Re: [election-services] Open letter in response to electronic votingarticles in the October Communications of the ACM journal (Volume 51 No.1010/08)

[Richard J Cardone <richcar@us.ibm.com>]

Here, here, David!

Your response to the Communications
article is engaging, provocative, and brings up points that are seldom
raised in polite company.  The recently developed cryptographic schemes
that target end-to-end (E2E) verifiability are interesting and important
research avenues, but I agree they won't provide a practical alternative
to today's systems anytime soon.  The public needs to believe that
its voting system is accurate, secure, and reliable; and verifiability
is an important characteristic in building confidence in a system.  E2E
verifiability, however, while being mathematically elegant, won't necessarily
convince the general public that a voting system is fair because it's logic
is not obvious or transparent--and what the public thinks matters.  Open
source and open standards bring transparency into the existing legal, social,
and logistical electoral frameworks by focusing on how systems are engineered.


I do not agree, however, that "an
underpinning that is founded on paper..." is the only or even the
best way to achieve verifiable election results that can be trusted by
the general public.  That said, the system that you describe has many
nice features and improves the current state of the art.  The ballot
marking component is software independent, and that certainly relieves
us of the problem of proving some of the software correct.  The other
technologies that are part of the system, like optical scan and EML, are
established and workable.  The voting experience would be similar
to what many people are used to, so public resistance should not be problem.
 

On the other hand, the proposed system
does face certain challenges.  First, the optical scan component has
the same strengths and weaknesses of existing optical scan systems, including
the problem of software corruption.  Also, by having 3 distinct records
of a vote, a well-defined reconciliation process is necessary to resolve
discrepancies when the records do not agree with each other.  Unfortunately,
reconciliation processes are often legally mandated with an emphasis on
quick resolution rather than correctness.  On a practical note, in
every election I've worked, temporary printer failure was the cause of
significant inconvenience.  The logistics and cost of dealing with
printers and paper needs to be weighed against paperless approaches, despite
the current movement back to paper.          
 

Rich