I’ve written an essay for @techpolicypress on the SIR and the faith in digitisation that has shaped Indian elections. I wanted to understand how decisions about who can vote came to depend on software whose workings are closed to public scrutiny. For the research, over some weekends in the IIC library, I read books by former Chief Election Commissioners. I enjoyed them, particularly the accounts of the logistical work involved in conducting elections at this scale. Technology, though, is almost always presented as a success. For instance, former CEC S.Y. Quraishi recalls facial recognition software in Manipur detecting a woman who had come to vote more than 60 times in different guises. H.S. Brahma began linking voter IDs with Aadhaar on the promise that almost every complaint about the rolls would be resolved. These accounts say much more about what technology promises than about its limits.That faith extends well beyond the Commission as across Indian public policy, digitisation is routinely offered as the answer to failures of administration (in panels called, "state capacity constraints"). However, when the rules, or more properly the updates in the software builds in a system are kept secret, or changed to vary from the officially notified forms, however, discretion becomes harder to examine. People affected by an arbitrary decision have fewer means of understanding or challenging it. I also went back to an earlier attempt to make voters prove their entitlement to remain on the rolls. In 1992, the Commission under T.N. Seshan instructed its officers that the burden of proving citizenship lay on the voter. An officer in Delhi then declared an entire slum colony of 18,000 residents suspected foreigners. More than 16,000 lost their vote. In Lal Babu Hussein in 1995, the Supreme Court held that an existing entry on the roll carried a presumption that the registering officer had followed the law. The intensive revision of 2002–04 proceeded on that basis. Enumerators did the verification, and existing voters were presumed to belong on the rolls. The SIR reverses that presumption through a process dependent on software. Today, more than 130 million names have come off draft rolls. The software treated “Md” and “Mohammed” as different people, flagged siblings born less than nine months apart, including twins, and found seven siblings sharing a father suspicious. Its rules, error rates and audit results have not been published. Voters are left to correct these errors, often by standing in queues and responding to notices in languages they cannot read. Nandan Nilekani, whose book advocated a centralised voter database and digitisation, has himself received an SIR notice. Finally, I would like to thank Tech Policy Press and its patient editorial team for being kind and gentle with me. Please read the essay and let me know what you think of it. techpolicy.press/indias-sir-…