Re: Text of count every vote act and nothing new under the sun.

From: Ed Kennedy <ekennedyx_at_yahoo_dot_com>
Date: Sun Feb 27 2005 - 23:43:13 CST

Hello Ron:

We've had seemingly endless discussions about bar codes versus optical
scanners generally with the arguments you have presented below. The
consensus has been that even with their drawbacks, bar codes are better due
to reliability information density. In terms of priority regarding
development, I'd suggest that the methodology of how machines read ballots
may be less important than you imagine. I'll state that I'm personally
inclined toward OCR but I'm quite aware of squirrelly current OCR technology
is. I believe the last discussion had something to do with a voter
inadvertently dropping a piece of chocolate
on their ballot. I'm willing to believe that the reliability of OCR can
improve but I don't think it is something that can be assumed. By the way,
when I say optically scanned ballots, I'm talking about mark sense scanning,
not OCR. Here's an excellent discussion by Doug Jones of that system:
http://www.cs.uiowa.edu/~jones/voting/miamitest.pdf.

Thanks, Ed Kennedy

Ron Crane wrote:
> On Feb 27, 2005, at 2:55 PM, Charlie Strauss wrote:
>
>> The Count every vote act (sens Clinton kerry) is available.
>> http://www.pfaw.org/pfaw/dfiles/file_493.pdf....
>>
>> The bill does not seem to anticipate that bar codes will get used
>> (they are part of the sequoia system and are integral to the accupol
>> system). Thus it does not spell out how bar codes get vetted
>> against the text during the audit or recount. This is a golden
>> opportunity for OVC to hammer this point home. It is the main
>> advantage of the OVC system over the Accupol system for example. And
>> certainly use of bar codes is going to be needed for efficient
>> handling of flimsy paper tape systems as well.
>
> Bar codes shouldn't be used. Though they improve the efficiency of
> ballot handling, they add security risk (e.g. voting station prints
> bar code that doesn't represent voter's selections even though text
> does) and reduce transparency (voter isn't sure how vote is represented).
> If
> the polls are busy, I'd guess that many voters would skip the ballot
> reader/verifier and go right to the ballot box, thus never finding
> bar-code mis-correspondence. And, of course, the ballot reader could
> be compromised to ignore this kind of cheating.
>
> OCR, though more error-prone [1] and less efficient than bar codes, it
> more secure and more transparent.
>
>> OVC could further improve its standing by making one of its
>> disavantages an heralded advantage. OVC requires that all the
>> ballots get hand scanned by poll workers. We all agree we like that
>> for many reasons but of course the down side is manual labor. However,
>> since the manula audit provision of the law means there is
>> going to be a manual audit phase this is an opportunity for OVC. OVC
>> should assert that the OVC process does not require a secondary
>> manual audit since that is part of its voting process. One could do
>> this voluntarily, as it does add to the integrity, but OVC should
>> take the stand that it is not required for OVC. this will reduce
>> the cost of operation of the OVC system compared to others that are
>> less manual-audit friendly.
>
> It'll do so only at a serious sacrifice of security. The manual audit
> is meant to make tabulation fraud much more difficult by
> cross-checking a randomly-selected, statistically-significant set of
> precinct
> tabulations against hand counts the gold standard of security.
> Nothing in the OVC system replaces this important cross-check, though
> possibly elections officials could hand-audit as they scanned ballots.
> That seems a clumsy process, and it doesn't satisfy the usual recount
> requirements of observers, partisan balance, two teams, etc. It also
> doesn't satisfy the legal requirements of, e.g. HR 550 (should it
> become law), which requires federal officials to perform the audit.
>
> More broadly, the main purpose of OVC's system should be to provide
> the most secure system possible, not necessarily the most convenient
> for elections officials to use. Of course it shouldn't be
> substantially harder than others to use if that can be avoided without
> sacrificing
> security.
>
> -Ron
>
> [1] Errors, unless the rate is high, concern me much less than
> cheating. Errors are, by definition, randomly-distributed, and thus,
> on average, won't skew elections. Cheating, on the other hand....
>
>
> _______________________________________________
> OVC discuss mailing lists
> Send requests to subscribe or unsubscribe to
> arthur@openvotingconsortium.org

_______________________________________________
OVC discuss mailing lists
Send requests to subscribe or unsubscribe to arthur@openvotingconsortium.org
==================================================================
= The content of this message, with the exception of any external
= quotations under fair use, are released to the Public Domain
==================================================================
Received on Thu Mar 31 23:17:03 2005

This archive was generated by hypermail 2.1.8 : Thu Mar 31 2005 - 23:17:09 CST