Re: Proposed Electronic Ballot Image Format for the Demo

From: Alan Dechert <adechert_at_earthlink_dot_net>
Date: Sat Aug 23 2003 - 17:24:39 CDT

Dennis wrote:
>
> In the interests of easing voter verification, I think it advisable
> to display the office being sought along with the name(s) of the
> candidate(s) voted for. The STCOPREXXXXY needs to appear no more
> than once per ballot. Some ballots have a large number of races. In a few
> cases, the same person can run for more than one office and in a few
> cases, such as County Central Committee, can hold more than one office.
>
I think we have some misunderstanding here. I have no idea why you say "the
STCOPREXXXXY (sic) needs to appear no more than once per ballot." Where do
you see it more than once?

We're talking about how the ballot data will appear in the electronic
record -- one ballot per row. STCOPRECTXXXXY is there only once per row.

> Whatever can be done to make it easier for the voter to verify what
> they voted for should be considered. Some voters aren't too bright.
> Others will have language questions or for whatever reasons will have
> trouble quickly reviewing their vote choices.
>
Okay.

> At the same time, we need the printed ballot to be machine readable.
> These two requirements are not mutually exclusive but they might also be
> somewhat of a question. ...
>
No problem. The printed ballot will be human readable and the font(s) used
should be easily read by scanners too.

> Is the barcode intended for machine readability?
>
Yes! But not human readable.

> If so, how does the voter know that what is written is the same as
> the bar code? If the machine uses OCR, it should be fairly easy but
> the technology is a bit trickier. But not much. I recommend using OCR.
>

http://gnosis.python-hosting.com/voting-project/August.2003/0218.html
Please see Item # 7 here in the Document describing what we need to demo.

     7. ELECTRONIC VOTING MACHINE FOR THE
         BLIND THAT INCLUDES A PAPER PRINTOUT
         BUT WILL NOT COMPROMISE SECRECY
         AND STILL PROVIDE VOTER VERIFICATION

This is a *MUST HAVE* requirement for the demo. If we can't do this and do
it well, there is no point in continuing the project. This is one of the
claims outstanding against the voter-verified paper trail: the League of
Women Voters attorney said that such a printed ballot would likely
compromise the right to a secret ballot for the blind. David Dill responded
to this ridiculous charge but THEY HAVE TO BE SHOWN.

The success of the scheme I describe depends on the barcode being NOT human
readable. This makes it possible for the blind voter to have their ballot
verified ON A MACHINE INDEPENDENT OF THE ONE ON WHICH THEY VOTED while not
having to remove the ballot from the privacy folder (only the bar code will
be visible).

The blind voter will be able to take the printout and put it in the privacy
folder unassisted (we have to test, actually... but we'll work with them to
make sure they can do it). At that point, they can be assisted with
locating the scanner and putting on the headphones without revealing the
contents of the ballot.

If we use OCR, how can you accomplish that? In order for the verification
process to have any validity, it must be done on a separate independently
programmed machine. I think it's way too complicated to expect a blind
voter to move to another machine and insert the ballot into an OCR scanner
(and put on another set of headphones) unassisted. If they are assisted,
then they will wind up revealing their ballot -- BOOM -- you validate the
League of Women Voters claim against voter-verified paper trail. At that
point, you can take our project and David Dill's work and flush it all down
the toilet.

The average voter won't know for sure if the bar code accurately reflects
what the printed text says. However, voters will know that it WILL BE
CHECKED. Large enough samples, randomly taken, will be checked such that it
will be statistically impossible for there to be enough errors in the bar
codes to affect the outcome.

Besides the official certification process where these things would be
checked, I recommend that normally sighted voters be allowed to put on the
headphones and check their ballots with the scanner. Most people won't take
the time to do this (and we don't want people lining up to use the
scanner -- certainly sighted voters would have to yield to any blind voters
that need to use the scanner). But if, say, one percent of the voters check
their ballots with the scanner, you have a very high degree of confidence
that the bar code was accurate -- even before the official certification
process begins.

Check the calculator here:

http://home.earthlink.net/~adechert/LOTAcceptanceCalculator.xls

The Acceptable Lot Fraction Defective would be plus/minus the margin of
victory.

> One could argue that the ballots will only be read during
> a manual recount.
>
No. Ballots will be sampled at various points in the process of certifying
the result. This is a must.

> I think that the ballots should be readable during a machine
> recount using a machine different from the one used in the precinct
> to generate the ballot.
>
Yes. That's partly why OCR is an impossible verification method for blind
voters. They have go to another machine and remove the ballot from the
privacy folder at the risk of exposing the ballot.

> Of course, it needs to be readable for the
> manual recount as well.
>
There is no problem there. The text of the ballot will be easy to read for
humans and scanners.

Thanks, Dennis.

Alan Dechert
==================================================================
= The content of this message, with the exception of any external
= quotations under fair use, are released to the Public Domain
==================================================================
Received on Sun Aug 31 23:17:14 2003

This archive was generated by hypermail 2.1.8 : Sun Aug 31 2003 - 23:17:18 CDT