Author: JT Smith
Open Source business –
Florida: It’s become a one-word description of how not
to conduct a political election. But last week I noticed something very
curious
about what was going on. Obscured by the vapid media reports and the
tiresome
political posturing is a far bigger problem than just one very
screwed-up U.S.
election. It’s a technical problem that may have tilted hundreds of
elections
around the world. And it’s a problem that the hacker community is in a
unique
position to help fix, if they want to.
I confess to being a political junkie. Although I had a hard
time working up much of a sweat about either of the major party
candidates, I
felt the same guilty thrill watching the election results one gets from
witnessing a car crash. But what engaged my inner geek was the largest
technical problem that affected the election. George Bush campaign
manager
Karen Hughes was right to complain about the selective recounts in
Florida
saying,”The votes in these four selective
counties are being counted differently than votes in Florida’s 63 other
counties.” Unfortunately, they were also tabulated differently, and the
various
systems have such a range of error rates that there is almost no way to
objectively settle the issue.
In most of the United States, and in many so-called “advanced” nations,
the processes used for tabulating voting results are inconsistent,
discriminatory against elderly and inner city populations, and use
technology that is downright third world. It’s a hidden scandal that
shouldn’t
be hidden any longer.
Over the last hundred years, the right to vote has gone from
the privilege of a few to a basic right; in some countries, a mandatory
responsibility of citizenship. Many countries have agonized over ways
of
increasing low voting rates. This is largely because differing voting
rates
have led to something less than representative government. Turnout
rates
have
traditionally skewed election results in favor of more educated, middle
class
issues. A
government
review of the 1994 U.S. congressional elections found that college
graduates
were three times more likely to vote than high school drop-outs. Low
voting
rates among less educated, more urban voters is not just an American
phenomenon. In the United Kingdom, even after years of agitating for home rule,
only
30% of
the London voters bothered to cast ballots for their Greater London
Assembly.
In local elections, poor
British
communities such as Conisbrough featured voting rates as low as 24%.
While most of us have learned to live with the fact that
many people don’t vote, it’s become far more upsetting to learn just
how
many
people lose their right to vote because of primitive voting systems,
which
disproportionately affect poor and urban districts. It’s downright
outrageous
to learn that most handicapped people have little recourse to a secret
ballot.
This is entirely due to the use of old, unreliable voting
technology.
In many urban areas and older industrial communities, voters
are faced with ancient mechanical voting stations or have to use
notoriously
unreliable punch card systems which have tripped up hundreds of
elections in
the United States and elsewhere.
Many urban areas in the United States still
rely on creaky old mechanical tabulators. For example, Philadelphia’s current
voting
machines are between 25 and 50 years old, weigh more than 1000 pounds,
were
built by companies that no longer exist, feature parts that can’t be
replaced,
and use a lever system that handicapped and elderly people find
intimidating
and difficult to use. The rate of breakdown is high. The result is
inaccuracy,
and long delays during major elections.
New York City has had similar problems. Daniel De Francesco,
executive
director of the New
York City Board of Elections, has said that in 1998, heavy turnout
and
“old machines” had a negative affect on totals. Things got so
bad
that on a recent Election Day, 70 repairmen were spread out across the
city
repairing machines while voters waited. Other equipment problems
arising
from
antiquated systems hung up elections in Detroit. In ’98, the Michigan
Democratic Party tried unsuccessfully to keep polls open for an
additional
three hours, and some voters were turned away or walked out of polling
places. This year, problems with
equipment led to lines in St. Louis. Political partisans
wrangled about whether to keep polling places open.
Other antique and inherently
flawed systems have spoiled ballots even after they were cast.
The worst of these systems has been
the punch card ballot. Derived from technology used to tabulate the U.S.
Census
of 1890, these system had problems even then, and as the recent snafu
in
Palm
Beach country demonstrated, they haven’t improved much.
In 1986, the respected John and Mary R. Markle Foundation
approached the National Institute of Standards and Technology to
learn more about the errors generated by such systems. The 1988 NIST
report
recommended the elimination of such punch card systems, particularly
“pre-scored” (partially cut) ballot cards, citing inaccurate voting
counts that
resulted from their use. A few years earlier the state of Illinois had
audited
its electro-mechanical tabulation systems and found that 28% of them generated tabulation errors.
The consequences of such errors can be severe.
In October of 1995 the voters of
Quebec went to the polls to vote on whether to secede from the rest of
Canada.
The “no” votes won by a razor-thin margin of only 1.12% or 52,488 votes.
Unfortunately, there were more than 86,000 “spoiled ballots”
in that
election, nearly half again the margin in the election!
In a recent primary election in
the state of Hawaii, more than
9,300 votes were spoiled by bad equipment.
Thousands more had their ballots kicked back or damaged and had to
revote, sometimes several times. In 1998, the spoilage rate was even
worse;
high enough to require a state-wide recount, even though most races
were
not
particularly close.
In Peruvian elections a few
years back, poorly maintained punch card systems were partially
responsible for
a ballot spoilage rate of that approached 6%. The result contributed
to
an international incident.
The U.S. State Department cited the antique vote counting systems used by
Peruvian officials as one of their many reasons to denounce the
election. “We do not
see the election as being valid… No president
emerging from such a flawed process can claim legitimacy.”
In a 20-year-old study, investigators
found that one in 20 Ohioans accidentally invalidated their vote
for
governor
because they “over-voted” in the gubernatorial election. Voters either
punched
or nicked more than one punch hole, or an extra tab simply fell out on
its own
during the tabulation process. In Cincinnati, attempts by election
workers to
clean up bad punch cards with tweezers after the 1985 election caused a
small but memorable
scandal.
