Subhashish Bhadra has handed out £9m in grants for digital identity research from the Omidyar Foundation over the past three years. He started by talking about seat belts, for which a patent was filed by Claghorn in 1885; Ford offered it from the 1950s and it became mandatory in 1965. But only 10–15% of people wore them, and eventually State Farm Insurance sued the government to get them to bring in mandatory seat belt laws. The design upgrade of inertia reel belts helped, but the key was the growing body of research into accidents. Similarly, getting decent ID systems fielded, especially in less developed countries, will take both technology and patient research into what works, in order to guide laws and judgments. Subhashish sees good ID as being governmental (or at least “issued”, such as university ID) while de-facto ID is the data trail. But digital identity and privacy are both important and sensitive topics.
My talk summarised what I say about naming in my security engineering book. Naming people (and things) poses a trilemma: names can be human-comprehensible, unique or distributed; in fact you can choose any two of these. When exploring what might work, Needham’s ten laws of naming may be helpful. The function of names is to facilitate sharing; The naming information may not be in one place, so resolving names brings the general problems of a distributed system; It’s bad to assume that only so many names will be needed; Global names buy you less than you think; Names imply commitments, so keep the scheme flexible enough to cope with administrative changes; Names may double as access tickets, or capabilities; Things are much simpler if an incorrect name is obvious; Consistency is hard, and is often fudged; Don’t get too smart. Phone numbers are much more robust than computer addresses; In general it’s a bad thing to bind early if you can avoid it.
Eddie Higgs was next, talking about the history of physiognomy. Aristotle believed that a beautiful soul would be in a beautiful body with the right symmetries; and physiognomy was accordingly codified by Lavater in the 1770s. Slavery and empire were justified by arguing that large African jaws were animal. Galton overlaid photos of criminals, syphilitics and Jews, and was the father of eugenics; appearance was genetic and predisposed you to particular lifestyles. Echoes of such ideas have come back in the digital age with various research groups talking about significant facial correlates of PF scores, whether in AU facial geometry features. Wang and Kosinski’s work on gay faces brought this to public attention in 2018; their view is that facial masculinity is an effect not just of fetal androgen levels but life history (which has echoes in classical Chinese writing).
Wendy Hunter has been studying the 7% or so of US citizens whose birthright citizenship is undermined by their lack of documentation. There are gaps in birth registration; the liberal state’s laissez-faire approach has side-effects. Birth registration became mandatory only on 1935, it’s a state matter, it’s poorly coordinated, and late registration is hard. If you never went to school, or saw a doctor, or got vaccinated, you can have problems. This is not just about children of illegal immigrants; the largest group is probably elderly African Americans, then Native Americans who thought their tribal cards would work but don’t, then prison leavers who just have no papers. The foster-care system is a mess in most states. The latest is the children of parents trying to live off-grid, some of whom just can’t get passports. Now delayed birth registrants in Texas (common enough given their midwifery system) can’t get passports as a matter of Trump administration policy. Wendy believes that the government must allocate serious professionals to the task of delayed registration rather than leaving it to volunteers and churches; and it needs serious coordination between the federal government and the states.
Amba Kak studies digital colonialism, which is not quite the same as the techlash! A growing number of serious policy documents from India, China and even the EU now tackle big tech practices as a colonising force. Xi Jinping talks of cyber sovereignty; he tells audiences not just in China but in Russia and Brazil that the Western narrative of their openness versus Chinese censorship is simply a cover for creating space for tech companies to predate on China. The European data-protection supervisor, Giovanni Buttarelli, prefers digital colonialism to surveillance capitalism as his opponent when promoting communitarian approaches where data are managed by collectives in the public interest. As India will be data rich before it’s rich in economic terms, this is a critical issue for Indian policy; it initially blocked Facebook’s Free Basics with the free-market argument that the country needed a level playing field, but the rhetoric now is unashamedly mercantilist.
Ian Watson started out by exploring Icelanders’ love of their ID number and has since become a scholar of all the Nordic ID traditions; his students have added English accounts of the Finnish and Norwegian numbers. In all countries, your number is more important than any card or other token. In Iceland you need someone’s ID number to send them money, as it’s embedded in their bank account number; this isn’t the case in Sweden, although people are open about there numbers there too, and in Finland. The Norwegian and Danish systems are slightly more closed.
Discussion started on the roots of Scandinavian ID systems in national statistical offices, and their takeover by tax offices; the various arguments for data localisation, sometimes about national sovereignty, at other times about business convenience, and also about their intersection in protecting local business. Many LDCs also lack decent data protection. At the core though is incentives: people register in Scandinavia because of the huge welfare state, and in South Africa because of the child support grant, but in many LDCs there’s no real reason to bother. Should a liberal state allow people to opt out of registering their children, or compel registration? There is also the issue of policy priority: is the main goal to ensure that all citizens have the vote, or to ensure that everyone has the right foundational documents? As Wendy’s talk brought out, you end up with quite different outcomes, and the choice is fundamentally a policy one. And finally, human beings have an insatiable appetite for divination.
Edwin Abuya gave us a survey of issues around ID and privacy in East Africa. In a Kenyan ID case, plaintiffs sued the attorney general and director of national registration and Edwin asked us not to blog the details as it’s in progress. In Burundi, journalists sued the AG, in a case about the protection of their sources. In Uganda, a farmer sued after her bank got hold of her ID photo and used it on an advertising billboard. One of the cases has already got to the East African Court of Justice. Practising lawyers start from the desired remedy and then try to build the case to get there; if they find missing steps in the ladder they make human-rights arguments that statute law is deficient.
Gautam Bhatia has been involved as a lawyer in two of the big Aadhar cases. In the Right to Privacy judgment the court ruled that the principle of proportionality should guide the trade-off between privacy and state goals; the Aadhar judgment applied a four-step proportionality test and ruled out the compulsory application of Aadhar in a number of applications including school exams and bank accounts. On the legislative track, the Srikrishna Data Protection Bill has been published but has attracted criticism including for the lack of a right to an explanation and the weakness of enforcement against the state. Gautam is also concerned about the independence of the enforcement body.
Marianne Henriksen described the Norwegian system. In 1964, municipal registration was centralised; there are now 5.3m residents and 2.2m “D-numbers” including workers, immigrants refugees staying less than six months. It supports a huge range of services. Their “identity triangle” links the ID number with biometrics captured when passports are issued. They have been trying to issue “Norwegian identity” cards for fourteen years now but there is resistance, even among refugees who might be significant beneficiaries. She would prefer it if service providers would insist on biometric ID but some significant services, such as the tax authorities, won’t. (The tax authorities also resisted the failed ‘Verify’ programme in the UK; perhaps, to see the ground truth, we need compare policy for systems the government cares about, like tax, with the rest.)
Jonathan Klaaren is a South African lawyer who has followed the litigation against Net1 in the Constitutional Court. Privacy arguments there had no traction, as the information was distributed in the cards they issued for social grants; yet this firm had grabbed hold of grant distribution for 16-17m South Africans, and in effect the CC nationalised it after the Competition Commission passed and the Information Commission had no staff or capacity. A supervised remedial order moved the grant stream back into state control. Courts can entertain broader public-interest arguments than classical antitrust market regulation. This isn’t a north-south issue but where privacy and competition overlap there can be regulatory arbitrage. Kenya has just worked through modernising its ID and privacy law; South Africa is at the white paper stage of a redrafting.
