On Tuesday 02 February 2010 18:21:16 Debayan Banerjee wrote:
> On 2 February 2010 16:40, jtd <[email protected]> wrote:
> > On Tuesday 02 February 2010 15:44:47 Debayan Banerjee wrote:
> >> http://uidai.gov.in/tenders/RFP_ASDMSA_Volume_II.pdf
> >
> > Fine document. With all the enrolment holes of existing id
> > schemes. On the authentication side I wonder if they have tried
> > fingerprinting after 8 hrs in rain / field / shrimp farm.
>
> Well they can wipe their hands before giving a fingerprint, right?

No you cant. Once the fingers are soaked in water for a few hours, 
They take a few hours to regain the original elasticity. So the poor 
sods for whom this billion rupee farce is being enacted are the ones 
who are the ones who will NOT be authenticated most of the time and 
not recieve whatever the government was doling out.

Children, the one area where all subsidies are  justified, cannot be 
biometrically authenticated (as correctly pointed out in this doc). 
So it is wide open to fraud - not that others in the populace are any 
better though. It is utterly trivial to spoof fingerprints.

In another report that this body has brought out (it is so unbelivably 
shoddy, a TY college project would be much better), to cover the 
actual finger print collection process, UIDA hopes to get good finger 
prints with proper training. And since one finger print will result 
in a high FAR, they will collect all ten finger prints. Never mind 
that it provides 10 times the opportunity (by mixing gummy fingers) 
for fraud.

In this report they propose to deliver the doc by post without the 
need for the persons prescence. Obviously someone realised that it 
would be impossible otherwise, but completely forgot that this is one 
of the most important steps in the verification process.

How do they propose to auth with address, 40% of our population who 
dont have a temp roof over their heads.

> > "Given the heterogeneity of fraud detection, one cannot expect a
> > single algorithm to address the situation. A layered Fraud
> > Detection engine must be implemented as a part of the UID project
> > which can detect/prevent various types of fraud. A systems-based
> > approach combined with manual inspection, investigation and
> > learning is essential for a complete fraud detection system.
> > The fraud detection system should be designed to detect who
> > committed the fraud at an individual or organizational level.
> > Depending on the method used, it may also be able to detect the
> > channel through which the fraud was committed. "
> >
> > Nice gymnastic leap of faith from the ground to algorithms.
>
> They might as well have not written this part at all. Would you
> have respected them more then?

>
> > 57 pages that do not address the most basic issues associated
> > with biometric ids let alone id ing 1 .2 billion.
>
> Well I did not read the entire 57 pages. I think its a document
> specifically for developers. Once the development starts, people
> will probably start looking at the document very frequently.

stop conning yourselves. This is the worst possible waste of tax 
payers money. 
The UIDA has neatly side stepped the first fundamental questions of 
why and how of the id process. And are bringing out documents about 
pinning the id uniquely to an individual using tech with KNOWN gaping 
and unfixable holes, algos and fraud prevention and advantage of this 
and that tech.


-- 
Rgds
JTD
_______________________________________________
network mailing list
[email protected]
http://lists.fosscom.in/listinfo.cgi/network-fosscom.in

Reply via email to