With less than a month before Super Tuesday, every vote counts. But will every vote actually be counted? One-by-one, states across the country are finding critical flaws in the accuracy and security of electronic voting machines. We speak with Clive Thompson, the author of a New York Times Magazine cover story titled "Can You Count on Voting Machines?" [includes rush transcript]
This is a rush transcript. Copy may not be in its final form.
JUAN GONZALEZ: The 2008 presidential race is in full swing, and political analysts say it’s shaping up to be the most open contest in eighty years. Wins by Senators Hillary Clinton and John McCain in New Hampshire have set the stage for a toe-to-to battle for the parties’ nominations. With less than a month before Super Tuesday, every vote counts.
But will every vote actually be counted? One by one, states across the country are finding critical flaws in the accuracy and security of electronic voting machines. Last spring, California and Florida decided to get rid of their electronic voting machines. In December, Colorado decertified about half of its touch-screen devices. In Ohio, Secretary of State Jennifer Brunner undertook an extensive review of electronic voting machines that concluded their use “may jeopardize the integrity of the voting process.” On Capitol Hill, two senators have sponsored a bill that would ban the use of touch-screen machines across the country by 2012.
AMY GOODMAN: Last week, the cover story of the New York Times Magazine was called “Can You Count on Voting Machines?” The nearly 8,000-word article takes an in-depth look at electronic voting machines in states across the country.
Clive Thompson is the contributing writer to the New York Times Magazine who wrote the piece, joining us in our firehouse studio. Welcome to Democracy Now!, Clive.
CLIVE THOMPSON: Good to be here.
AMY GOODMAN: Begin where you began your piece, these stories, these anecdotes around the country.
CLIVE THOMPSON: Sure. Well, what I was looking at was the performance of touch-screen machines in a bunch of key states — Ohio, Florida, swing states — and what I discovered when I went and observed them in action is that they have a well-known propensity to sort of, about 10% of the time, have some sort of a failure rate. So when I was in Cuyahoga County the day of the election —-
AMY GOODMAN: Ohio.
CLIVE THOMPSON: In Ohio, that’s right, Cleveland. They have a touch-screen machine, and it has a little paper attachment, sort of -—
AMY GOODMAN: Explain what a touch-screen machine is.
CLIVE THOMPSON: A touch-screen machine is basically — you walk up to it, and it’s almost as if there’s a large flat screen, just like a computer screen. When you vote, you touch the screen.
AMY GOODMAN: Like an ATM.
CLIVE THOMPSON: Like an ATM, yeah, like an ATM machine. So you’ll see the names, you’ll see the candidates like that, and you’ll touch the screen, and then it’s supposed to be recorded on the computer chips inside. Some of the machines also have a paper trail attached. They’re sort of almost like a — imagine like a cash register roll. You can’t actually take it with you, but what you’re supposed to do is, after you vote, you look down at the paper, and it sort of prints up what your vote is, so you can verify with your own eyes that that’s — it’s recording your vote correctly.
In the case of Ohio, interestingly, that paper trail is legally the vote. What’s on the computer chip is just a convenient way to tally the vote quickly. But in the event of a recount, they go back and they uncork the paper, and they recount the paper trail. But the interesting thing in Cleveland is that with these machines, about 20% of the time, the paper trail jams and doesn’t print for some reason, maybe only one or two of the votes, maybe all of the votes. And they’ve had this problem over and over again. In November, they had about ten very close elections. And when they went back and pulled the paper out to do a recount, they discovered that 20% of the time there was at least one or maybe many errors with the paper trail. So, effectively, the votes are gone on paper, and you have to trust what’s on the computer chip.
JUAN GONZALEZ: Well, I think that, to me, was one of the most astonishing parts of your piece, because it’s almost like the second generation of defenders of computerized voting have said, well, if you produce a paper trail, then at least that guarantees that the original computer vote or the count from it, you’ve got that backup. But now your article seems to indicate that there are huge problems. I know New York state was, I think, adopting a situation of computer voting with paper backup. So this seems to indicate that even the, quote, "fixes" of the first problems are now creating additional problems, as well.
