He and his wife have already been making significant contributions on the NGO side... but to do this full-time - kudos to him!
I have been reading also about the various "obstacles" that he is likely to face, how it's all going to be a failure and hearing from cynics about how Infosys will make money, and how this will never work out.
I say - put a good man with drive to do something - something good will come out of it - inspite of all the obstacles. Whatever Unique ID we get from the quirkily named Unique Identification Authority of India (UIDAI) will be far better with Mr. Nilekani heading it...
Having said that - obstacles are there and duplication, errors, misuse, etc are all big ones.
My suggestion to the Chairman is that first attack at the base so that no new data gets generated wrongly. In other words - at hospitals at the time of birth - make it mandatory that every child when born gets a Unique ID - like a birth certificate - like a Social Security Number in the US.
Next attack the other end - people dying need a death certificate - make sure each such event gets logged into the UID - either to someone already in the system or by creation of a new record in the system.
This will at least get the new data correct - and then begins the onerous task of getting the existing population into the system.
For existing data, a good way to quickly get off the ground would be to provide a self-service portal where people with existing ID (like a passport) can logon and self-register, provide proof-of-ID and submit an application. This will make it much easier than to do a data collection effort. The applications can be scrutinized against the other ID database (passport) and if okay - can be integrated into the UID.
Of course that would take care of more urban than rural population - the primary political target for the UID. But there will be ways - for example take the Yashashwini database in Karnataka and you will get a good bunch of the population into the system with relatively "correct" info!