Re: Secret ballot and the Colorado constitution

[Paul E Condon <pecondon@xxxxxxxxxxxxxxxx>]

On Tue, Feb 28, 2006 at 12:00:34PM -0500, alkolwicz@xxxxxxxxx wrote:
> Paul,
> 
> If I understand your concern, it is resolved in Colorado Election Law which
> provides for pre-printed ballot-serial-numbers reocrded on "removable
> stubs".  Colorado law says that the the "issuing judge" is to record the
> stub number alongside the voter name/id in the pollbook, and that the stub
> is removed by the election judge BEFORE the ballot is cast by the voter and
> retained (for audit).  In this way a secret ballot is created.  The voter
> and election officials have no way to associate this ballot number to the
> ballot   (short of comparing the ragged edges).

Computer programs that match up edges in images exist and are used successfully
in the processing of satelite images of the Earth. They could be used to
match images of torn off stubs to images of ballots. 

Also, I seriously doubt that the writing of the number onto the pollbook
would be reliable enough to have the data be useful for audit. So, even if
one matched the stub image with the ballot image, matching the ballot image
with the voter would produce a lot of false positives. 

So, I think the law is unenforceable, useless for audit because the
data that it would produce would be very unreliable, a waste of effort
of the judges, and generally the kind of dumb idea that legislators
come up with.  But if it did provide reliable data, that reliable data
could be used to match ballots with voters.

But perhaps if the stub with unique ID were had a double line of tear
strip so that there were a third strip of paper in between the main
ballot and the small stub containing the unique ID and these extra
strips were guarenteed, for sure, never to be saved or scanned
... perhaps then it could be useful.  But that won't happen.

Somewhat off topic, but worthy of consideration: The Census Bureau
collects data on individuals, but has a very strict system of
separating peoples' names for the other data at a very early stage in
the processing of the data. The Census Bureau bureaucrats work hard at
maintaining the separation.  I think they are generally
successful. There have been research projects for which access to the
name data has been allowed, but for a project to be granted access it
must be review and justified and documented in great detail.  It is
hard to get access, and every civil servant employee of the Census
Bureau knows that getting permission is a big deal and trying to get
access without permission is a serious offense. This spirit of respect
for individuals and respect for data is far less prevalent in the
election bureaucarcies in the states and counties. But, maybe there is
something to be learned from how the Census Bureau works.

> 
> The secondary ballot serial number, which would be printed at the time the
> ballot is scanned for counting, is a way to uniquely identify ballots and
> physically locate them in batches.  This second number would have no
> association to a specific voter.  Its purpose is to facilitate comparison
> of the markings on the physical ballot to the interpretation of the ballot
> by the computer. 
> 
> Al
> 
> 
> 
> Original Message:
> -----------------
> From: Paul E Condon pecondon@xxxxxxxxxxxxxxxx
> Date: Mon, 27 Feb 2006 22:05:45 -0700
> To: cvv-discuss@xxxxxxxxxxxxxxxxx
> Subject: Re: Secret ballot and the Colorado constitution
> 
> 
> No ID marks until the moment of scanning was Paul W.'s idea intended
> to address the problem of some spy-ware or some hidden cameras getting
> information that could be used to associate a particular marked ballot
> with a particular voter. Maybe these problems need to be addressed in
> some other way. Perhaps, we can:
> 
> 1) Discourage hidden cameras and other non-computer spy techniques by
> application of traditional techniques of physical security, the kind
> of stuff that security guards are trained to do. Use State Troopers, etc.
> 
> 2) Address computer code spy-ware by requiring that *all* software in
> *any* aspect of vote processing be Open to general inspection by any
> concerned person. In this I really mean "all" and "any". Some will say
> that this is too extreme, but if computers are to be used, and there
> is strong pressure to use them, then the software must be truly verifiable.
> There *are* techniques for ensuring that the actual software used is
> identical
> to the software approved by some testing authority. There are techniques
> for ensuring that the 'machine code' software was actually derived from
> some particular version of 'source code'. The trouble with these techniques
> is that they leave very little wriggle room for 'proprietary' software,
> and some people believe that 'proprietary' is the American Way. 
> 
> Any other ideas as to how to have both unique ID that precludes ballot
> stuffing and end-to-end voter privacy? Without some, we seem to have
> an unresolvable contradiction.
> 
> In the past, paper ballots were printed on special paper stock and given
> watermarks and other special treatment that were very hard to copy. But
> 'modern' ballots that are printed on laser printers are really rather
> easy to reproduce exactly with modern laser printers. 
> 
> This laser printer ballot box stuffing seems to me to be a problem for
> all forms of vote counting that use ballots printed on laser printers,
> including hand counting of votes. So if one allows the possibility of
> laser printed ballots, one commits to requiring a lot more physical
> security than most people think is reasonable. There is a real citizen
> education problem here. 
> 
> On Mon, Feb 27, 2006 at 08:31:27PM -0700, Some Guy wrote:
> > Problem --- if I photocopy ballots and stuff the box and there are no
> > pre-printed serial numbers on the ballots, then how can you tell that
> there
> > are extra ballots? Eh?
> > 
> > paul t
> > 
> > -----Original Message-----
> > From: Paul E Condon [mailto:pecondon@xxxxxxxxxxxxxxxx]
> > Sent: Monday, February 27, 2006 1:17 PM
> > To: cvv-discuss@xxxxxxxxxxxxxxxxx
> > Subject: Re: Secret ballot and the Colorado constitution
> > <snip>
> > I had been thinking, rather narrowly, on only the situation of
> > pre-printed ballots that are handed to voters by judges in a precinct
> > voting place. Your comment points to the value of a print head in the
> > scanner at the ballot processing stage. If each ballot is marked at
> > the time it is scanned, that mark can be used to ensure that ballots
> > are not double counted. But it plays hob with reprocessing ballots
> > that will need to be done because of computer malfunction.
> > 
> > Putting the scanners in the polling place could spoil your good idea.
> > If the ballot is scanned and marked while the voter is in a vacinity
> > there is still an opportunity for learning the association of vote
> > with voter.
> > 
> > --
> > Paul E Condon
> > pecondon@xxxxxxxxxxxxxxxx
> > 
> > --
> > No virus found in this incoming message.
> > Checked by AVG Free Edition.
> > Version: 7.1.375 / Virus Database: 268.1.0/269 - Release Date: 02/24/2006
> > 
> > --
> > No virus found in this outgoing message.
> > Checked by AVG Free Edition.
> > Version: 7.1.375 / Virus Database: 268.1.0/269 - Release Date: 02/24/2006
> > 
> > 
> 
> -- 
> Paul E Condon           
> pecondon@xxxxxxxxxxxxxxxx
> 
> 
> --------------------------------------------------------------------
> mail2web - Check your email from the web at
> http://mail2web.com/ .
> 
> 

-- 
Paul E Condon           
pecondon@xxxxxxxxxxxxxxxx