Improperly punched cards are only part of
the problem. Another can be poorly made, imprecise cards. In
southeastern North
Carolina, Duplin County election officials found that the punch card
holes
didn’t line up properly with candidate names so elections director
Rosemary
Futrell sent them back. The second batch was no better. “The third
time,
they
got it right,” said Futrell. North Carolina, like many states (like
Florida!) uses a hodgepodge
of systems, and spoiled ballot percentages that vary accordingly,
damaging the credibility of close votes.
So what happens if you are a
local official in a close race being decided by punch card ballots? You
spend
lots of money and wait. A
local coroner’s race in Missouri took over two years to resolve,
nearly
bankrupting the two major contestants.
Government officials outside the
United States have been vigilant about ballot spoilage, particularly in close
elections.
In a Toronto city
election during the mid-’80s, the spoilage rate amounted to
little more
than 1%, but the margins in several minor races that year were less
than
a
third of that spoilage rate. Although in most districts no races were
at
stake,
the city went ahead with a comprehensive recount of all ballots across
the
city. Election officials explained their action as an attempt to audit what had
gone
wrong with the count.
No amount of vigilance is a substitute for a better system, however.
Most people assume that better system should be computer-based.
The ACM developed a suggested
set of Characteristics
of a Good Electronic Voting System some years back. Many of the
electronic
systems being offered lack significant numbers of those
characteristics.
One commercial developer of
electronic voting machines, Microvote,
has developed a system it calls “Infinity Voting Panels.” These
systems
were
used in a number of general elections in the United States this year.
They have
many advantages. They are lightweight, and designed to be easily
portable, even
to the extent of being taken out to the curb to allow the mobility
impaired to
vote from their car. But the systems use dull, gray-on-gray liquid
crystal
displays, making for potential visibility problems. The company
says
that
the machine has what they call “transflextive”
lighting, which allegedly brightens or darkens in response to the light
in the
area. But they are still more difficult to read than a traditional
ballot.
That’s important. An inability to read the ballot is what started
the whole mess up
in Century City, Florida, a Palm Beach community of senior citizens,
many
of
whom complained they couldn’t read the ballot they were given. Asone advocacy
group put
it, “Access to a secret ballot is guaranteed by most state
constitutions, but it is only a dream for those unable to use current
voting
technology.” Many
advocates for the handicapped prefer a design similar to eSlate’s
Disabled
Access Unit, the DAU 5000, a computer system developed by Hart InterCivic which
features a
specially modified computer for use by the handicapped. The machines
cost about
$2,500 each, with additional modifications such as the voice
synthesizer
adding
another $1,000 to each unit.
In addition the software for many of these systems is less than
perfect. Morgan County, Indiana, used Microvote machines, but eyebrows
were raised when the system seemed to have the candidates’ names listed on ballots in reverse order of what was intended, leading to last-minute
reprogramming,
and complaints by the losing Democrats.
Some activists maintain that the
software in most electronic
voting systems is insecure and allege that it would not take a
great
of
manipulation to defraud elections. In California, Gov. Gray Davis cited
a
variety of
security concerns when he vetoed a bill that would have let local
officials use
an electronic voting system based on Internet technology at several
polling
stations.
Electronic voting critics were nearly
proved right in the most recent elections. In New Mexico’s Bernalillo
County,
election officials nearly missed thousands of votes logged on
electronic
systems developed by Global Election Systems Inc. because “someone in
the
county
clerk’s office” failed to “click a box on the computer menu” to
activate
the
system’s ability to count straight-party-ticket voting. The result was a
two-day
delay to manually re-tabulate votes.
Many observers have suggested that the
only means of guaranteeing that votes are fairly tabulated is to
develop
a system based on an auditable, Open Source software platform.
This could help the other problem
holding up widespread adoption: the cost of new technology.
Philadelphia
is going to employ new
electronic voting systems similar to those now found in nearby suburbs,
but it
won’t be cheap. The city’s Capital Programs Office budgeted for a $21
million
expenditure in 2001 and that may not be nearly enough. In Los Angeles,
a
projected bill of better than $98 million kept that county from
adopting
an
electronic voting system using touch-screen computers.
But low cost, partly open systems helped
reduce the cost of electronic voting systems to the point that the
relatively
poor nation of Brazil could afford to adopt an electronic voting system
nationwide. The Brazilian government chose to use a system based on
automated
bank teller technology, and in a test
vote
with 33 million voters it had a failure rate of only 0.9%. The
system is
expected to go nationwide shortly, with more than 77,000 voting stations
across the
country. Ultimately the system should cover all of Brazil’s 323,000
voting
stations. The system has
been endorsed by the International Foundation for
Election Systems and Roy G.Saltman, one of the leading experts on
voting
systems technology, and a leading
critic of most U.S. based systems.
Want to grow your own voting
system? You could do worse than to look
at the SENSUS
system, which was developed by AT&T’s Dr. Lorrie Faith Cranor
while she was still
a grad student at Washington University in St. Louis. The Sensus system
still has a number of conceptual and technical flaws, but it is as
close
to an Open Source electronic elections system
that I’ve seen anywhere. You’re invited to download
it for
testing and evaluation purposes. Id be interested to hear what you
think.
Not that it’s going to do any
good this year…
NewsForge editors read and respond to comments posted on our discussion page.
