The dark side of the ID database

Mr Nilekani has taken on the responsibility over the most ambitious government programme of national identification in recent years.

Having a single nationally accepted identification and identity number, it is generally thought, will enable the government to not only better manage internal security, but also enable direct contact between the state and the citizen.

This in turn can help better target benefits to the citizen without the need for a deep hierarchy -- the associated leakages should also then naturally reduce dramatically. It is difficult to argue against any of these presumptions.

In theory, if all works well, direct contact between citizen and state can introduce much efficiency in the implementation of a whole range of government programmes as well as enable better allocation of limited resources.

And the national identification database that exhaustively covers the whole population is perhaps the only way of ensuring a direct two-way interaction between the state and the citizen; as the unique ID is necessary for the state and its agencies to correctly identify the unique individual on the other end.

But all of this is in theory, and the large potential benefits rest on the technical and administrative mechanisms that go into building and maintaining such a large and potent database. Mr Nilekani and his team are about to come up against a range of hurdles.

Contrary to what some may think, various agencies of the state will continue to use their own database for their activities; the only change will be that each record in the many different databases will have the national ID number added to it.

In other words, there will continue to be a passport database with the external affairs ministry, a PAN database with the finance ministry, a terrorist database with the home ministry, a BPL database with the state governments, the election commission's voters database, and so on. Each of these databases will eventually have one more piece of information -- the national ID number. (Mr Nilekani is too smart to try to integrate all of them under one roof, and there will be many legal constraints in doing so.)

Therein lies the first of Mr Nilekani's problems. The clerks who will eventually assign the national ID number to the various databases will not be under his control.

True, they may be trained for a week or two, and manuals will be written for their use, but they will be the same people with the same skill sets who make a mess of names and addresses in all government-citizen engagement. 

Someone less careful than Mr Nilekani might use the time-tested government approach -- shoot first and aim later. In other words, make the database first, get the accolades for the timely completion of a difficult task, and leave the matching mess for others to sort out.   

Only time will tell how this problem will be addressed by Mr Nilekani and his team. But one likely solution will be to build the national ID database on the back of the various pre-existing databases. This will reduce, though not eliminate, the role of the clerk. But this will face other problems that will need to be addressed.

Almost all government databases have names, parent's name, address as the basic information. And this information will be used to initially match the records in different databases.

However, many neighbours have same or similar names, spellings differ across databases for the same person, addresses change with time or are not precise enough, and the language and scripts also differ. Mr Nilekani will try to use his network in the science and engineering establishment to develop automated indexation and matching technologies.

He will try out transliteration technologies to address the multiple languages and scripts problem.

The problem is that not only is the underlying data flawed, even the technologies have some error rates. Given the large numbers involved, even a 5 per cent error rate will translate into 50 million records being matched incorrectly.

To address this and associated problems, each citizen will need to be contacted with a form, and asked to fill out all the information he has, including the various identifiers such as PAN, Passport No. and so on.

This form will then be 'fed' into the national database. The problem is, here as well, there are significant errors. Electronic scanning and matching technologies are not 100 per cent error-free. The only way to cross-check this would be to contact the citizen again and get his confirmation on the information.

If the voter id cards are any indicator, many readers will vouch for the magnitude of this problem. Innumerous instances of ids with wrong names, addresses and photographs have been observed. That is how Sania Mirza's photograph found its way on a BPL card.

Technologies such as fingerprinting, biometrics, etc only have an impact once the basic problem is fixed. But it is not merely matching and ID technologies that are the issue. Without proper love and care an ID card scheme can be highly biased against the underprivileged.

For it is here that low literacy levels, lack of proof of domicile, missing certifications, rapidly changing or missing addresses, inability to maintain papers and cards in a safe place, all are most severe. It is also this segment that is least able to fight the power of the lower level government functionary in extracting rents.

It is a known fact that most district commissioners have to maintain a list of "VIPs". They also have to ensure that at the time of elections these names are in the electoral roll so that there are no uproars. Attention at this level is not provided nor can be expected to be provided to the common man.

Though much of this will be beyond Mr Nilekani's specific ambit, he should ideally ensure that the government functionary at the local level does not have veto power over a citizen's entry into the database or ID card assignment.

One way this can be achieved is through the community's active participation. (But not merely routed through a Panchayat functionary -- the death certificate is one such example, which is taken out by the administration but routed to the recipients through the Panchayat. Complaints of corruption in this process are common.)

There are a range of solutions to such problems, but whatever be the solution finally implemented, it will need to be based on an underlying principle that the individual citizen is more important than the database.

The current set of people who are involved in database creation, maintenance and certification neither have the skill-sets nor an appreciation of the great difficulties they cause when they key in incorrect or imprecise information. And many also have the desire and wherewithal to extract rents from citizens.

What is more, they are spread across ministries and manifold agencies of the state, each reporting to a different master. And they will need to change the way they function. Mr Nilekani will need all the help that he can get from the Cabinet, PM and his wellwishers.

Peeyush Bajpai and Laveesh Bhandari are with Indicus Analytics.