Aadhaar

Nandan Nilekani, co-founder and ex-CEO of Infosys, was one of the guest speakers at a tech symposium at my office earlier today. The topic of his talk was not technology, rather, it was about the Aadhaar card initiative that he drives.

He started with the reason behind the need for such a card: unique proof of identity that can’t be duplicated. He pointed if you have such a non-duplicatable ID, a lot of systems can be built on top of that. So no, this card isn’t just about getting subsidies on that gas cylinder or for use as a KYC proof. A lot more possibilities exist, but more on that later.

So why not build on top of an existing ID system? Like the passport or PAN card?
1)      Those can (and have been) forged. If you try and build a database where some entries are already suspect, nobody knows which part of it to trust!
2)     Most people have nothing to prove their identity anyway, not even birth certificates. Hence biometrics were thought to be the only way out.
3)     Most existing ID systems weren’t designed for either a billion entries or to handle a million applications a day. To do something on this scale requires building a system from the ground up.

Next he got into how the system has been built. Turns out it has been built on cheap hardware (is there another kind?) and open-source (aka free) software. The only part that isn’t free or low cost was the biometrics parts. And even there Nilekani wanted 3 different suppliers to have leverage while negotiating and to avoid problems if one of them went out of business. And, if needed in the future, all of this is scalable. I guess that’s what you get when you use Internet related tools.

Then there’s the part about being able to track the agents who record the data (like time to process, corrections made, reattempts etc). The operators can then either be re-trained or dropped from the system based on performance. Well ok, at least the possibility exists.

So what else can this data be used for? Well, Nilekani has plans to provide API’s (that’s programming talk for providing ways for others to access your system) to let others access the data and to build their own applications. Does that sound like the app store for your smartphone or what? Of course, there are security aspects to be considered, but if sharing data electronically was one of your intentions all along, you can design for it. As Nilekani pointed out, even Western systems don’t have these abilities, but that’s one of the advantages we have when we build a system using today’s capabilities instead of yesteryear’s.

Are you uncomfortable with the government having so much data about you? Nilekani dismisses it saying Facebook and Google have more info on you! And your cellphone operator knows where you are all the time! A bit more seriously, he pointed out that the info in the system is minimal and can’t be used for surveillance anyway.

Now for the million billion dollar question: how much does it cost? I guess Nilekani, being an IT guy, thinks in dollars! He said the system cost around $2 billion; and costs around $2 to process each application. The PDS system transfers around $60 billion annually; so even a 5% improvement in efficiencies and fraud reduction would mean the entire investment has been recovered in 1 year! I’d never heard anything to do with government talk about the money involved, let alone ROI (return on investment)!

So why did Nilekani do it? I loved his answer: because he was given a chance. Not some grand “I always cared about poor people” or “I always wanted to do something for my country” reason.

Most of the points about this system (recovering the money spent, thinking ahead about how to expand the use of this data, mechanisms to detect the performance of operators) are so not what we associate with anything government. But hey, put a business guy who made his billions by working with the West, in charge of things, and it becomes possible, even in India.

Comments

Popular posts from this blog

Student of the Year

The Retort of the "Luxury Person"

Animal Senses #7: Touch and Remote Touch