CLIVE THOMPSON: So far, yeah. I mean, definitely it would be possible to produce a machine that had a much more robust, high-quality paper attachment. The problem seems to be, with this one, is that, from people that have looked at it and had to fix it — I talked to some — I talked to some technicians who were trying to fix them on the —- they were broken on the day of the Cleveland election -— that they’re just simply not high enough quality. I mean, it’s something — from what I could see, it looks like something that’s on the par of what you’d see at a local, you know, corner store cash register, basically. You could certainly make one that was much better quality, but they haven’t been made that way.
AMY GOODMAN: Sarasota, 2006?
CLIVE THOMPSON: Now, in Sarasota, you had a very interesting situation. So you’re down in Florida. It’s a fight between Vern Buchanan, a Republican, Christine Jennings, a Democrat, to replace Katherine Harris, a very famous, high-profile race.
AMY GOODMAN: Congressman.
CLIVE THOMPSON: Congressman, that’s right. And so, it was a very, very closely fought race. It came to under 400 votes, which is incredibly tight.
But when they did the recount, they found that there was an 18,000 undervote. Now, what that means, an undervote is when someone goes to the polls and there’s twenty or thirty races, and they vote in every race except for one. That’s an undervote. And that’s what happened in this case. People voted in every other race — 18,000 people voted in every other race, except for that race. Statistically, it’s completely off the scale. In a normal undervote, it’s about 3% or 4% . This was an undervote of about 13% .
So the question became: what caused that 10% difference? And Christine Jennings, who lost, has been mounting a case for two years, arguing that it was a machine malfunction, that people intended to vote for her, and the machines, for various reasons — maybe the screen was out of whack; maybe something went wrong with the software; maybe it’s a bug that hasn’t been disclosed, hasn’t been found — it did not record votes for her, and that she lost because of the machine. She’s still fighting that.
JUAN GONZALEZ: And there has been no resolution of what potentially could have caused the problem?
CLIVE THOMPSON: No. And part of the problem is — and this is where things get really interesting — is that it’s often very hard, because they’re computers, to figure out what went wrong or if anything wrong, because you’ll have voters that will walk out of the voting booth, and they’ll say things like, “I saw my vote flip in front of my eyes,” like “I touched one candidate’s name, and I saw another candidate’s name an inch below get highlighted.” But the problem is, because you can’t be there with them, you don’t know whether it was their mistake or the machine’s mistake, right?
So, similarly, when you have an unusual result, a very unusual result like this, an incredibly high undervote, you can never really figure out what happened, because you can’t — you can never recreate those situations. And everyone who knows computers knows a computer code, when it gets really long and really complex, is just going to have a certain amount of bugs that need to be fixed and that can unpredictably cause problems. Computers cause problems unpredictably and of unpredictable scale. They work fine 99% of the time, but the 1% of the time they stop working, you don’t know whether it’s going to be a tiny problem or a big problem.
JUAN GONZALEZ: Now, you mentioned also the one case of a small town where no votes appeared for the candidate for mayor, but he was sure that he, at least, had voted for himself.
CLIVE THOMPSON: Yes, exactly. Eighty-person town. The mayoral candidate went in and voted, and there was no votes recorded for him. He was pretty sure that he had — he claims that eight or nine friends should have voted for him also.
AMY GOODMAN: This is in Waldenburg, Arkansas?
CLIVE THOMPSON: Yes, that’s right. But again, you see, this is the interesting thing, is was it his and his friends’ mistake or was it the machine’s mistake? You can never really know. So the real problem with the machines is that they engender an enormous lack of trust. The losers have a lot of reasons to never accept the fact that they lost, because they can always point to a potential computer bug. They can’t prove it, and their opponents cannot prove that there wasn’t a bug. So you wind up in this gray zone, which is almost like custom-made for legal battles.
So what people worry about with the next election is that you’re going to have a battleground state, where there’s a closely fought presidential race that tips the election. The odds are on Pennsylvania, because in Pennsylvania you have about one-third — a little more than one-third of the people voting on the exact same machine that caused the problems in Sarasota.
AMY GOODMAN: And that machine is?
CLIVE THOMPSON: The iVotronic by ES&S. It has no paper trail. It just has the chips. You have to trust what’s on the chips. And it has this record of being involved in a controversial race. So, essentially, the scenario that scares people is this: the presidential race comes down to a few thousand votes in Pennsylvania, twenty-one Electoral College seats. It’s a pretty big chunk of votes. And because it comes down to a few thousand votes, there’s an automatic recount. And the loser, be they Republican or Democratic, discovers, as they probably will, that 10% of the machines displayed some form of aberrant behavior — crashes, glitches, complaints by voters that things flipped. So that will give them enough of a basis to just start a legal fight, saying, you know, I don’t trust the results of this election. And that could wind up back at the Supreme Court.