Ranjit Singh has been doing a PhD on Aadhar since 2015. Aadhar started out as a general program to identify people, without a stated bureaucratic purpose; the Supreme Court has since defined its purposes to some extent by limiting them. The authority that runs it deliberately collects minimal information about each resident that it enrols; that establishes uniqueness. Seeding is a secondary process whereby the Aadhar number is added to other public and private database records in order to clean them. The third step is authentication: the authority records only the time, the requesting entity and the response, not the location or the purpose. What’s more, nic.in requests identity checks for a lot of government departments, shielding the metadata. This privacy by design led the Supreme Court to conclude that the architecture, together with the Aadhar Act, do not tend to create a surveillance state. Yet although Aadhar is not an archive, it is becoming a common index to multiple archives; he believes the court was wrong to miss this. Ranjit argues that Aadhar should not be blamed when, for example, a poor person suddenly fails to get welfare grains from which she got welfare grain for years; there the blame should fall on the specific government application systems that now call Aadhar.
Edgar Whitley introduced the afternoon session on identity for credit. Can credit scoring systems substitute for national ID, and how do you deal with people whose credit files are thin? In any case, how do you link credit histories dependably to individuals, especially in the absence of good ID? Is the goal to provide national identity systems, or to provide identity systems for nationals? In a global world there’s a big question about your nationals living abroad, and foreigners resident in your country. And in 1975 Paul Armer noted that if you wanted to build an unobtrusive system for surveillance, you couldn’t do much better than an electronic funds transfer system. Is that where we’ve got?
Johann Bezuidenhoudt works on financial inclusion. Cash was inclusive and pervasive, but electronic services exclude many people, include those with no formal national ID. What alternative approaches can be accepted and included? You’d need the means of re-establishing a functional authenticator, such as by getting a new SIM card if it’s stolen; and authentication good enough to block imposter theft. Practically, you need value and volume limits, back-end analytics to mitigate smurfing, and profiles of location and use. Authentication is probably a SIM and re-issue may be a voice biometric. A good financial inclusion product could be bought in a supermarket from cash; “Smart” was fielded in the Phillippines in 2000, but later closed. In Nigeria you can still self-assert to open a mobile money account. Could use de-facto, chained, associative, situational and intrinsic (biometric) ID: let’s tighten the definitions. The imperative is to break the KYC status quo. M-pesa would never have got to where it is if KYC had been as tight ten years ago!
Roland Claussen has been studying the predictive power of credit scoring. The effectiveness of institutions depends on their social embeddedness – the degree to which people trust it and accept their role in its ecosystem. The other factor in his research is the specificity of contextual data, which greatly boosts the power to predict future arrears. Big data aren’t needed. You can use existing socially-embedded institutions such as the 25k existing microfinance institutions all over Africa. His company is providing software to try to standardise all this in one app.
Gabriel Deval worries about the thinness of credit markets in Africa; Zambia has only 500 mortgages and 2,199 business loans were made in 2017. There is so little credit flowing through banks into the private sector that it makes no difference, and the central bank statistics are generally of little value. There are issues with personal and company registers, asset registration, credit bureaux and contract enforcement. Tanzania has over 18% loans to MSMEs defaulting and over 40% to agribusiness; it has only 3,600 mortgages (which are hard when you can’t identify people or trust the land registry). Business practices like invoicing and invoice discounting are also tough. The basic problems are legal and political; without fixing them, tech fixes are pointless – unless perhaps we allow some “Wild west behaviour” to get the car moving, and worry about the seat belts later. Digital finance might hope to make a difference by getting its hooks into distribution and collection.
Bulelani Jili has been studying the Chinese social credit system – an attempt to create a credit market with socialist/Chinese characteristics in that it would detect political expression as well as spending history. It does seem to be helping with access to credit as well as helping deal with governance in China. Collaboration between state entities and private actors has helped create punishment mechanisms but the punishment comes mostly from violating industry norms and codes. In the case of Ethiopia, the government is also trying to instrumentalise data as a tool of social governance, and privacy laws are also weak. However Bulelani’s concern is not so much a social credit system as the bulk surveillance tools sold by firms in Italy and Germany – together with the training and doctrine around their use, which comes from the USA.
Discussion ranged from the extent to which social infrastructure should be private or run by the state; why nobody seems willing to push back on the FINCEN requirement that everyone show ID and address proof when opening a bank account; the weakness of the underlying authentication of most mobile money systems; the South African experience that fixing property registration, courts etc gets stuff to work better than messing about with fintech; the fact that many big-data startups like Wonga went bust or nearly so, while successful firms like Safaricom use simple credit scoring; the huge amount of credit used in sixteenth-century England when there was a shortage of specie and there were a million court cases for four million people; the use of government-mandated systems, such as fingerprint readers in South Africa, being used by crooks to loot people’s bank accounts; the real problems facing firms in Africa; and whether the west’s history of growing building societies might be of relevance, or whether it just invites a replay of Victorian banking scandals or the Mafia.
Bidisha Chaudhuri and Syed Mohammed Faisal have been studying Aadhar takeup. It’s available as a paper form you can laminate, and also as an Aadhar card and as online biometric checks: but in practice the paper form is demanded and sometimes even the card is not enough. Your functional ID, namely your ration card, is demanded too. In effect, Aadhar has become something demanded in addition to your ration card, and the autonomy of local institutions to fix a problem with the ration-card system is compromised as they don’t know what’s wrong and have to report it up the line. Traditional foundational IDs such as voter cards don’t really work for welfare any more but are still accepted for opening bank accounts. As for credit, people depend on relatives, moneylenders, and credit societies; Aadhar tries to universalise this. Recall that Gandhi rejected the pass book in 1905 in South Africa as he understood that it would compromise the family system. Similarly, Aadhar tramples over local ways of organising things such as credit cooperatives.
Marielle Debos works on electoral technologies in Africa, starting with a case study in Chad, and is now studying the electoral biometric market in Africa. The IDEA database records that 23 countries use photos and fingerprints out of the 50 they record, and only 13 use no biometrics. These technologies have been widely criticised (see later speakers), and their fetishization has attracted separate comment. However the technology debate needs to be combined with a wider perspective on democracy assistance.
Alan Gelb has been doing fieldwork on the Krishna district in Andhra Pradesh, the place the World Bank takes delegations. Is the hype justified? They’ve pushed Aadhar hard, tried hard to improve connectivity and to fix accountability for service delivery by making the village revenue officer responsible for technology failure; pushed portability so you can pick up your rations anywhere, backed by better stock management; and given users a voice by mandatory robocalls to check satisfaction. Alan’s mission was to do an independent assessment of beneficiary experience and understand whether the tracking and redress systems actually worked. 390 people thought things better and 155 worse; rations are more accurate and dependable, but on the other hand the authentication service and the point of delivery can be flaky. People who authenticate first try are pretty happy but some need three goes to get a fingerprint and tend to think the system worse. People who couldn’t get food went to the village revenue officer and almost all got food within seven days. Mobility is viewed very favourably even though only about 12% of beneficiaries move around; 96% of dealers have served people from out of their area. The headline benefit is huge cost savings: perhaps 35%. The main losers are dealers who can no longer just keep unclaimed food thanks to the improved reconciliation. (Their margins had to be improved, cutting savings to 33%.) The payback period for the automation is about 2 months. The robocall service monitor is important; response is 15% so each dealer gets about 60 ratings each month, and official behaviour is now significantly constrained, with corruption significantly curbed. In short, this seems to be a frontier in digital governance; you can build a lot of good things on a platform. Some other states collect data but don’t use it in this way. It’s critical to have a real-time feedback-accountability loop, and to have politics that prioritises service delivery.
