Anyone who was not in
a
coma in No vember 2000 remembers the agony caused by the now infamous
butterfly ballots and hanging chads. Concerns about a possible repeat
of events almost caused the California recall election to be delayed.
Following the election
debacle in Florida, Congress became determined that in the next
elections the winners actually would be determined by all the votes
cast. Last October, they passed the Help America Vote Act in order to
help states prepare for the next election. Unfortunately, the solutions
being proposed, involving an assortment of computer-voting systems, may
be worse than the problems they were designed t
We are used to
depending on computers for almost every aspect of our lives, from
governing our bank accounts to controlling our cars. So it doesn't seem
highly radical to suggest computer-aided voting. That is, until you
think of the possible problems.
How can you be assured
after
you vote that the machine actually recorded your vote? With a paper
ballot, even a flawed ballot, at least there is a semi-permanent record
that we can return to - and argue over, if necessary. Would you buy an
airplane ticket by computer if there was no way to obtain a printed
receipt of your transaction?
There already have
been
problems. For example, in the 2002 election, the new computer voting
systems in Florida lost more than 100,000 votes due to a software
error.
Voting is not like a
physics
experiment. We learned in Florida that even if the first attempt is
flawed, no large-scale election is likely to be repeated merely to
verify the result - as one would do in any good scientific measurement.
Thus, you have to get it right the first time and allow some method of
secure verification.
It is not surprising,
therefore, that one of two Ph.D. scientists in Congress, physicist Rush
Holt of New Jersey, has proposed new legislation that would require a
paper record of every vote and require that all software for use in
elections be verified in advance.
In spite of this,
various
states have indicated a willingness to go ahead with systems that
experts in the field find suspect. As reported in the New York Times
last month, software flaws in a popular voting machine, the Diebold
Accuvote-TS machine, make it vulnerable to manipulation. More than
33,000 of these machines are used in 38 states.
In the Science
Applications
International Corporation report, commissioned by of Maryland (which
nevertheless plans to use the Diebold machines in its next election),
"several high risk vulnerabilities" were identified - even based on the
assumption that the machines are isolated and not connected to the
Internet. But in a March primary in California, the Diebold machines
were connected to the Internet with election tallies posted on the
Internet before polls closed.
It is interesting in
this
regard that Walden O'Dell, the CEO of Diebold, an Ohio company, was
quoted in The Plain Dealer as telling Republicans in a recent
fund-raising letter that he is "committed to helping Ohio deliver its
electoral votes to the president next year."
As we rush to install
computer voting systems, we should remember the admonition of a former
chief scientist at Sun Microsystems Inc., who said in a television
interview following the 2000 election: "If your life depended on the
measurement of a single ballot, would you prefer it be read by a
machine, or examined carefully by three different human beings?"
If we are to avoid a
host of
articles on this page explaining how the election of 2004 might have
been stolen, state governments must step back from the current headlong
rush to install computer-voting systems until the necessary
verification systems and security guarantees, certified by outside
experts, are in place. Certainly no one wants to relive the frustration
that followed the 2000 election - without any possibility of rechecking
the results.
Krauss is chairman of
the physics department at Case Western Reserve University.