JUAN GONZALEZ: Well, what about the whole issue of hacking into computers held by counties that actually count the vote? That was raised numerous times by citizen groups who have been following this and raising questions about the past two elections.
CLIVE THOMPSON: It’s true. For the first long while — I guess around, you could say, the early 2000s — all the concerns about election machines were about their vulnerabilities, about hacking. And it is true that every time a computer scientist has been able to examine the machines, which is difficult, because they’re considered a trade secret — the guts are considered a trade secret. You’re not just allowed to ask for one and look at it. But in a few rare situations — either the code has been leaked or a state has decided to subject the machines to a review — they’ve discovered a lot of vulnerabilities, ways of introducing viruses that would swap votes, ways of hacking the machines just by guessing fairly easy-to-guess passwords on the servers that tally votes.
So the fact of the matter is, yes, there’s a lot of evidence that you could hack and throw a vote if you wanted to. But the truth is, when you actually talk to the computer scientists who have studied this stuff and ask them, “Do you think anyone has ever done it?” they say no. The likelihood is pretty small. What they’re much more concerned about — and this is where it’s interesting — is the everyday malfunctions that appear to be going on regularly 10% of the time. That’s what scares them right now.
JUAN GONZALEZ: I’d like to ask you also about the human factor. Obviously, those of us who have been voting for years are familiar very much with going into a voting booth, and the people who are working there are inevitably elderly people who have been doing it for years, because most local governments don’t pay very much to their Election Day volunteers. So the — and the folks that are working these polls very often are people who — for whom the entire computer age revolution is something that they just haven’t gotten into. And the question about the human factor in all this?
CLIVE THOMPSON: Yeah, the human factor is a big deal, because, you’ve got to understand, running an election is like setting up an army that you deploy only a few days a year. And the volunteers have incredibly complex jobs. They have to be interpreting elections law, determining who can vote and who can’t, in addition to, you know, running the machines. And every time the machines are being deployed, the first couple of elections have always had a lot of problems, because — again, you’re precisely right — you’ve got a lot of senior citizens who have been working diligently for decades running elections, but they did it with paper machines, paper that they understood very well. Here, along came machines that, even when they work perfectly, had a certain amount of errors that were a little baffling.
In Cuyahoga County, they had an absolute catastrophe the first time they ran it in the 2006 primaries. The polls were open late into the evening. They lost dozens and dozens of the memory cards that had the votes on them. It took well into the next day to count them. They lost all sorts of equipment, security equipment, like stuff just went missing. So it was — it’s very, very hard to train people to run an election. It’s even harder when you have very, very complex machinery.
AMY GOODMAN: Clive Thompson, explain now how states are dealing with this. I mean, it’s amazing it doesn’t get a lot of attention, as state by state says they’re getting rid of their Diebolds or whatever the machines they have are. What are we going to see in 2008?
JUAN GONZALEZ: Well, Diebold no longer even exists, right? They changed their name.
CLIVE THOMPSON: They changed their name to Premier Election Solutions, that’s right. Now, the — well, Diebold, the company, exists. The division that makes the voting machines is now called Premier Election Systems.
What’s going to happen in the next year, it’s a really interesting question, because the fact of the matter is, over the last few years, all the concerns that were originally raised by citizens’ groups and by computer scientists that were originally sort of dismissed out of hand, i.e. that these machines are hackable or that they have unacceptable failure rates, those have risen to the point where now secretaries of state accept them, and they openly talk about stuff that I only heard computer geeks saying eight years ago, like the fact that they think that the privatization of voting machinery is wrong and that the systems ought to be open, that anyone can look at the software. This is absolutely remarkable. I’ve never seen a transition like this.
And so, what’s happening is a bunch of states, one by one, have decided to get rid of the machines. The high-profile ones are Florida — after that fiasco in Sarasota, they just said, “Forget it. All the big counties that have these touch-screen machines, you’re going back to paper.”