Jaap van der Straaten studies the economics of elections. He has become pessimistic about elections in Africa after the progress up till 2005. Measurements by political scientists such as Staffan Lindberg and Daniel Kaufman have collected lots of data on the positive effects that democracy has on economic growth in less developed countries. Subsaharan Africa is second-worst on democracy, beating only central Asia; it is last by a mile on birth registration at about 40%, with South Asia second on 60-ish and everywhere else in the 90s. As a result, the cost per capita of running elections in Africa is huge; about 40 times as much as in the rest of the world. Sub-Saharan Africa has spent about $50bn on 300 elections since 2000 and could have saved a third of that if costs had gone up only as much as in South Asia. Jaap believes that African countries need a longer-term approach to population registration, but at present citizens are cynical, and with some justification. Since 1989 the number of countries holding elections has climbed from under half to over 90% but the proportion considered free and fair has sunk from over 90% to about 60%. Democracy is being stress-tested around the world.
Discussion included the difficulty of changing an administration conditioned by history, ethnic rivalries and limited capacity; the difficulties of dealing with diverse and distributed pressures on governments with constrained means; the conflict between trying to achieve quantum leaps using technology and the need to continue building administrative institutions, checks and balances; the underlying assumptions about state behaviour in the democratisation realm, with increasing bad faith by incumbents, just as western states have displayed bad faith over surveillance and privacy; our reliance on leaks, and to a lesser extent lawsuits, to find out what’s actually going on; vendor lock-in and the lack of open-source or other public-good govtech; the need for realistic maturity models for govtech; the South African experience of doing adult registration first, then child registration for benefits, then birth registration; and whether we can fix the registration problem more generally by paying people to register their children.
Amiya Bhatia has been looking at regstration rates in 67 low and middle income countries from 1999–2015, Most did much better with 14 achieving almmost complete registration, but 15 got worse, three much worse. The in-country disparities were urban/rural and rich poor but there were no gender differences. In many countries, registration improved among wealthier families. Amiya concludes that it’s not a last-mile problem as some say but about underling inequality.
Laura Bingham has been studying the role of community engagement. Identity systems tend to reflect wider prejudices and institutional dynamics such as remote government offices and concentration of power in security ministries. Her research framework is ‘jurisgenerative politics’, which means increasing people’s legal and political agency so that can assert claims.
Sanjay Dharwadker has been studying incompatibilities between border crossing and national identification. Countries have multiple systems of registration, identification and entitlement. The boundaries become ever more complex; after passports you have visas, then you have ESTAs, then you have advanced passenger information, then the passenger name record (which card was used to buy the ticket); the increasing role on non-state players makes governance hard. APIs between all these systems create incompatibilities (in the ways names are written down, for example) and make governance hard. Privacy does fail seriously for some: Indians who just want to be known as ‘Rakesh’ so their caste is not disclosed find that electronic systems will not accept mononyms.
Ursula Rao studies the last mile, particularly in India. Digital systems are often tough on the poor, and are not necessarily cheap: why is voting expensive in Africa? Because the supporting institutions we have in Europe often don’t exist there, and digital systems can’t stand alone. Her studies show that if you relax security and make systems more inclusive, they work better. For example, to get an Aadhar registration in India, you have to show proof of birth and proof of address, but the system doesn’t take copies, so you have to bring them again and again when you use the system; all it does is to assign a number to your fingerprint. The Indian welfare system is a pain because people have to prove every five years that they’re below the poverty line – but the benefits are so meagre that the middle class wouldn’t be interested, so why not just cover everyone? The public distribution system only started to work when they allowed relatives to register, as a workaround for people whose fingerprints didn’t work because of wear, burns etc. Even so you abandon the most needy, such as grandmothers whose sons have gone off to work in town. In short, the securitisation of the state is not compatible with inclusiveness.
Discussion included unplanned uses of identity, such as money lenders taking cards as security; this doesn’t work unless at least one critical component is hard to replace, so Indian money lenders now demand your birth certificate as well as your Aadhar card. Authentication failures are also an issue: 90% of those who can’t authenticate are old, and the most vulnerable are old women whose bodies are not machine-readable. If your eyes are bad and your fingers are worn, or too dry, or damaged by henna, then you may be unable to register. If you live in a village then neighbours might care for you, but the unregistered elderly in cities are perhaps the most precarious. As for Africa, very little works; people are starving, there are no jobs, there’s no access to housing, a bureaucracy gets in the way of everything, and the pressing need is for ways of distributing welfare cheaply. This argues for universal social grants based on biometric transfer. But we now fight two sources of new obscurity; technology (particularly biometrics) and the lawyers who always find new ways to mess everything up.
Bronwen Manby studies nationality law and statelessness, including refugees, irregular migrants and citizens who are falsely accused of being such.
Fernando de Medina Rosales works for a Norwegian agency on the obstacles faced by displaced persons to access civil registration and identification, in Columbia, Lebanon and Afghanistan.
Natalie Brinham has been working for the last ten years on Rohingya populations, recording their oral histories and experiences, including of registration processes – both in Myanmar and in states to which they’ve fled, including India, Bangladesh and Saudi Arabia. In Myanmar they are highly visible and controlled, passing through six or more checkpoints between home and market, often requiring different credentials. Systems in India and Saudi share information with Myanmar.
Eves Hayes de Kalaf is doing doctoral research on Dominican citizens’ experience accessing legal identity. In 2013 a constitutional tribunal ruled that because a lady had been erroneously recorded as a citizen, as her parents had not got her an identity card; civil registry officials got engaged in confiscating birth certificates,
Emrys Schoemaker works with Caribou Digital, a consultancy that analyses emerging markets and that started looking at ID some three years ago. Their identities project did a study of Aadhar, and they’re now studying refugees in Jordan, Lebanon and Uganda.
The first panel discussion topic was what’s needed for an enabling environment. This may be hardest for the Rohingya who suffer explicitly racially discriminatory laws; they have to falsely claim to be foreign (Bengali) to get documents which thus carry shame and destroy ethnic identity rather than asserting it. The lived environment has built-in persecution. The Dominican Republic was different as it was under pressure from the World Bank: the country had the worst registration in the Americas. As systems were improved, gaps appeared and officials were under pressure to do something about “el tojo”, the mess. A common theme is the doctrine that citizenship is a matter of state discretion; international institutions fight shy of any criticism. A second was dignity in the context of registration cards: in Lebanon and Jordan, being registered as a refugee means you can’t work, and many people prefer working to handouts (using refugee cash cards is felt to be humiliating). In Uganda the environment is better in theory but connectivity in the northern camps is limited (deliberately, for security reasons) so a lot doesn’t work in practice. The Rohingya resisted electronic registration unless their ethnicity and religion were recorded properly. The UN had thought that non-ethnic cards would cut the risk of discrimination, but the Bangladeshi government was sharing the data with Myanmar anyway and the refugees saw registration as something that “makes you foreign”. Columbia is different again, as Venezuelan refugees see it as a normal, trustworthy state. This raises issues of consent, the third question. Is it essential? The Norwegian view is that aid should be available on the basis of need but that can be hard to implement. In Dominica, anyone without one of the new biometric ID cards is in effect having citizenship taken away. In many refugee contacts, registrants simply don’t know what will be done with their data; UN systems are being networked, NGOs don’t have any real choice, and refugees are afraid that registration data will be given to the regime they’re fleeing (which happens sometimes, as with the Rohingya). There are all sorts of catch-22; Sudanese in Egypt can’t register kids because you need ID, and renewing your passport costs $200 every two years; even if you get a refugee card, you need a marriage certificate and Sudan doesn’t do those, and in any case you need the husband to be there or his death certificate. Then you need a kid to be registered before it can go to school; it’s just impossible. So you maybe register your child as the child of a neighbour. The UN just won’t engage with bureaucratic thickets like this. Just sometimes, the bureaucracy can be empowering; women and children usually flee first, and the UNHCR treats the first adult in each family as the head of household, which empowering for many women, but leads to domestic violence of others.
