Voting Process too Important to Leave to Technology
By Andrew Kantor, USATODAY.com columnist
You can't trust technology, but somehow we always do.
Many objects technological have become background noise - literally or
figuratively. You don't think about it unless it breaks - there's no
dial tone, or the heat doesn't come up, or the engine explodes. We
expect things to work. Most of the time they do.

It's not a matter of how old something is. Powered flight's been
around for 100 years and I'm still sure the wings are going to come
off the MD-88 I'm on.

But the modern Internet is fewer than 10 years old and I always expect
my e-mail to arrive in seconds. If you use a spreadsheet and put
"=2+2" in a box, you expect to see a "4" appear, George Orwell
notwithstanding. But there's a danger to treating any gizmo like an
unfailing "black box." There are always human beings involved, and
human beings make mistakes. Or worse.

Last month, we - well, some of us - voted. Depending on where you
live, you may have stuck a piece of paper in box, or thrown a little
mechanical lever, or punched a hole in a card. Or pushed a button -
beep! John Smith gets your vote for school board president.

Or does he?

Electronic voting machines, it turns out, may or may not be counting
your votes properly, if at all.

Detractors - and there are more and more of them - call it "black box
voting." You assume the machine's software is counting the votes
correctly, but there's no way to know. But the government must have
tested these machines before entrusting our very democracy with them,
right?

Maybe. Maybe not.

With black box voting systems, the machine records each vote onto its
internal memory via software. And software can be hacked. Coding it to
switch every 50th vote from Smith to Jones would be trivial.

Can't happen, you say? There's that trust in technology I mentioned.
It can happen. Someone broke into the computers of Diebold, one of the
largest makers of electronic voting machines, and downloaded hundreds
of staff memos regarding the company's voting systems.

They're a scary read - software bugs, faked demos to governments,
discussions of how easy it is to break into the machines' databases
that store the votes. (The memos have since spread far and wide
onilne. A search on "Diebold memos" will find them.)

OK, you say, so the software had bugs. That doesn't mean there was any
malice involved, or that anything actually went wrong.

Would that it were the end of it. But it's not. First, there was
Diebold's CEO, one Walden O'Dell, who told the Cleveland Plain Dealer
in Augustthat he was "committed to helping Ohio deliver its electoral
votes to the President next year." Coming from the head of a voting
machine company, that's scary.

OK, you say, that was a stupid thing to say. But only a conspiracy
theorist would believe it's more than hyperbole from an overzealous
exec. There's no indication the machines don't work.

Unless you're in, say, Fairfax, Va., where the county's new e-voting
machines (made by Advanced Voting Solutions, not Diebold) apparently
subtracted about one out of every hundred votes for Rita Thompson,
Republican candidate for school board. She lost by fewer than 1,700
votes.

Oops.

Or in Boone County, Ind., where the software showed 144,000 votes
cast. Trouble was, there are only about 19,000 registered voters.

Or Alameda, Kern, or Plumas counties in California - which do use
Diebold machines - where the e-voting systems reported, somehow, that
every single voter cast a ballot for the recall election; that is, no
one abstained. In every other county, between one-half and 9% of
voters skipped the recall question, but the Diebold machines in these
three counties showed 100% participation. That means either the
machines discarded thousands of votes (those who abstained) or cast a
vote for them. Which do you think is better?

A true cynic (good for you!) might say that we also trust the folks
who make and use the mechanical voting systems. But mechanical systems
offer two things an e-voting machine doesn't. First there's the clear
feedback to the voter - a piece of paper or a resounding 'click' -
that tells you your vote's been cast. I bet the folks in Fairfax would
have appreciated that. Second, it's harder to "hack" a mechanical
voting system. Anyone can look inside see how it works: Here are the
paper ballots, here is where the tape is punched. A lot of people have
sufficient mechanical aptitude to verify the workings. Not so with
software.

Further, it's impossible to get such a system to shift its votes just
a little bit. You could make one cast every vote for Jones or for
Smith, but that would be obvious. Tricking it into switching, say, one
out of every 50 Smith votes into a Jones vote would be darned near
impossible.

There have been calls - loud calls, in some cases - for
"voter-verifiable paper ballots" from black-box machines: something
that says "I voted for Smith." If you vote for Smith but your receipt
says you voted for Jones (or that you didn't vote at all), you can
complain and have something to back you up.

The Electronic Frontier Foundation is working to have this kind of
machine be mandatory. But for now it's not. So the next time your
expensive piece of software crashes - or does something unexpected -
think about how you'll be casting your ballot in 2004.

Beep.

Andrew Kantor is a technology writer, pundit, and know-it-all living
in Columbus, Ohio; he's also a former editor for PC Magazine and
Internet World. Read more of his work at kantor.com.