of Assam; while less than one percent of those enrolled in Aadhaar did not already have at least one form of identity document (State of Aadhaar 2017-18, 2018, p. 5; State of Aadhaar: A People’s Perspective, 2019, sec. 2.2). Indeed, it seems as though the original concept may be turned on its head: rather than providing a means to prove identity delinked from legal status in the country, the government of Narendra Modi is reported to be planning to link the Aadhaar platform to the controversial revival of a National Register of Citizens (NRC) and to refuse Aadhaar numbers to those not in the NRC (Chakravarty, 2018; Masiero, 2019). Legal identity and the recognition of citizenship What can be done to avoid the undesirable outcomes, and enable the envisaged positive impacts of ‘identification for development’, especially in countries where state capacity is weakest and the rule of law hardest to establish? There are already numerous examples of better practice. Several of the former French territories in Africa have quite strong due process protections in decisions relating to the determination of nationality, most importantly the authority of a tribunal to determine who is a citizen from birth and who is not. In some jurisdictions, this includes the power to recognise nationality on the basis that a person has always been treated as a citizen in practice (is in possession d’état de national), a procedure that responds realistically to a context in which civil registration was and remains very incomplete (Manby, 2018a, Chapter 6). These models have already been drawn on as inspiration for the design of procedures to mitigate the extreme executive discretion that is the heritage of the former British territories – for example in Kenya, to implement new provisions in the 2010 constitution for the grant of nationality to children of unknown parents (Manby, 2018b). We can also draw on the lessons from international support for democratic institutions, in particular for independent electoral commissions. The existence of a body named an ‘independent electoral commission’ is clearly no guarantee of genuinely free and fair elections, without considering the broader context. Yet it is not for no reason that African citizens and opposition parties have spent several decades insisting on the principle of independent election management in negotiations over new constitutions and electoral legislation. We can extend to identification systems generally the conclusion that ‘the greatest gains from digitization come in countries where the quality of democracy is higher and the electoral commission more independent’ (Cheeseman et al., 2018, p. 3; see also Piccolino, 2016). This argument is particularly significant because voter registration cards have functioned as informal identity cards in countries where no centralised national identity card was established until recently. In Africa, these states include some of those which also have the lowest rates of birth registration, among them Democratic Republic of Congo, Ethiopia, Liberia, Malawi, Nigeria and Tanzania. When a new national identity card has been introduced, voter cards are generally not accepted as sufficient proof of an applicant’s citizenship, even if they feature as supporting documents (in a mass enrolment exercise in Malawi, a voter registration card contributed 40 points towards a total required score of 100 (Malik, 2020)). At the same time, voter registration cards are no longer accepted as identity documents for the range of purposes for which they were previously used. Even in states that have always had a national identity card, the effort to ensure that all adults are enrolled in a new system – and to insist on proof of enrolment to access services – means 14