Group discussion started with the Dominican Republic. Children of Haitian descent often couldn’t get healthcare or go to school; even though they were citizens in theory they couldn’t exercise their rights in practice. Eventually the constitution was changed to exclude them, with Trumpian demonisation. The country did on the other hand welcome Jewish refugees. Something similar happens with Somalis in Kenya; some are citizens, some are refugees. Some register as both, but this strategy is being undermined by digitisation. Myanmar based citizenship from independence on both residency and ethnicity, but removed citizenship by residence in 1982; laws have effectively been racist since then. They have colour-coded ID cards and the Rohingya’s are pink; the issue has come up with the UN again and again but they see it as neocolonialism to interference with such things. In the Anglophone colonies in Africa, rule was indirect (typically by tribe) and systems just didn’t want to know about outsiders; in Uganda officials created a whole lot of fictitious citizens in order to siphon off money. In Britain itself, we’ve had the Windrush scandal where UK citizens have been deported as they could not satisfy the demands of junior officials for unreasonable amounts of documentation. A continuing theme is the lack of interaction between analogue and digital systems; the latter often demand ever greater quantities of bumf, as in the Dominican Republic where officials constantly demand fresh certified copies of birth certificates. Citizenship is often unclear, and governments obstruct, as when the UK destroyed the Windrush landing cards in 2010. If you try to solve such problems by giving people of uncertain citizenship a registration card, it stigmatises them, perhaps for generations (as with the many people in Lebanon with registration cards saying they’re not registered). There’s a lack of political will for ministries and other organisations to share data sensibly between complex systems. Once this is understood, identity as assertion (as promoted by the blockchain startups) is clearly inadequate, as identity in the real world is about recognition. The UK government will never let you stay in Britain because you put something on a blockchain; it’s just not relevant. The real questions are who makes the decision? Can it be appealed? Do you need to be rich enough to go to the high court, or is there something else that works? And does it work for divorces, for adoptions, for unaccompanied children?
John Effah has been studying ID problems in Ghana, whose ethnic groups spill over to neighbouring countries. People can therefore come in and pass as locals, which can cause tension during elections. The government therefore proposed in 2003 a multipurpose biometric ID for the voter register and everything else too. Data were captured manually in the field followed by deduplication which turned out to be harder than expected; people also had to re-register. After the government changed from NPP to NDC in 2009, the NDC one didn’t support the project any more. Further issues arose with ‘ghost names’, non-existing people on the government payroll. Biometrics were captured but verification was manual, by the head of institution, leading to further problems. The overall experience was that biometrics were over-promoted; they were advertised as solving various political and administrative problems but given the implementation they didn’t really. Now the NPP are back in power and want to resurrect the project. The underlying problems are basically political; governments attempting such projects should build consensus and make them national projects rather than partisan ones.
Christian Lund has been studying land registration, most recently in Medan, Sumatra, most of which has been built illegally since the 1990s on plantation land, or on old plantation leases that expired. This is not just slums but hotels, shopping centres, and even Centre Point – which an entrepreneur built on railway land and which has been litigated for years. Entrepreneurs often hired youth gangs to push people off land so they could build on it. In 2003 the Governor did a study and concluded that 5873 ha had been irretrievably been taken and should be formally distributed. This caused a rush to have ‘your’ land registered, with massive corruption and eventually times the amount of land being registered. Medan has now been declared a ‘smart city’.
Cláudio Machado Cavalcanti is studying the history of registration in Latin America. In Brazil, people are very anxious about documents: they’re what you save if your house catches fire or is flooded. Current institutional arrangements go back to the Vargas dictatorship in 1934, with a thicket of records at federal, state and municipal levels whose interrelationship is so complex that poor rural people cannot navigate it. It is mainly controlled by the police, and essentially authoritarian; to get a card, you go to a police station. The state has been trying but failing to reform them for the last 20 years.
Alena Thiel has been studying database interoperability and administrative population data, and compares Ghana, Rwanda and Kenya as a case study. Most recently Ghana ordered a halt to all national ID investments in 2016 to get back to the NPP program (see John Effah’s talk, above); the discourse in Ghana, as in Rwanda and Kenya, is ‘harmonisation’, and the current push is to rely on the birth certificate as the only acceptable foundation document for biometric registration (though a passport will do, as passports need birth certificates). Court cases followed and commissioners of oaths have now been hired to do retrospective birth registration. They currently have 65% registration; dissidents within the administration say that trying to get to 95% is a waste of money as they’re good at using the data they have.
Ornit Shani studies the history of India, and has written a book on How India Became Democratic: Citizenship and the Making of the Universal Franchise. India’s 1947 decision in favour of universal adult suffrage was brave as Partition was tearing the country apart, and a third of the territory consisted of princely states that still had to be integrated; 76% were illiterate and the female half of the population didn’t all recognise themselves as individuals. The state of Travancore persuaded Delhi officials that the government had to take the initiative and go door-to-door to register people while being flexible about details such as language and age (an astrological chart would do instead of a birth certificate). They decided to divorce the voter roll from everything else, including the census, as they had to make everyone on the roll equal in the most hierarchical society in the world. The formal criterion was that voters had to be at least 21 and resident for 180 days, so what about the 18m refugees? Officials looked at the collapse of empires after WW1 and found no precedent, so they decided to register any refugee who declared they would stay for 180 days. There were various attempts to disenfranchise other groups such as dalits, who resisted as they realised that the vote was they key asset. Fraud was fought by dyeing voters’ fingers and nails to prevent double voting. This was all done when there was no law or constitution yet, and in due course parliament wrote the laws so as to justify the voter roll on which its legitimacy was based.
The discussion noted the effects of colonial and pre-colonial history; Indonesia had a lot of small princely states, like India. Some countries have a long history of independent electoral oversight commissions, and there it’s repressive to just roll it all into a national identity card. What are the underlying political processes? Are the forces driving voter registration in Brazil similar to those in the USA? Can we make sense of stories like Ghana’s without a clearer idea of what the parties stand for and the stories of the people who lead them? There is also a big difference between being registered and being able to prove you’re registered, which technology might possibly narrow; but there are still many countries where the rights you get as a result of registration depend in complex ways on your parents or other context.
The final wrap-up discussion continued with a discussion of Brazilian history; resistance to identity registration goes back to the slavery period and the registration drive was driven by senior police officers who were involved not just in torture but in supporting fascism. In India, on the other hand, the electoral commission became an independent pillar of the state; citizenship law came along only two years later and was informed by Partition and by the permit laws that had been passed meantime. The constitution followed still later once registration had been tidied up. As a result, being on the roll wasn’t proof of citizenship. Ownership of land was another big issue. In Ghana, ID is poisoned by politics to the extent that the border with Togo is closed during elections if the NPP’s in power but opened when the NDC are in office; this can make a difference of tens of thousands in one constituency. So the combination of documents needed to register to vote really matters for election outcomes. There’s also an issue with Nigerians getting Ghanaian passports when they shouldn’t. The middle classes in Accra mostly think it’s a waste of money to have seven different biometric ID cards of various vintages. Finally, could problems with land registration in LDCs be mitigated with squatters’ rights formalised into adverse possession rules, as happened in the UK? Sometimes but it tends to depend on whether the other party is powerful or not. Some countries have made less progress than others on the “long march from status to contract”.
