: David Moss
: 20 November 2009 15:23
: Duncan Hine
: XXXXXXXXXX; Face
recognition. NIST? Or the FBI? You can't both be right.
Dr Duncan Hine
Integrity and Security
Identity & Passport Service
Dear Dr Hine
Thank you for your letter dated 16 November 2009 in response to mine
dated 30 October 2009.
You kindly offer to have Kevin Sheehan, Director of Identity Assurance
and Document Services, "take this forward".
Gratitude needs to be tempered with scepticism. On 18 September 2009,
Lin Homer wrote promising me a meeting with Daniel Souter, Assistant
Director of the Border Force, please see attached. Two months later,
the meeting hasn't even been arranged, let alone taken place. That meeting
is meant to discuss the questions about eBorders raised in my letter
dated 8 August 2009.
It is to be hoped that "take this forward" means "promptly
inform the public convincingly why we should trust IPS's and UKBA's
choice of face recognition and flat print fingerprinting technology
to help fight crime and terrorism and to improve the efficiency of public
You said in a letter to the Guardian (15
January 2009) that: "We plan to use all 10 fingerprints and
facial biometrics to ensure someone can only enrol on the scheme once,
thereby preventing multiple identities being established". As you
must know, if only from my letter published on 25
January 2009, identification like that is impossible.
Mr Sheehan needs, therefore, to concentrate on the use of biometrics
for verification, not identification. The UKPS
biometrics enrolment trial (please see §1.2.1) revealed that
face recognition fails to verify identity around 50% of the time, and
flat print fingerprinting aound 20% of the time.
The Home Office claim that that wasn't really a trial. Which gives Mr
Sheehan a problem. Does that mean that the Home Office are spending
taxpayers' money on technology without first conducting trials to see
if it works?
Not quite. You have subsequently conducted trials of face recognition
at Manchester airport which provoked the memorable claim that this technology
cannot distinguish between Gordon
Brown and Mel Gibson, nor between Osama bin Laden and Winona Ryder.
The smart gate technology has nevertheless now been deployed at 10 UK
airports. And despite the entreaties of the House of Commons Science
and Technology Committee, no trial results have been published:
"We are surprised by the Home Offices unscientific approach
... Given the relative lack of information available publicly regarding
the performance of biometrics in a national scheme, we recommend that
once the scheme is established the Home Office publishes details of
the performance levels of the technology" (para.91).
You haven't published any performance results but you have told The
Register magazine that: "We will not give the error rates
or technical specifications of the gates for commercial and security
reasons". Mr Sheehan will need to explain why a company selling
equipment that works would want to keep that fact confidential and why
our security would be threatened by telling us how their equipment would
make us more secure. He will need to explain also why the FBI are wrong
when they say that the face recognition algorithms just dont exist
to deliver the highly reliable verification required.
The Home Office haven't conducted any trials of flat print fingerprinting
since the UKPS biometrics enrolment trial. But 1,800 business schools
around the world have. For two years, candidates sitting the GMAT
entrance exam had their identities verified by their flat print fingerprints.
It didn't work, fingerprints
have been dropped, and the schools are now trying veinprint technology
instead: "The new technology is a more accurate, more efficient
and less invasive way to ensure that each test taker has a single GMAT
record, preventing people from taking the test for others".
In the absence of any large-scale field trial, the only defence
the Home Office advance for their trust in the biometrics they have
chosen is a "chimera trial", as Nigel
Sedgwick calls it, conducted by the US National Institute of Standards
and Technology (NIST).
NIST's record on predicting the reliability of biometrics is poor.
An enquiry has been sent to the lead author of this report, please see
attached. Who is right? NIST? Or the FBI? He hasn't answered. Can Mr
We have a Home Secretary who fires scientists who tell him the truth
and an Identity Minister who thinks we should spend a fortune on systems
to help us collect parcels from the sorting office. We have smart gates
installed at 10 UK airports against the best advice of the FBI and against
four decades of evidence of the failure of the face recognition biometrics
industry. As far as we know, flat print fingerprinting fails to verify
the identity of about 20% of people. And now we learn, via a press release
in Paris, that IBM, who have a £265
million contract to create the biometric National Identity Register,
have contracted with Sagem
to supply face recognition and flat print fingerprinting technology
before we even know that the technology works.
Mr Sheehan has a lot to "take forward". He must provide independently
audited large-scale field trial results promptly which will convince
the British public that our money is not being wasted and that our security
really is being enhanced. Two months is not long enough for the UK Border
Agency to justify its eBorders strategy. Can the Identity & Passport
Service do better? Can Mr Sheehan lay his hands on the documentary evidence
which justifies the spending of £5 billion by the Home Office
alone on the National Identity Service in, say, two weeks? A sort of
Christmas present to the long-suffering British taxpayer who finds it
harder and harder to respect
anyone in government, whether politician or civil servant?
Business Consultancy Services Ltd
The attachment(s) to this email, if any, have been checked by Norton