Electronic voting machines have come under heavy attack in the days leading
up to next week’s election, leaving voters with the uneasy feeling that the
will of the people will not be respected.
Deforest Soaries, the former chairman of the Elections Assistance Commission
(EAC), says voters should be concerned.
According to Soaries, the guidelines published by the agency he once oversaw
fail to ensure that voting machines are accurate and secure.
Now, he said, the federal government should right a wrong by taking
responsibility for developing a prototype, rather than allowing vendors to
create and sell inferior equipment to the states.
“There are some things that the federal government just must do and therefore must see to it that they are done right,” he told internetnews.com.
That might sound like a radical idea, but Ted Selker, a highly-regarded voting technology expert and co-director of the CalTech-MIT Voting Technology Project, noted that this is precisely how voting machines are developed and rolled out in Brazil.
Selker said that three separate government entities in Brazil are responsible for implementing voting technology in that country.
One government agency develops the specifications, one develops the software
and a third does the testing.
Private-sector companies then bid for the opportunity to manufacture the
equipment for general use.
“It’s resulted in tremendous improvements in their processes and equipment,”
said Selker.
Selker suggested that a further improvement on this system would be to have
three different software programs running simultaneously on each machine.
The different programs should each produce the same results, and if they
didn’t, the results of the outlier could be thrown out and an investigation
launched.
“Diversity in the code gives you a more reliable and less hackable system,”
he said.
Selker also suggested a raft of measures that jurisdictions should take to
improve voting security and accuracy.
He said the poll workers should never be left alone with a machine, and that
at least two poll workers should be present when results are transmitted or
transported for final tabulations.
He also said that parallel testing of machines be made a standard practice.
In parallel testing, randomly selected machines are taken out of service and
undergo a simulated election under the same conditions as the real election
taking place on the same date.
If the results produced by the machine don’t match the ballots prepared by
the auditors, then officials know that something is wrong with the software.
There are at least a half-dozen vendors of direct recording electronic
voting machines (DREs) in the United States, the three largest of which are
Diebold Election Systems, based in Allen, Texas, Election Systems and
Software (ES&S), based in Omaha, Neb., and Sequoia Voting Systems, of
Oakland, Calif.
Vendors estimate that DREs are currently in place in one-third of all
polling stations in the United States.
DREs display ballots and record votes by means of touch screens and record
voting data in both permanent and modular memory components.
Moreover, computer security experts have demonstrated that the machines are vulnerable to manipulation through the introduction of malware.
They have also discovered simple interfaces located at the rear of some models of DREs that can be used to change votes.
Other flaws have emerged in the machines in recent weeks, including votes that switch from one candidate to another, screens that freeze when the device memory overloads and truncated names on summary screens.
The irony is that all the machines have been certified by
independent testing authorities using specifications and best practices
published by the EAC last year.
Soaries said that the current system cannot inspire confidence.
One reason is that, as in many other industries, the vendors themselves pay
the independent testing agencies (ITAs).
Soaries says that elections are too important to allow the appearance of a
conflict of interest.
“If you come up with software or hardware that you want certified, I don’t
think you should be the person paying the ITA because the ITA considers you
their client — not the government, not the voter, not democracy,” he said.
He also said that the standards published by the EAC fall flat because they
merely prescribe a set of features to be tested, but do not prescribe
expected results.
“It’s not performance-based standards,” he said.
But Soaries said that ultimately, the blame for poor standards lays with
Congress, which he said failed to ever appropriate $30 million that they had
promised to spend on researching the issue.
“Many of these machines that we’re using were paid for by the federal
government, which didn’t take the time to do the proper research in advance
to say if your machine can’t do this, doesn’t do this, won’t do that, then
we won’t buy it,” Soaries said.
“Congress never appropriated one penny for research that was
supposed to precede the creation of standards and the purchase of machines,”
he said.
Experts are keeping their fingers crossed that elections go
smoothly on November 7.
But they’re also hoping that, regardless of how the machines perform this
time, steps are taken to improve their performance and reassure voters in
the lead-up to the next presidential election two years hence.
“Hopefully there are going to be enough informed people this time to really
insist that Congress not do a watered-down version of election reform but
really get at the hard answers,” Soaries said.