AMY GOODMAN: So, you’re going back to chads?
CLIVE THOMPSON: Well, no, actually. They’re going back to a paper system that’s considered much more reliable. The problem with the hanging chads is that you were punching holes in paper, and that’s actually a very weird thing to do. We don’t do that on a day-to-day basis, so we don’t know how to do it. And people would half-punch a hole, and that was the hanging chad. You get an argument over did they intend to vote or not.
What they vote on now is something much simpler. It’s a paper ballot with — just like if you did a test in college, where you’re filling in dots. And anyone knows how to do that. And then you scan them. So you get the best of human ability with computer ability. The humans know how to do this very accurately, and the computers know how to count them very quickly. And if you have a recount, you just pull the paper out and look at the paper. And traditionally, there’s not a whole lot of argument over voter intent in those situations.
AMY GOODMAN: So that’s what Florida is doing?
CLIVE THOMPSON: That’s what Florida is going back to, and that’s what a lot of the states want to go back to. Other ones include: California decertified a whole bunch of machines; Colorado, just last month, decided that it was going to decertify a bunch of machines; and several other states are looking at it also. So the dominoes are sort of toppling.
JUAN GONZALEZ: But this whole issue of, as you alluded to, the privatization of this process, for many people, it’s just astounding that such a central part of our political process has been increasingly sort of turned over to these private firms to be able to do the — not only the production of the machinery, but then actually the servicing of it, and so forth.
CLIVE THOMPSON: That’s right.
JUAN GONZALEZ: Do you see any indication that that’s going to change, that government is going to bring a lot of these processes back within government itself?
CLIVE THOMPSON: Well, you know, that’s a really good question. There’s sort of two parts of it. One is, do they want — are they worried about having lost control of this? And I think a lot of secretaries of state are, but some aren’t. I mean, this is a — the thing about elections is that it’s a completely local affair, right? So you’ve got some counties and some states that are like, “This is fine,” other ones that are very concerned that what they’ve — what’s happened with touch-screen machines is that they’ve bought machines that are of such high complexity that no one at the county level or the state level fully understands it. When things are broken, they have to hire the vendor who sold the machines to come in and fix it and have to trust them to deal with them honestly.
So the counties that do want to regain control can do so to a certain extent by going back to this paper-based balloting, simply because the recounts, because they’re done just with looking at paper, they have a lot more control over. And the machines that scan them, they are computers, so they require, you know, some complexity, but they’re just much, much less complex than touch-screen machines. So they can establish a fair amount more control over that.
The interesting question is: can this be done in time for the 2008 election? Probably not, because the fact of the matter is, it takes a long time, eighteen months at a minimum, to make a healthy switch over to a new system. There isn’t enough time for all the states that want to switch over to do so. So by the 2008 election, presidential election, probably one-third of the electorate will still be voting on touch-screen machines.
AMY GOODMAN: Do you know about what these voting machines were, used in New Hampshire, LHS, what its record is — LHS having exclusive programming contracts for all New Hampshire voting machines, and its record?
CLIVE THOMPSON: I confess, I haven’t done much research into that. I’ve read a little bit about it. It’s paper-based balloting, so in the event of recounts, they do have paper to look at. The concern tends to be, in these situations, is the automatic counting on the scanning being futzed or hacked in some sort of way. It is possible with those scanning machines for there to be errors. It’s possible for them to be hacked. But the point is, if you demand a legal recount, you go back to the paper and look at the paper.
AMY GOODMAN: The bill in the Senate that would ban all electronic voting machines by 2012?
CLIVE THOMPSON: Yeah, the Nelson bill. That’s currently moving along. It’s slow. It’s very controversial, because Nelson and several congressmen, senators, would like to get rid of all these machines. They’re worried by the error rate. They’re worried about the hackability. But it’s very expensive. It cost the federal government $4 billion to put all these touch-screen machines in place. And a lot of the counties do not want to have to be forced to go back to something new. And there’s a question as to whether or not the Nelson bill will pay for it. If there’s no money mandated for it, most of those counties will probably stick with it, and there will be big fights over whether or not the federal government can force them to get rid of their machines.
AMY GOODMAN: Clive Thompson, thanks very much for joining us and writing the piece. It’s called “Can You Count on Voting Machines?” It was the cover story of the New York Times Magazine this past Sunday.