social, ideological and imaginary functions of technical systems and their purported use, but apart from its discussion of the transformation of the concept of the international border, generally it does not offer an alternative understanding of biometrics, taking them only at face value, as technical devices, and not repositioning them as the manifestations of particular set of ideas or desires on the part of their makers and users.
Magnet argues in her final pages that biometric technology does not need to be perfected, because it cannot ever achieve what it claims, to map a range of cultural ideas onto a biological, but ever changing, body. This is where the thrust of the book, and its intervention lies, not in discussing the technical failures of biometrics, but in identifying some of the key and catastrophic problems caused by a partially fictional technology that claims to do the impossible. The result is a useful synthesis of the activities contemporary biometric industry, providing a compelling dissection of how the idea of a science of biometrics fails, and a timely critique of an industry and its claims which exploit a misconception about the human body and use technological fetishism to posit the solution as a high tech – and therefore ethically unproblematic -solution to a range of problems that have serious political and humanitarian consequences.
Daston, L., and Galison, P., (2007) Objectivity. New York: Zone.
Dumit, J., (2004) Picturing Personhood; Brain Scans and Biomedical Identity, Princeton and Oxford: Princeton University Press.
Hornblum, A.M., 1999. Acres of Skin: Human Experiments at Holmesburg Prison. Routledge.
Joyce, K.A., (2009) Magnetic Appeal: MRI and the Myth of Transparency, New York: Cornell University Press.
Saunders, B., (2008) CT Suite: The Work of Diagnosis in the Age of Noninvasive Cutting, Durham & London: Duke University Press.