Great! Looking forward …
Subhashish Bhadra has handed out £9m in grants for digital identity research from the Omidyar Foundation over the past three years. He started by talking about seat belts, for which a patent was filed by Claghorn in 1885; Ford offered it from the 1950s and it became mandatory in 1965. But only 10–15% of people wore them, and eventually State Farm Insurance sued the government to get them to bring in mandatory seat belt laws. The design upgrade of inertia reel belts helped, but the key was the growing body of research into accidents. Similarly, getting decent ID systems fielded, especially in less developed countries, will take both technology and patient research into what works, in order to guide laws and judgments. Subhashish sees good ID as being governmental (or at least “issued”, such as university ID) while de-facto ID is the data trail. But digital identity and privacy are both important and sensitive topics.
My talk summarised what I say about naming in my security engineering book. Naming people (and things) poses a trilemma: names can be human-comprehensible, unique or distributed; in fact you can choose any two of these. When exploring what might work, Needham’s ten laws of naming may be helpful. The function of names is to facilitate sharing; The naming information may not be in one place, so resolving names brings the general problems of a distributed system; It’s bad to assume that only so many names will be needed; Global names buy you less than you think; Names imply commitments, so keep the scheme flexible enough to cope with administrative changes; Names may double as access tickets, or capabilities; Things are much simpler if an incorrect name is obvious; Consistency is hard, and is often fudged; Don’t get too smart. Phone numbers are much more robust than computer addresses; In general it’s a bad thing to bind early if you can avoid it.
Eddie Higgs was next, talking about the history of physiognomy. Aristotle believed that a beautiful soul would be in a beautiful body with the right symmetries; and physiognomy was accordingly codified by Lavater in the 1770s. Slavery and empire were justified by arguing that large African jaws were animal. Galton overlaid photos of criminals, syphilitics and Jews, and was the father of eugenics; appearance was genetic and predisposed you to particular lifestyles. Echoes of such ideas have come back in the digital age with various research groups talking about significant facial correlates of PF scores, whether in AU facial geometry features. Wang and Kosinski’s work on gay faces brought this to public attention in 2018; their view is that facial masculinity is an effect not just of fetal androgen levels but life history (which has echoes in classical Chinese writing).
Wendy Hunter has been studying the 7% or so of US citizens whose birthright citizenship is undermined by their lack of documentation. There are gaps in birth registration; the liberal state’s laissez-faire approach has side-effects. Birth registration became mandatory only on 1935, it’s a state matter, it’s poorly coordinated, and late registration is hard. If you never went to school, or saw a doctor, or got vaccinated, you can have problems. This is not just about children of illegal immigrants; the largest group is probably elderly African Americans, then Native Americans who thought their tribal cards would work but don’t, then prison leavers who just have no papers. The foster-care system is a mess in most states. The latest is the children of parents trying to live off-grid, some of whom just can’t get passports. Now delayed birth registrants in Texas (common enough given their midwifery system) can’t get passports as a matter of Trump administration policy. Wendy believes that the government must allocate serious professionals to the task of delayed registration rather than leaving it to volunteers and churches; and it needs serious coordination between the federal government and the states.
Amba Kak studies digital colonialism, which is not quite the same as the techlash! A growing number of serious policy documents from India, China and even the EU now tackle big tech practices as a colonising force. Xi Jinping talks of cyber sovereignty; he tells audiences not just in China but in Russia and Brazil that the Western narrative of their openness versus Chinese censorship is simply a cover for creating space for tech companies to predate on China. The European data-protection supervisor, Giovanni Buttarelli, prefers digital colonialism to surveillance capitalism as his opponent when promoting communitarian approaches where data are managed by collectives in the public interest. As India will be data rich before it’s rich in economic terms, this is a critical issue for Indian policy; it initially blocked Facebook’s Free Basics with the free-market argument that the country needed a level playing field, but the rhetoric now is unashamedly mercantilist.
Ian Watson started out by exploring Icelanders’ love of their ID number and has since become a scholar of all the Nordic ID traditions; his students have added English accounts of the Finnish and Norwegian numbers. In all countries, your number is more important than any card or other token. In Iceland you need someone’s ID number to send them money, as it’s embedded in their bank account number; this isn’t the case in Sweden, although people are open about there numbers there too, and in Finland. The Norwegian and Danish systems are slightly more closed.
Discussion started on the roots of Scandinavian ID systems in national statistical offices, and their takeover by tax offices; the various arguments for data localisation, sometimes about national sovereignty, at other times about business convenience, and also about their intersection in protecting local business. Many LDCs also lack decent data protection. At the core though is incentives: people register in Scandinavia because of the huge welfare state, and in South Africa because of the child support grant, but in many LDCs there’s no real reason to bother. Should a liberal state allow people to opt out of registering their children, or compel registration? There is also the issue of policy priority: is the main goal to ensure that all citizens have the vote, or to ensure that everyone has the right foundational documents? As Wendy’s talk brought out, you end up with quite different outcomes, and the choice is fundamentally a policy one. And finally, human beings have an insatiable appetite for divination.
Edwin Abuya gave us a survey of issues around ID and privacy in East Africa. In a Kenyan ID case, plaintiffs sued the attorney general and director of national registration and Edwin asked us not to blog the details as it’s in progress. In Burundi, journalists sued the AG, in a case about the protection of their sources. In Uganda, a farmer sued after her bank got hold of her ID photo and used it on an advertising billboard. One of the cases has already got to the East African Court of Justice. Practising lawyers start from the desired remedy and then try to build the case to get there; if they find missing steps in the ladder they make human-rights arguments that statute law is deficient.
Gautam Bhatia has been involved as a lawyer in two of the big Aadhar cases. In the Right to Privacy judgment the court ruled that the principle of proportionality should guide the trade-off between privacy and state goals; the Aadhar judgment applied a four-step proportionality test and ruled out the compulsory application of Aadhar in a number of applications including school exams and bank accounts. On the legislative track, the Srikrishna Data Protection Bill has been published but has attracted criticism including for the lack of a right to an explanation and the weakness of enforcement against the state. Gautam is also concerned about the independence of the enforcement body.
Marianne Henriksen described the Norwegian system. In 1964, municipal registration was centralised; there are now 5.3m residents and 2.2m “D-numbers” including workers, immigrants refugees staying less than six months. It supports a huge range of services. Their “identity triangle” links the ID number with biometrics captured when passports are issued. They have been trying to issue “Norwegian identity” cards for fourteen years now but there is resistance, even among refugees who might be significant beneficiaries. She would prefer it if service providers would insist on biometric ID but some significant services, such as the tax authorities, won’t. (The tax authorities also resisted the failed ‘Verify’ programme in the UK; perhaps, to see the ground truth, we need compare policy for systems the government cares about, like tax, with the rest.)
Jonathan Klaaren is a South African lawyer who has followed the litigation against Net1 in the Constitutional Court. Privacy arguments there had no traction, as the information was distributed in the cards they issued for social grants; yet this firm had grabbed hold of grant distribution for 16-17m South Africans, and in effect the CC nationalised it after the Competition Commission passed and the Information Commission had no staff or capacity. A supervised remedial order moved the grant stream back into state control. Courts can entertain broader public-interest arguments than classical antitrust market regulation. This isn’t a north-south issue but where privacy and competition overlap there can be regulatory arbitrage. Kenya has just worked through modernising its ID and privacy law; South Africa is at the white paper stage of a redrafting.
