[EL] More on the Voter Impersonation Fraud Case in Fort Worth

Lillie Coney coney at epic.org
Thu May 3 12:54:42 PDT 2012


Biometrics are interesting as an identifier, they have been in use for
a very long time and most are just taken as routine such as signatures,
fingerprints, photographs, etc. Biometrics are any thing about a person
that can uniquely identify that person or persons. Do any searches on
EPIC.org on the subjects below for more information.

When biometrics are easy for persons without specialized training to 
determine authenticity then they can be used effectively such as in the 
case of signatures.  

The more difficult it is for a person to view a biometric and make decisions
based on its accuracy when compared with an original the problems
occur.  When analysis of a biometric requires expert training there will be 
questions regarding cost benefit.  When the population of people who
must be authenticated is large and the time that the authentication must take 
place is limited then automation is the only way to accomplish the task.

Automation Problems:

No one has built an automated biometric identification system for a very
large population--say the entire voting age population of the US.  This 
would necessarily include people who are allowed to vote as well as persons
who are not allowed to vote based upon state or federal law e.g. persons
who are younger than 18 on election day.

The largest issue would the false positive and false negative rates and
what to do about them.  A false positive occurs when the systems says
that someone is someone they are not and the false negative rate occurs
when its says some is not someone they claim to be.

Jane claims to be Sam and the System Says that Jane is Sam (False Positive)
Jane is Jane but the system says Jane is Sam  (False Negative)

Privacy 

Voter records are in most states treated as public information. Personal information 
provided for voting or commercial transaction purposes.  The White House proposed 
a very good set of privacy protections for commercial data "Consumer Privacy Bill 
of Rights." http://www.whitehouse.gov/sites/default/files/privacy-final.pdf

Issues:

1) Collecting biometrics
2) Real time capture of biometrics to be compared with a database
3) Reporting of results
4) Public Records nature of voter files
5) Compromise of biometric data, corruption of data, failures of communication
between verification and authentication of information

Signatures have a good track record and are easy for individuals to compare 
and can remain stable over over a period of time.  People are good at what
they have experience in doing. A signature is a biometric and it can be 
captured on the voter registration document.  However, with the use of 
electronic signature capture at store checkouts consumer control over 
signatures may become an issue.  In addition, the signature at the store 
checkout does not resemble signatures made on paper or can be influenced 
by other factors.  In any case the problem with digital information associated 
with an individual is the hidden commercial market for that data "Data Broker 
Industry."  Also digital information can be open to abuse, misuse or manipulation.

Different Types of Biometrics

Fingerprints as a biometric have secondary uses and would be very attractive
to law enforcement.  These would likely be valuable and added to consumer
profiles, which would promote commercial use of this biometric. One print will
give a limited degree of accuracy, the more prints captured the greater the
accuracy.  

Eye Scans can be read very accurately at a distance.  Collection of data would
need to be the population of people involved. The data would be attractive for 
commercial uses and law enforcement.  Linking eye scans to other personal 
information can lead to profiles or dossiers. The larger issue is not knowing 
when your identify was being checked because scanning can take place even 
when wearing sunglasses at a distance. Eye scan recognition is very advanced
and should not be seen as limited by a few feet, but this is the envelop and
distance accuracy will improve.

Voice recognition--voice capture is taking place and has been for sometime
as telephone messaging systems moved from the home to telecommunication
service providers.  The question is how much data is being used to develop
voice recognition for telecommunications purposes or commercial purposes.
Voice recognition is different from speech recognition, being able to translate
spoken words into text or commands that a digital device or person can act on. 
Voice recognition is with a measurable degree of accuracy associate the spoken
word of a person or persons with their identify. Voice recognition also feeds into 
voice analysis that looks at determining a person's emotional state not just words 
or associating the spoken word or sounds to a particular person.
 
DNA presents the same issues as fingerprints in a law enforcement context,
but would also introduce information about family, health, or potential health
issues that are not related to the original collection unless it was for these purposes.
Like other forms of personal information it would be considered valuable
outside of the context of the original collection.

Lillie Coney
Associate Director
Electronic Privacy Information Center (EPIC) 
Defend Privacy. Support EPIC. 
http://epic.org/epic/support.html
1718 Connecticut Avenue NW Suite 200
Washington, D.C.
http://epic.org/
202-483-1140 x 111




On May 3, 2012, at 2:20 PM, Kathay Feng wrote:

> I am interested in the reasons why a thumbprint or other biometric would be good/bad for policy reasons as an identifier or deterrent to the rare case of impersonation.  Lori?
> Sent via BlackBerry by AT&T
> 
> -----Original Message-----
> From: Lori Minnite <lminnite at gmail.com>
> Sender: law-election-bounces at department-lists.uci.edu
> Date: Thu, 03 May 2012 10:50:34 
> To: law-election at UCI.edu<law-election at uci.edu>
> Subject: [EL] More on the Voter Impersonation Fraud Case in Fort Worth
> 
> _______________________________________________
> Law-election mailing list
> Law-election at department-lists.uci.edu
> http://department-lists.uci.edu/mailman/listinfo/law-election
> 
> _______________________________________________
> Law-election mailing list
> Law-election at department-lists.uci.edu
> http://department-lists.uci.edu/mailman/listinfo/law-election

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://webshare.law.ucla.edu/Listservs/law-election/attachments/20120503/d4fc4a01/attachment.html>


View list directory