[Politech] Australian gvt tries to make face recognition look good [priv]

From: Declan McCullagh (declan@private)
Date: Sun Feb 08 2004 - 21:46:12 PST

  • Next message: Declan McCullagh: "[Politech] FBI wants computer repair shops to report Kazaa, DeCSS users? [priv]"

    ---
    
    Date: Sat, 7 Feb 2004 16:11:20 +1100
    From: Roger Clarke <Roger.Clarke@private>
    Subject: Face-Recognition Technology - FOR DISTRIBUTION
    
                    PLEASE DISTRIBUTE TO ALL RELEVANT LISTS
    
    
             AUSTRALIAN GOVT TRIES TO MAKE FACE-RECOGNITION LOOK GOOD
    
    ________________________________________________________________________
    
    Applications of face-recognition technology have been uniformly 
    unsuccessful.  Technology providers and security agencies are desperate for 
    some good news.
    
    It's very likely that a project called SmartGate, conducted by the 
    Australian Customs Service, will be trumpeted throughout the world as the 
    good news that face-recognition technology has been waiting for.
    
    This email contains an assessment of the extreme manipulation of data, 
    truth and the media on which such 'good news' stories will be based.
    
    ________________________________________________________________________
    
    Background
    
    SmartGate is the Australian Customs Service's pilot face-recognition 
    system, using German Cognitec technology.  It has been trialled for a year 
    on QANTAS aircrew.  The stated intention of the Australian Government is to 
    extend it further.  This project is running in parallel with a Passports 
    Office project to change Australian passports to conform with U.S. demands.
    
    Customs have been trying to convince suppliers to finance the development 
    of the scheme.  It does not appear that any independent testing has been 
    performed;  but a couple of international experts were invited to review 
    the internal test results.  A media event was held on Friday 6 Feb 04.  The 
    information made available to the public is listed below.
    
    ________________________________________________________________________
    
    Analysis of the Available Information
    
    The brazenness of Custom's manipulation of media and public opinion exceeds 
    the standards normally expected of the Government.
    
    (1)  No data has been provided, despite assurances given in the past that 
    data *would* be provided.
    
    (2)  The invited experts who were paraded by Customs are arguably about the 
    world's two foremost designers of testing for biometric technologies.
    
    But is appears that:
    (a)  they did not *perform* the tests
    (b)  they did not *design* the tests
    
    All that they were permitted to do was explained as follows:
    -   they "reported on three reports"
         (only one of which was a technical test)
    -   they "conducted independent obervations" (i.e. watched it in operation)
    -   they "reviewed SmartGate data logs [and other, derivative documents]"
         (which are meaningless without additional information, and analysis)
    
    (3)  The only report that matters was undertaken by the Defence Science and 
    Technology Organisation (DSTO).  There are some serious problems here:
    
    (a)  no information whatsoever has been provided about that study, i.e. the 
    public is expected to accept at face-value the assertions made about the 
    methods used, the controls applied, and the outcomes;
    (b)  DSTO is not in any sense independent.  It is a government agency.  Its 
    assurances need to be regarded with the same degree of scepticism as any 
    other arm of a government that has demonstrated itself to regard national 
    security as paramount and what it releases to the public as a means to an end.
    
    (4)  The best quotation that Customs seem to have been able to extract from 
    the two experts was that "the scheme's performance is remarkably good for 
    an operational facial recognition system".
    
    That seems quite positive, until you realise that the other attempts around 
    the world have been abject failures, and pilot after pilot has been quietly 
    abandoned.
    
    In other words, the quotation can be readily interpreted as "it doesn't 
    work very well, but it's better than the other disasters we've seen".
    
    This is borne out by an answer by one of the experts to a reporter's 
    question.  He said: "In one test 100 company employees attempted to 
    impersonate someone other than themselves and eight of them were falsely 
    accepted by the system. That is a very low rate of false accepts".  At 
    that"very low" level of false acceptances, every 747-load of people can 
    include 25-30 terrorists;  and some (undeclared) number of people will miss 
    the plane because they were false rejections, and they're still waiting in 
    the queue for interrogation.
    
    _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
    
    The Minister had promised that the announcements about the tests would 
    vindicate his claims that the scheme works.
    
    This media event has done two things.
    
    It has confirmed that the Government intends to continue constructively 
    misinforming the public about Smartgate and about biometrics technologies.
    
    And it has, in the process, tarnished the reputations of the two scientists.
    
    ________________________________________________________________________
    
    The Available Information
    
    (I wasn't invited to the media event, and haven't been provided with any 
    information by Customs.  That's fair enough:  I'm not a member of the 
    media.    But on the other hand I've been much-quoted in the media;  and 
    Customs have previously sought meetings with me, and adopted the pretence 
    of wanting to keep me informed).
    
    The media release is at:
    http://www.customs.gov.au/site/index.cfm?area_id=5&content_id=14859
    
    This points to two marketing brochures (1 x 1-page and 1 x 2-page) at:
    http://www.customs.gov.au/resources/Files/media%20background%20smartgate.pdf
    http://www.customs.gov.au/resources/Files/media%20background%20smartgate%20information.pdf
    
    No further information was provided.
    
    The unlinkable page reached via:
    -   'protecting our borders' / 'border technologies' / 'SmartGate'
    has had all previous documents withdrawn from it, and the only links 
    available are the two marketing brochures pointed to above.
    
    None of the other 17 hits arising from a search on 'SmartGate' on Customs' 
    web-site appear to be new.
    
    ________________________________________________________________________
    
    
    Additional background information is at:
    http://www.anu.edu.au/people/Roger.Clarke/DV/SmartGate.html
    including:
    -  links to the Customs site
        http://www.anu.edu.au/people/Roger.Clarke/DV/SmartGate.html#SmartSrces
    -  links to media reports
        http://www.anu.edu.au/people/Roger.Clarke/DV/SmartGate.html#Media
    
    Sceptical reports about face-recognition technology, including the 
    authoritative ACLU materials and EPIC's resource-page, are indexed at:
    http://www.anu.edu.au/people/Roger.Clarke/DV/BiomBibl.html#RSFR
    
    ________________________________________________________________________
    
    
    -- 
    Roger Clarke              http://www.anu.edu.au/people/Roger.Clarke/
    			
    Xamax Consultancy Pty Ltd, 78 Sidaway St, Chapman ACT 2611 AUSTRALIA
                     Tel: +61 2 6288 1472, and 6288 6916
    mailto:Roger.Clarke@private            http://www.xamax.com.au/
    
    Visiting Professor in the eCommerce Program, University of Hong Kong
    Visiting Professor in the Baker Cyberspace Law & Policy Centre, U.N.S.W
    Visiting Fellow in Computer Science, Australian National University
    
    _______________________________________________
    Politech mailing list
    Archived at http://www.politechbot.com/
    Moderated by Declan McCullagh (http://www.mccullagh.org/)
    



    This archive was generated by hypermail 2b30 : Sun Feb 08 2004 - 22:13:55 PST