Rage Against the Machine

Rage Against the Machine

by Ronnie Dugger

http://www.afd-online.org

Post date 11.22.00 | Issue date 12.04.00

The Bush campaign has made a lot of questionable assertions in
the last two weeks: about the authority of the Florida
secretary of state, about the propriety of hand counts, about
the need for \"finality.\" But one claim that probably struck
most people as incontrovertible was James Baker\'s statement
that \"[m]achines are neither Republicans nor Democrats and
therefore can be neither consciously nor unconsciously biased.\"

Literally speaking, of course, he\'s right: Machines don\'t have
political beliefs. But computer programmers do. Just as a
dishonest and determined vote inspector can claim to see a
hanging chad where none exists, so can a dishonest and
determined vote-counting-machine programmer add votes to a
candidate\'s total, transfer votes from one candidate to
another, or determine an outcome with a specified percentage
spread.

Voting machines, in other words, can be as \"biased\" as their
human masters want them to be. And when they are it\'s a lot
harder to detect. To steal votes in the South Florida hand
recounts, you\'d have to escape the eyes of an army of
bipartisan observers and reporters. Stealing votes by
reprogramming the voting machines, by contrast, would take
place deep within the zeros and ones of computer code,
invisible to public officials or suspicious candidates.
Indeed, even the software\'s own designers admit that without
security checks on the process and a hand count as a fallback,
we\'d have no way to know whether such fraud had taken place.

Although Baker has called ballot computers \"precision
machines,\" there are plenty of documented cases of machine
errors. In Missoula County, Montana, in 1968, machines counted
ballots cast for Richard Nixon as votes for Hubert Humphrey,
and vice versa. In 1978, a candidate for comptroller in
Illinois successfully reversed the outcome of an election in
one county because the computer had \"flipped\" the results.
Computers in Orange County, California, in 1980 gave about
15,000 Democratic-primary votes for Jimmy Carter or Edward
Kennedy to Lyndon LaRouche or Jerry Brown. That same year, in
San Antonio, computers were unable to count more than 9,000
votes for any office on the ballot, depriving Ronald Reagan
and other candidates of votes.

Most of these problems resulted from faulty equipment or
programming errors.

But fraud can occur, too. The nation\'s more than 10,000 voting
jurisdictions buy their computer equipment from privately owned
companies. The companies write the basic \"source code\" on which
the machines operate, but those codes must then be customized
for each jurisdiction. Sometimes the companies do this
customizing themselves. Often, though, local election districts
bring in their own outside computer specialists to do the work
--creating additional opportunity for foul play.

In fact, it would be relatively easy for somebody with
programming knowledge to instruct a computer to add extra votes
or to discount some--and to do so in a way that would escape
notice. \"There are at least a half-dozen places, maybe a few
less, where you could lay in a Trojan horse [for instance, a
secret vote-counting control algorithm] in that source code
--lay in routines to do whatever you wanted to in an election,\"
Wayne Nunn, a computer specialist who examined the source code
in a contested West Virginia election, once told me.
(Then, to prove his point, he sat down at his computer while we
were talking and, within ten minutes, had rewritten code so the
computer made two and two add up to five.) \"I have no
question,\" says Randall H. Erben, President Reagan\'s special
counsel for ballot integrity, \"that somebody who\'s smart enough
with a computer could probably rig it to mistabulate.... It\'s
going to be virtually undetectable if it\'s done correctly.\"

Election officials insist that they run \"logic and accuracy
tests\" to prevent tampering. But these tests are often woefully
inadequate. Dr. Peter Neumann, a senior computer specialist at
SRI International in Menlo Park, California, says a skilled
user of a conventional computer system can \"do almost anything
he wants and leave no trace.... You effectively have to trust
the entire staff of the corporation that is producing your
software. Every single member has to be trusted.\" Because of
poor security controls in many election districts, Neumann
says, it is \"relatively easy\" to start the counting with, say,
3,000 votes for one candidate and zero for the other, even
though the counters would indicate zero for both. \"I program
it,\" he explains, \"so that, after the test is run ... it
simply adds thousands of votes.\"

And there\'s another impediment to safeguarding software against
manipulation: The election companies consider their codes
private property and are willing to go to court, citing
intellectual property laws, to keep candidates or public
officials from looking at it. This is precisely what happened
in--you guessed it--Florida twelve years ago, when Democrat
Buddy MacKay lost his Senate bid to Republican Connie Mack by
a razor-thin margin after a recount. Immediately after the
election, suspicious of \"a problem in the software,\" MacKay
asked to examine the vote-counting computer codes in three of
the very same predominantly Democratic counties at the center
of the current battle. But those counties (and two others)
refused, saying the codes were the private property of the
vote-counting companies involved. Discouraged, MacKay gave up.

\"My understanding is that it\'s possible to program a computer
for it to count wrong for a while and then straighten itself
out,\" MacKay says. \"Now that I realize the level of
sophistication, I realize how easy it would be for somebody
to, in effect, rewire that program for a brief period of time
and then have it straighten itself out. I don\'t know how anybody
could prove that.\"

None of this means Mack or his supporters stole the election
through software sabotage--any more than it means somebody in
the Bush camp (or, for that matter, the Gore camp) stole votes
this year. The point is less that fraud has occurred than that
it could--and we might never know. \"Of course, it would have to
be a national election,\" mused David Dunbar, former president
of Computer Election Systems, the company that built the
Votomatic punch-card machines on which the ballots of 37
percent of American voters are still counted. \"You\'d have to
concentrate on a few states and counties and precincts that
you\'d need. It isn\'t a matter of millions of votes. You\'d need
ten votes in this precinct, 50 there, 100 there.... You\'d have
to be careful about the politics and the precincts. You\'d have
to rifle in on the ones that were crucial. From there on it
would be easy.\"

Dunbar told me he has lost sleep worrying that someone might
bribe his programmers. So should the rest of us. If the Florida
fiasco is going to prompt a national reexamination of the way
America runs elections, that reexamination should start with a
recognition that the machines that count our votes are not
today--and probably never will be--completely safe. It\'s just
one more reason that counting ballots the old-fashioned way,
by hand, is still the best guarantee that the voters\' will is
observed. No matter what James Baker says.

RONNIE DUGGER is completing a book on the history of voting.


\"The only true aging is the erosion of one\'s ideals.
They\'re people of increasingly low expectations.
That\'s the definition of a frightened liberal.\"
- Ralph Nader

http://www.greens.org - Globalize Democracy.

http://www.votenader.org - Run with Ralph.

http://www.bfi.org - Utopia or Oblivion?

http://www.tompaine.com - Common Sense.

http://www.futurenet.org - Yes! Positive Futures.

http://www.indymedia.org - Independent Media Center.

http://www.transaction.net - What is Money?

http://www.grb.net - Global Currency for the 21st Century.

http://www.cinetopia.net - The Power of Light.

http://www.freespeech.org - Share Your Mind.