Ranjit Singh has been doing a PhD on Aadhar since 2015. Aadhar started out as a general program to identify people, without a stated bureaucratic purpose; the Supreme Court has since defined its purposes to some extent by limiting them. The authority that runs it deliberately collects minimal information about each resident that it enrols; that establishes uniqueness. Seeding is a secondary process whereby the Aadhar number is added to other public and private database records in order to clean them. The third step is authentication: the authority records only the time, the requesting entity and the response, not the location or the purpose. What’s more, nic.in requests identity checks for a lot of government departments, shielding the metadata. This privacy by design led the Supreme Court to conclude that the architecture, together with the Aadhar Act, do not tend to create a surveillance state. Yet although Aadhar is not an archive, it is becoming a common index to multiple archives; he believes the court was wrong to miss this. Ranjit argues that Aadhar should not be blamed when, for example, a poor person suddenly fails to get welfare grains from which she got welfare grain for years; there the blame should fall on the specific government application systems that now call Aadhar.
Edgar Whitley introduced the afternoon session on identity for credit. Can credit scoring systems substitute for national ID, and how do you deal with people whose credit files are thin? In any case, how do you link credit histories dependably to individuals, especially in the absence of good ID? Is the goal to provide national identity systems, or to provide identity systems for nationals? In a global world there’s a big question about your nationals living abroad, and foreigners resident in your country. And in 1975 Paul Armer noted that if you wanted to build an unobtrusive system for surveillance, you couldn’t do much better than an electronic funds transfer system. Is that where we’ve got?
Johann Bezuidenhoudt works on financial inclusion. Cash was inclusive and pervasive, but electronic services exclude many people, include those with no formal national ID. What alternative approaches can be accepted and included? You’d need the means of re-establishing a functional authenticator, such as by getting a new SIM card if it’s stolen; and authentication good enough to block imposter theft. Practically, you need value and volume limits, back-end analytics to mitigate smurfing, and profiles of location and use. Authentication is probably a SIM and re-issue may be a voice biometric. A good financial inclusion product could be bought in a supermarket from cash; “Smart” was fielded in the Phillippines in 2000, but later closed. In Nigeria you can still self-assert to open a mobile money account. Could use de-facto, chained, associative, situational and intrinsic (biometric) ID: let’s tighten the definitions. The imperative is to break the KYC status quo. M-pesa would never have got to where it is if KYC had been as tight ten years ago!
Roland Claussen has been studying the predictive power of credit scoring. The effectiveness of institutions depends on their social embeddedness – the degree to which people trust it and accept their role in its ecosystem. The other factor in his research is the specificity of contextual data, which greatly boosts the power to predict future arrears. Big data aren’t needed. You can use existing socially-embedded institutions such as the 25k existing microfinance institutions all over Africa. His company is providing software to try to standardise all this in one app.
Gabriel Deval worries about the thinness of credit markets in Africa; Zambia has only 500 mortgages and 2,199 business loans were made in 2017. There is so little credit flowing through banks into the private sector that it makes no difference, and the central bank statistics are generally of little value. There are issues with personal and company registers, asset registration, credit bureaux and contract enforcement. Tanzania has over 18% loans to MSMEs defaulting and over 40% to agribusiness; it has only 3,600 mortgages (which are hard when you can’t identify people or trust the land registry). Business practices like invoicing and invoice discounting are also tough. The basic problems are legal and political; without fixing them, tech fixes are pointless – unless perhaps we allow some “Wild west behaviour” to get the car moving, and worry about the seat belts later. Digital finance might hope to make a difference by getting its hooks into distribution and collection.
Bulelani Jili has been studying the Chinese social credit system – an attempt to create a credit market with socialist/Chinese characteristics in that it would detect political expression as well as spending history. It does seem to be helping with access to credit as well as helping deal with governance in China. Collaboration between state entities and private actors has helped create punishment mechanisms but the punishment comes mostly from violating industry norms and codes. In the case of Ethiopia, the government is also trying to instrumentalise data as a tool of social governance, and privacy laws are also weak. However Bulelani’s concern is not so much a social credit system as the bulk surveillance tools sold by firms in Italy and Germany – together with the training and doctrine around their use, which comes from the USA.
Discussion ranged from the extent to which social infrastructure should be private or run by the state; why nobody seems willing to push back on the FINCEN requirement that everyone show ID and address proof when opening a bank account; the weakness of the underlying authentication of most mobile money systems; the South African experience that fixing property registration, courts etc gets stuff to work better than messing about with fintech; the fact that many big-data startups like Wonga went bust or nearly so, while successful firms like Safaricom use simple credit scoring; the huge amount of credit used in sixteenth-century England when there was a shortage of specie and there were a million court cases for four million people; the use of government-mandated systems, such as fingerprint readers in South Africa, being used by crooks to loot people’s bank accounts; the real problems facing firms in Africa; and whether the west’s history of growing building societies might be of relevance, or whether it just invites a replay of Victorian banking scandals or the Mafia.
Bidisha Chaudhuri and Syed Mohammed Faisal have been studying Aadhar takeup. It’s available as a paper form you can laminate, and also as an Aadhar card and as online biometric checks: but in practice the paper form is demanded and sometimes even the card is not enough. Your functional ID, namely your ration card, is demanded too. In effect, Aadhar has become something demanded in addition to your ration card, and the autonomy of local institutions to fix a problem with the ration-card system is compromised as they don’t know what’s wrong and have to report it up the line. Traditional foundational IDs such as voter cards don’t really work for welfare any more but are still accepted for opening bank accounts. As for credit, people depend on relatives, moneylenders, and credit societies; Aadhar tries to universalise this. Recall that Gandhi rejected the pass book in 1905 in South Africa as he understood that it would compromise the family system. Similarly, Aadhar tramples over local ways of organising things such as credit cooperatives.
Marielle Debos works on electoral technologies in Africa, starting with a case study in Chad, and is now studying the electoral biometric market in Africa. The IDEA database records that 23 countries use photos and fingerprints out of the 50 they record, and only 13 use no biometrics. These technologies have been widely criticised (see later speakers), and their fetishization has attracted separate comment. However the technology debate needs to be combined with a wider perspective on democracy assistance.
Alan Gelb has been doing fieldwork on the Krishna district in Andhra Pradesh, the place the World Bank takes delegations. Is the hype justified? They’ve pushed Aadhar hard, tried hard to improve connectivity and to fix accountability for service delivery by making the village revenue officer responsible for technology failure; pushed portability so you can pick up your rations anywhere, backed by better stock management; and given users a voice by mandatory robocalls to check satisfaction. Alan’s mission was to do an independent assessment of beneficiary experience and understand whether the tracking and redress systems actually worked. 390 people thought things better and 155 worse; rations are more accurate and dependable, but on the other hand the authentication service and the point of delivery can be flaky. People who authenticate first try are pretty happy but some need three goes to get a fingerprint and tend to think the system worse. People who couldn’t get food went to the village revenue officer and almost all got food within seven days. Mobility is viewed very favourably even though only about 12% of beneficiaries move around; 96% of dealers have served people from out of their area. The headline benefit is huge cost savings: perhaps 35%. The main losers are dealers who can no longer just keep unclaimed food thanks to the improved reconciliation. (Their margins had to be improved, cutting savings to 33%.) The payback period for the automation is about 2 months. The robocall service monitor is important; response is 15% so each dealer gets about 60 ratings each month, and official behaviour is now significantly constrained, with corruption significantly curbed. In short, this seems to be a frontier in digital governance; you can build a lot of good things on a platform. Some other states collect data but don’t use it in this way. It’s critical to have a real-time feedback-accountability loop, and to have politics that prioritises service delivery.
