WORKSHOP ON DEVELOPING A RESEARCH AGENDA FOR ELECTRONIC
VOTING TECHNOLOGIES
September 17-18, 2004
Convened by the American Association for the Advancement of Science
Main | Program | Participants | Synopses
George Gilbert
AAAS WORKSHOP ON DEVELOPING A RESEARCH AGENDA FOR ELECTRONIC VOTING TECHNOLOGIES
Synopsis of Issues
The unreliability of hand counted paper ballots was evident well before the end of the nineteenth century. Each subsequent technological "advance" has engendered its own unique set of management problems. Direct record electronic (DRE) voting is no exception.
Never-the-less, properly managed, DRE voting is the most versatile, the most reliable and the most secure method of voting yet devised. This assertion is particularly applicable to American elections which tend to involve lengthy, complex ballots.
Obtaining repeatable, accurate tabulations, historically, has been among the most challenging tasks facing election officials. During 16+ years of DRE voting in Guilford County, NC, our voters have cast nearly 2 million ballots on this equipment. Recounts of at least one contest, have been conducted in a majority of the 51 elections conducted during this period. As expected, the recounted results from our DRE voting equipment has never varied from the original count.
Insuring the security of voted ballots is another major imperative. Of the
near 2 million DRE ballots cast in Guilford County, only 36 ballots were not
properly recorded and only 4 ballots were so defective so as prevent tabulation.
Of the 36 ballots which were not properly recorded, all were identified and
the voters casting those ballots were invited to cast another ballot; 32 did
so and all of their ballots were counted. This is a total error rate of .002%,
or one out of every 50,000 votes. From the perspective of elections officials,
the chief importance of this finding is not that the margin of error is statistically
infinitesimal (though this is certainly important), but that such errors were
detected and corrected (to the extent possible).
Results such as these are accomplished not because the technology is perfect,
but by thoroughly and rigorously verifying and documenting the accuracy or our
voting machines and tabulation system before the election, and by an even more
thorough and rigorous auditing of the results after the election; with both
procedures being open to public scrutiny.
------------------------------------------------------------------------------------------------------------
It is significant that there is no presumption of perfection in election law.
Every state has its procedures for handling election protests. In North Carolina,
the fundamental issue addressed by election protest rules is, "is there
substantial evidence to believe that a violation of the election law or other
irregularity or misconduct did occur and that it was sufficiently serious to
cast doubt on the apparent results of the election."
"Irregularities" occur in every election. Most are minor or would
not affect the outcome. The closer the contest and the nearer the top of the
ballot, however, the more brightly the lights shine on these incidents. Unlike
every other office on the ballot, the timeframe for resolving presidential election
protests is limited. Similarly, the option of a new election does not exist
as a remedy in the presidential contest. If the uncertainty involves the tabulation
of votes, it is imperative that the voting system be capable of rapidly and
accurately recounting the ballots. That imperative cannot be met by hand counting
paper ballots.
------------------------------------------------------------------------------------------------------------
While this workshop has been identified as an "AAAS workshop on E-Voting
Technologies," of necessity the discussion must encompass all voting systems.
The strengths and weaknesses of one type of voting system can only be meaningfully
evaluated in relation to other types of systems. In the current climate, the
effectiveness, cost and reliability of electronic records must be compared to
the effectiveness, cost and reliability of alternative voting mechanisms.
The evidence regarding "effectiveness" is overwhelmingly in favor of DRE voting. While no system is, or can be made perfect, DRE voting machines offer the possibilities of far greater accessibility, ease of use and accuracy than do any type of paper based voting systems. These possibilities are acknowledged even by most of the harshest critics of DRE systems.
Cost must be assessed in several contexts. The direct or upfront costs of most voting systems is not particularly significant from a national perspective. It is often very significant, however, at the local level where most of the cost is typically incurred. One time federal grants will do little to alter the local, long term perspective.
Among the most significant cost factors that is generally disregarded until
too late, is the cost of a recount. The cost of a DRE recount (by which I mean
retabulation of the ballot images) is negligible. The cost of recounting ballots
using an optical scan system, while somewhat greater, is not a significant budgetary
issue. The cost of recounting ballots by hand is potentially several times the
cost of conducting a whole new election. The variable costs of conducting the
presidential election of 2000 in Guilford County, NC, was roughly $200,000 to
$250,000. The estimated cost of a full ballot hand recount of paper ballots
is $1.2 million. More importantly, such a count would never be reliable.
------------------------------------------------------------------------------------------------------------
The decentralized character of election administration in this country remains
the strongest protection against fraud or manipulation impacting the outcome
of an election. The trend toward centralization, both in voter registration
and voting system selection, while intended to enhance uniformity and quality
in election management, also enhances vulnerabilities as well as the scope of
impact of either fraud or error.
Error being the greatest threat to the integrity of elections, DRE voting systems,
in general, lack one crucial feature….that is a mechanism for storing and
securing the electronic ballot images (the ballots) from being destroyed by
administrative error or electro-mechanical failure. Electronic ballots can be
made even more secure that paper ones. The technology is available or, at least,
a very short step away. It is time for it to be applied.