The Truth of the Error

Making Identity and Security through Biometric Discrimination

ELIDA K. U. JACOBSEN AND URSULA RAO

As moderns, our task and our obligation is to be attentive to ways around knowledge that claims to be universal when it is contingent, unified when it is at best partial, and autonomous and sovereign when it is dependent and immature.

— RABINOW, French Enlightenment: Truth and Life

"This had to be a fraud! Or could you imagine a family with over 100 members?," the quality assurance officer Amit Chatterjee asked rhetorically. What has happened? The computer at the central data-processing unit of the Unique Identity Authority of India (UIDAI) showed an error. The automatic filter of the new biometric registration system highlighted more than 100 individuals as problematic because they had all been authorized by one single Head of Family (HoF), indicating that this person had proof of being directly related to each individual.

During an interview, he recounted in an animated fashion the story about how their initial astonishment concerning the extent of the fraud had turned into curiosity and, rather than cancelling the enrollments straight away, the team decided to travel to the region and find out what had happened. When they reached the distant border region in Mizoram, in the northeast of India, they found to their surprise the "biggest family in the world," totaling 184 members. "Can you imagine a man with 39 wives?" Amit repeated dramatically and continued to marvel at how enrollment for India's new biometric database has brought the team in touch with even the most remote people of India.

Today many countries are experimenting with biometric identification systems that use smart cards or central databases. With over 1.2 billion enrollments to date, India's aadhaar (Unique Identity, UID) is not only larger than any other similar project but is a "frontier case" that will influence developments in other countries, such as Indonesia or Papua New Guinea (Jacobsen 2012; Zelazny 2012; Gelb and Clark 2013b). According to the World Bank, it is also pioneering because it promises to achieve maximum interoperability by linking a national ID program to multiple sectoral interventions, such as welfare projects, security operations, or commercial applications (World Bank 2015). And indeed, the notion of interoperability captures well the ambition of the architects of UID, who launched the project in 2009 to provide a streamlined means of identifying India's entire population and linking millions to national digital networks of information. The system is meant to biometrically enroll all residents of India and give every person a unique twelve-digit identification number (aadhaar number) that is connected to a record containing their personal biometric data — fingerprints, iris scan data, and photograph — and to a skeleton set of social data — name, address, and gender. It can be used for online verification of identity at any time and any place. Proponents of UID are confident that the new technology will solve India's identification crisis by supplying reliable information to public and private service providers about who is who, thus making all transactions transparent and secure. Currently the aadhaar number is required for identification in most official contexts, such as applying for a passport, receiving welfare payments, or getting a bank loan. It can be used for instant activation of a SIM card, purchasing a train ticket, or conducting internet transactions (Bhatia and Bhabha 2017).

India's biometric project is participating in a global shift toward states using new digital technology in the management of population flows. The contemporary world capitalist system not only depends on the rapid flow of people and goods, but also produces heightened concerns over the unwanted movements of illegal migrants, terrorists, or smugglers, persons who may present a threat to national security and prosperity (Fuller 2003). Biometric technology provides automatized surveillance at crucial checkpoints in order to protect spaces of privileged sociality against unwanted entrants — in short, it is a means to separate "bad" flows from "good" flows (Aas 2006; Amoore 2006; Lebovic 2015; Amicelle and Jacobsen 2016). While surveillance studies scholars analyze the increased usage of networked biometric technologies in managing risks and contingencies (Amoore 2006; Muller 2011; Jacobsen 2013; Lidén, Boy, and Jacobsen 2016), development studies literature emphasizes the role and societal effect of biometric technology for creating more efficient and fraud-free welfare states (Rao 2013; Donovan 2015; Singh and Jackson 2017). The introduction of security logic into welfare contests follows on from neoliberal suspicion about wasteful states and worries over inefficient targeting, corruption, and leakage. By tracking goods and people, governments seek to undercut false reporting or "double dipping" — the illegal diversion of limited resources that impoverish states and contribute to distributional injustice. Regardless of whether biometric surveillance systems face inward or outward — that is to say, to include the undocumented or to exclude unwanted foreigners (Breckenridge 2014) — they fulfill a key purpose of making transactions traceable by employing a binary distinction. On the one side, there is the production of the documented person, the wanted traveler, or the needy citizen, which is mirrored on the other side by its opposite: the imposter, the fraud, or the criminal.

The attractive clarity of the binary logic of biometric classifications is disturbed every time technicians or users encounter an error. An error appears as a red warning on the screen when fingerprints are unreadable or recorded data appears as incoherent. The story from Mizoram is a case in point that was resolved positively, unlike many other cases of data errors that have led to rejections of aadhaar registration because applicants wishing to enroll have washed fingers, damaged irises, or unlikely names. The "failure to enroll" has its complement in the "false reject" of...