Jaap van der Straaten studies the economics of elections. He has become pessimistic about elections in Africa after the progress up till 2005. Measurements by political scientists such as Staffan Lindberg and Daniel Kaufman have collected lots of data on the positive effects that democracy has on economic growth in less developed countries. Subsaharan Africa is second-worst on democracy, beating only central Asia; it is last by a mile on birth registration at about 40%, with South Asia second on 60-ish and everywhere else in the 90s. As a result, the cost per capita of running elections in Africa is huge; about 40 times as much as in the rest of the world. Sub-Saharan Africa has spent about $50bn on 300 elections since 2000 and could have saved a third of that if costs had gone up only as much as in South Asia. Jaap believes that African countries need a longer-term approach to population registration, but at present citizens are cynical, and with some justification. Since 1989 the number of countries holding elections has climbed from under half to over 90% but the proportion considered free and fair has sunk from over 90% to about 60%. Democracy is being stress-tested around the world.
Discussion included the difficulty of changing an administration conditioned by history, ethnic rivalries and limited capacity; the difficulties of dealing with diverse and distributed pressures on governments with constrained means; the conflict between trying to achieve quantum leaps using technology and the need to continue building administrative institutions, checks and balances; the underlying assumptions about state behaviour in the democratisation realm, with increasing bad faith by incumbents, just as western states have displayed bad faith over surveillance and privacy; our reliance on leaks, and to a lesser extent lawsuits, to find out what’s actually going on; vendor lock-in and the lack of open-source or other public-good govtech; the need for realistic maturity models for govtech; the South African experience of doing adult registration first, then child registration for benefits, then birth registration; and whether we can fix the registration problem more generally by paying people to register their children.
Amiya Bhatia has been looking at regstration rates in 67 low and middle income countries from 1999–2015, Most did much better with 14 achieving almmost complete registration, but 15 got worse, three much worse. The in-country disparities were urban/rural and rich poor but there were no gender differences. In many countries, registration improved among wealthier families. Amiya concludes that it’s not a last-mile problem as some say but about underling inequality.
Laura Bingham has been studying the role of community engagement. Identity systems tend to reflect wider prejudices and institutional dynamics such as remote government offices and concentration of power in security ministries. Her research framework is ‘jurisgenerative politics’, which means increasing people’s legal and political agency so that can assert claims.
Sanjay Dharwadker has been studying incompatibilities between border crossing and national identification. Countries have multiple systems of registration, identification and entitlement. The boundaries become ever more complex; after passports you have visas, then you have ESTAs, then you have advanced passenger information, then the passenger name record (which card was used to buy the ticket); the increasing role on non-state players makes governance hard. APIs between all these systems create incompatibilities (in the ways names are written down, for example) and make governance hard. Privacy does fail seriously for some: Indians who just want to be known as ‘Rakesh’ so their caste is not disclosed find that electronic systems will not accept mononyms.
Ursula Rao studies the last mile, particularly in India. Digital systems are often tough on the poor, and are not necessarily cheap: why is voting expensive in Africa? Because the supporting institutions we have in Europe often don’t exist there, and digital systems can’t stand alone. Her studies show that if you relax security and make systems more inclusive, they work better. For example, to get an Aadhar registration in India, you have to show proof of birth and proof of address, but the system doesn’t take copies, so you have to bring them again and again when you use the system; all it does is to assign a number to your fingerprint. The Indian welfare system is a pain because people have to prove every five years that they’re below the poverty line – but the benefits are so meagre that the middle class wouldn’t be interested, so why not just cover everyone? The public distribution system only started to work when they allowed relatives to register, as a workaround for people whose fingerprints didn’t work because of wear, burns etc. Even so you abandon the most needy, such as grandmothers whose sons have gone off to work in town. In short, the securitisation of the state is not compatible with inclusiveness.
Discussion included unplanned uses of identity, such as money lenders taking cards as security; this doesn’t work unless at least one critical component is hard to replace, so Indian money lenders now demand your birth certificate as well as your Aadhar card. Authentication failures are also an issue: 90% of those who can’t authenticate are old, and the most vulnerable are old women whose bodies are not machine-readable. If your eyes are bad and your fingers are worn, or too dry, or damaged by henna, then you may be unable to register. If you live in a village then neighbours might care for you, but the unregistered elderly in cities are perhaps the most precarious. As for Africa, very little works; people are starving, there are no jobs, there’s no access to housing, a bureaucracy gets in the way of everything, and the pressing need is for ways of distributing welfare cheaply. This argues for universal social grants based on biometric transfer. But we now fight two sources of new obscurity; technology (particularly biometrics) and the lawyers who always find new ways to mess everything up.
Bronwen Manby studies nationality law and statelessness, including refugees, irregular migrants and citizens who are falsely accused of being such.
Fernando de Medina Rosales works for a Norwegian agency on the obstacles faced by displaced persons to access civil registration and identification, in Columbia, Lebanon and Afghanistan.
Natalie Brinham has been working for the last ten years on Rohingya populations, recording their oral histories and experiences, including of registration processes – both in Myanmar and in states to which they’ve fled, including India, Bangladesh and Saudi Arabia. In Myanmar they are highly visible and controlled, passing through six or more checkpoints between home and market, often requiring different credentials. Systems in India and Saudi share information with Myanmar.
Eves Hayes de Kalaf is doing doctoral research on Dominican citizens’ experience accessing legal identity. In 2013 a constitutional tribunal ruled that because a lady had been erroneously recorded as a citizen, as her parents had not got her an identity card; civil registry officials got engaged in confiscating birth certificates,
Emrys Schoemaker works with Caribou Digital, a consultancy that analyses emerging markets and that started looking at ID some three years ago. Their identities project did a study of Aadhar, and they’re now studying refugees in Jordan, Lebanon and Uganda.
The first panel discussion topic was what’s needed for an enabling environment. This may be hardest for the Rohingya who suffer explicitly racially discriminatory laws; they have to falsely claim to be foreign (Bengali) to get documents which thus carry shame and destroy ethnic identity rather than asserting it. The lived environment has built-in persecution. The Dominican Republic was different as it was under pressure from the World Bank: the country had the worst registration in the Americas. As systems were improved, gaps appeared and officials were under pressure to do something about “el tojo”, the mess. A common theme is the doctrine that citizenship is a matter of state discretion; international institutions fight shy of any criticism. A second was dignity in the context of registration cards: in Lebanon and Jordan, being registered as a refugee means you can’t work, and many people prefer working to handouts (using refugee cash cards is felt to be humiliating). In Uganda the environment is better in theory but connectivity in the northern camps is limited (deliberately, for security reasons) so a lot doesn’t work in practice. The Rohingya resisted electronic registration unless their ethnicity and religion were recorded properly. The UN had thought that non-ethnic cards would cut the risk of discrimination, but the Bangladeshi government was sharing the data with Myanmar anyway and the refugees saw registration as something that “makes you foreign”. Columbia is different again, as Venezuelan refugees see it as a normal, trustworthy state. This raises issues of consent, the third question. Is it essential? The Norwegian view is that aid should be available on the basis of need but that can be hard to implement. In Dominica, anyone without one of the new biometric ID cards is in effect having citizenship taken away. In many refugee contacts, registrants simply don’t know what will be done with their data; UN systems are being networked, NGOs don’t have any real choice, and refugees are afraid that registration data will be given to the regime they’re fleeing (which happens sometimes, as with the Rohingya). There are all sorts of catch-22; Sudanese in Egypt can’t register kids because you need ID, and renewing your passport costs $200 every two years; even if you get a refugee card, you need a marriage certificate and Sudan doesn’t do those, and in any case you need the husband to be there or his death certificate. Then you need a kid to be registered before it can go to school; it’s just impossible. So you maybe register your child as the child of a neighbour. The UN just won’t engage with bureaucratic thickets like this. Just sometimes, the bureaucracy can be empowering; women and children usually flee first, and the UNHCR treats the first adult in each family as the head of household, which empowering for many women, but leads to domestic violence of others.
Group discussion started with the Dominican Republic. Children of Haitian descent often couldn’t get healthcare or go to school; even though they were citizens in theory they couldn’t exercise their rights in practice. Eventually the constitution was changed to exclude them, with Trumpian demonisation. The country did on the other hand welcome Jewish refugees. Something similar happens with Somalis in Kenya; some are citizens, some are refugees. Some register as both, but this strategy is being undermined by digitisation. Myanmar based citizenship from independence on both residency and ethnicity, but removed citizenship by residence in 1982; laws have effectively been racist since then. They have colour-coded ID cards and the Rohingya’s are pink; the issue has come up with the UN again and again but they see it as neocolonialism to interference with such things. In the Anglophone colonies in Africa, rule was indirect (typically by tribe) and systems just didn’t want to know about outsiders; in Uganda officials created a whole lot of fictitious citizens in order to siphon off money. In Britain itself, we’ve had the Windrush scandal where UK citizens have been deported as they could not satisfy the demands of junior officials for unreasonable amounts of documentation. A continuing theme is the lack of interaction between analogue and digital systems; the latter often demand ever greater quantities of bumf, as in the Dominican Republic where officials constantly demand fresh certified copies of birth certificates. Citizenship is often unclear, and governments obstruct, as when the UK destroyed the Windrush landing cards in 2010. If you try to solve such problems by giving people of uncertain citizenship a registration card, it stigmatises them, perhaps for generations (as with the many people in Lebanon with registration cards saying they’re not registered). There’s a lack of political will for ministries and other organisations to share data sensibly between complex systems. Once this is understood, identity as assertion (as promoted by the blockchain startups) is clearly inadequate, as identity in the real world is about recognition. The UK government will never let you stay in Britain because you put something on a blockchain; it’s just not relevant. The real questions are who makes the decision? Can it be appealed? Do you need to be rich enough to go to the high court, or is there something else that works? And does it work for divorces, for adoptions, for unaccompanied children?
John Effah has been studying ID problems in Ghana, whose ethnic groups spill over to neighbouring countries. People can therefore come in and pass as locals, which can cause tension during elections. The government therefore proposed in 2003 a multipurpose biometric ID for the voter register and everything else too. Data were captured manually in the field followed by deduplication which turned out to be harder than expected; people also had to re-register. After the government changed from NPP to NDC in 2009, the NDC one didn’t support the project any more. Further issues arose with ‘ghost names’, non-existing people on the government payroll. Biometrics were captured but verification was manual, by the head of institution, leading to further problems. The overall experience was that biometrics were over-promoted; they were advertised as solving various political and administrative problems but given the implementation they didn’t really. Now the NPP are back in power and want to resurrect the project. The underlying problems are basically political; governments attempting such projects should build consensus and make them national projects rather than partisan ones.
Christian Lund has been studying land registration, most recently in Medan, Sumatra, most of which has been built illegally since the 1990s on plantation land, or on old plantation leases that expired. This is not just slums but hotels, shopping centres, and even Centre Point – which an entrepreneur built on railway land and which has been litigated for years. Entrepreneurs often hired youth gangs to push people off land so they could build on it. In 2003 the Governor did a study and concluded that 5873 ha had been irretrievably been taken and should be formally distributed. This caused a rush to have ‘your’ land registered, with massive corruption and eventually times the amount of land being registered. Medan has now been declared a ‘smart city’.
Cláudio Machado Cavalcanti is studying the history of registration in Latin America. In Brazil, people are very anxious about documents: they’re what you save if your house catches fire or is flooded. Current institutional arrangements go back to the Vargas dictatorship in 1934, with a thicket of records at federal, state and municipal levels whose interrelationship is so complex that poor rural people cannot navigate it. It is mainly controlled by the police, and essentially authoritarian; to get a card, you go to a police station. The state has been trying but failing to reform them for the last 20 years.
Alena Thiel has been studying database interoperability and administrative population data, and compares Ghana, Rwanda and Kenya as a case study. Most recently Ghana ordered a halt to all national ID investments in 2016 to get back to the NPP program (see John Effah’s talk, above); the discourse in Ghana, as in Rwanda and Kenya, is ‘harmonisation’, and the current push is to rely on the birth certificate as the only acceptable foundation document for biometric registration (though a passport will do, as passports need birth certificates). Court cases followed and commissioners of oaths have now been hired to do retrospective birth registration. They currently have 65% registration; dissidents within the administration say that trying to get to 95% is a waste of money as they’re good at using the data they have.
Ornit Shani studies the history of India, and has written a book on How India Became Democratic: Citizenship and the Making of the Universal Franchise. India’s 1947 decision in favour of universal adult suffrage was brave as Partition was tearing the country apart, and a third of the territory consisted of princely states that still had to be integrated; 76% were illiterate and the female half of the population didn’t all recognise themselves as individuals. The state of Travancore persuaded Delhi officials that the government had to take the initiative and go door-to-door to register people while being flexible about details such as language and age (an astrological chart would do instead of a birth certificate). They decided to divorce the voter roll from everything else, including the census, as they had to make everyone on the roll equal in the most hierarchical society in the world. The formal criterion was that voters had to be at least 21 and resident for 180 days, so what about the 18m refugees? Officials looked at the collapse of empires after WW1 and found no precedent, so they decided to register any refugee who declared they would stay for 180 days. There were various attempts to disenfranchise other groups such as dalits, who resisted as they realised that the vote was they key asset. Fraud was fought by dyeing voters’ fingers and nails to prevent double voting. This was all done when there was no law or constitution yet, and in due course parliament wrote the laws so as to justify the voter roll on which its legitimacy was based.
The discussion noted the effects of colonial and pre-colonial history; Indonesia had a lot of small princely states, like India. Some countries have a long history of independent electoral oversight commissions, and there it’s repressive to just roll it all into a national identity card. What are the underlying political processes? Are the forces driving voter registration in Brazil similar to those in the USA? Can we make sense of stories like Ghana’s without a clearer idea of what the parties stand for and the stories of the people who lead them? There is also a big difference between being registered and being able to prove you’re registered, which technology might possibly narrow; but there are still many countries where the rights you get as a result of registration depend in complex ways on your parents or other context.
The final wrap-up discussion continued with a discussion of Brazilian history; resistance to identity registration goes back to the slavery period and the registration drive was driven by senior police officers who were involved not just in torture but in supporting fascism. In India, on the other hand, the electoral commission became an independent pillar of the state; citizenship law came along only two years later and was informed by Partition and by the permit laws that had been passed meantime. The constitution followed still later once registration had been tidied up. As a result, being on the roll wasn’t proof of citizenship. Ownership of land was another big issue. In Ghana, ID is poisoned by politics to the extent that the border with Togo is closed during elections if the NPP’s in power but opened when the NDC are in office; this can make a difference of tens of thousands in one constituency. So the combination of documents needed to register to vote really matters for election outcomes. There’s also an issue with Nigerians getting Ghanaian passports when they shouldn’t. The middle classes in Accra mostly think it’s a waste of money to have seven different biometric ID cards of various vintages. Finally, could problems with land registration in LDCs be mitigated with squatters’ rights formalised into adverse possession rules, as happened in the UK? Sometimes but it tends to depend on whether the other party is powerful or not. Some countries have made less progress than others on the “long march from status to contract”.