By Jon Stokes
11/14/2006 from www.arstechnica.com
Though you wouldn't know it from watching the election night and post-election coverage on the Cable news shows, preliminary reports from local papers and from a host of electronic voting activist groups and researchers indicate that there were widespread and significant problems with the new electronic voting machines used in the November 7th mid-terms. The many groups who're working on collecting and summarizing the information gleaned from nationwide hotlines and poll watching efforts have a ton of data to sort through, but a few of the broader outlines are clear already. In a nutshell, the kinds of problems highlighted by the two post-mortems done on Cuyahoga County, Ohio's May 2nd primaries proved exemplaryin terms of the types of problems and their relative frequenciesof what the nation as a whole faced on November 7th.
Many activist groups kept logs, like this one, of problems that cropped up on election day. Among the most publicized collections of problems is the database run by Common Cause. Common Cause logged 16,000 calls on the 1-866-MYVOTE1 hotline. The calls registered all sorts of problems, with registration problems being the most common. The biggest change from their 2004 hotline was the percentage of calls reporting mechanical failures, up from 3 percent to almost 17 percent. Poll access was also a big problem, and one that was greatly exacerbated by the mechanical failures. There were reports of people waiting in line for three and four hours, due to issues like poor machine allocation and voting machines that didn't work.
Most of the House and Senate races on Nov. 7th were very, very close, with many being decided by a few thousand (or in some cases, a few hundred) votes. With margins of victory so narrow, the kinds of problems I'll describe below are simply unacceptable, and in some instances these problems could've decided the race. (Whether they actually did decide a race or not is impossible to determine, which is the problem.)
Voting machine failures
Problems with voting machines that either wouldn't start or would stop working surfaced across the country. In many cases, these problems caused delays in the opening of polling places, and in others the poll workers were forced to fall back on paper and/or provisional ballots. Michigan, Maryland, Pennsylvania, Indiana, Ohio had some of the most widely reported problems of this type, but there were many other states that also reported machine failures.
For instance, in Cuyahoga County, Ohio, 43 out of 573 precincts reported problems getting machines started and working. This resulted in poll opening delays and the use of paper ballots. There were also reports from Cleveland of machines that stopped working in the middle of voting, and calls from other many other parts of Ohio reporting a litany of problems that range from registration and voter ID confusion to printer jams and vote flipping. You can get a quick list of some of the reports from this site. (If the state of Ohio decides to do the kind of independent post-mortem analysis of the mid-terms that Cuyahoga did of the primaries, those reports will make for even more miserable and embarrassing reading than the Cuyahoga reports.)
In Pennsylvania, polling hours were extended after some polling places opened late due to problems getting the machines started. There were also reports of machines not going through all of the election screens, machines that were not functioning at all, machines automatically shutting down early due to timing problems, and so on. Indeed, Pennsylvania was another state that saw extremely widespread problems with electronic voting machines across multiple counties. Like Ohio, it will yield an embarrassing batch of lengthy post-mortem reports if state officials opt to investigate.
In Cook County, Illinois, where I voted, there were multiple reports of problems with the touchscreen voting machines from Sequoia. The one touchscreen machine at my precinct was frozen, and the poll worker that I talked to said that other precincts in the area were reporting the same problem. The BOE was so swamped with calls, though, that they couldn't get a technician out to the site to fix the machine. All of us ended up voting with the optical scan ballots. Other e-voting activist sites are also investigating the problems in Cook County, as is the state.
At any rate, I'm not going to go on here, because you get the idea. You can find much more of the kinds of these kinds of stories by checking the links in the "Further reading" section at the end of this post. If you take a look at Election Protection 365's short, preliminary, and very incomplete (I can vouch for that) state-by-state run-down of election problems, then you what you get is a picture of a massive, distributed denial-of-service (DDoS) attack. A lot of machines in a lot of precincts across the country just did not work as advertised, and far too many voters were either forced to leave without voting or given provisional ballots.
Before moving on from the DDoS theme, let me ask you this: if you were fired up about voting in this election, and you were turned away from the polls two or three times in the morning before giving up because you had to go to work, would you feel like the mid-terms were a success? I know I wouldn't, regardless of what I thought about the outcome of the election.
I concur with computer scientist and voting activist David Dill's assessment of the prominent reports of stories that machines across the country were flipping votes on Nov. 7: this problem is a national disgrace.
The most basic thing that a voting machine should be able to do is register the proper vote for the proper candidate, but time and again voters in many states found that the ballot summary screen indicated that they'd voted for a different roster of candidates than they had actually voted for. In some instances, repeated attempts to go back and correct the problem were to no avail, and those voters' votes simply were not counted correctly.
Texas, Florida, and New Jersey were the sources of the majority of these reports, but voters in other states (Ohio, Louisiana, Maryland, South Carolina, Georgia, and others) reported instances of it.
These vote-flipping problems were so widespread and so well-documented, that statements like the following from e-voting machine vendors just boggle the mind. From a new Computerworld article on vote flipping:
E-voting machines are far more secure, accurate and auditable than the mechanical lever-operated voting machines and other systems they replaced, the [Sequoia] spokeswoman contended. She called vote-flipping concerns a "conspiracy theory from activists and bloggers."
A spokesman at e-voting equipment vendor Diebold Election Systems also said that e-voting machines don't cause vote flipping. "It's not a problem," he said. "It doesn't exist. This again falls into the 'what if' scenario."
The two leading candidates for the cause of vote-flipping reports are touchscreen calibration and user error. Most researchers reject the idea that deliberate fraud is to blame. (As I said in a previous article, if you're smart enough to compromise a machine to steal votes, then you would certainly write your vote-stealing Trojan so that it shows the voter his or her correct vote choices on the summary screen.)
Nonetheless, there are enough reports of voters who tried multiple times to cast a correct ballot, and even called poll workers over for help, that I think it's impossible that user error is to blame for all the cases. Touchscreen calibration may turn out to be the ultimate culprit, but what's needed is a full investigation into this phenomenon.
Missing votes in Florida
The Miami Herald reports that 18,382 votes either weren't cast or weren't counted in the race for House District 13, Katherine Harris's old seat. What this means is that of all of the voters who went to the polls in that district, signed in, and cast ballots, 18,382 of them either didn't vote in the House race or didn't have their votes recorded. So one out of every seven ballots cast in that race did not register a vote for the House seat; this is an extremely high number of undervotes. From the Herald:
But on Election Day on touchscreen machines, voters had a 14 percent undervote rate and 18 percent undervote rate in the early-voting week before, when complaints surfaced.
Dr. Richard Malkin said he and his wife voted early and realized their votes in the congressional race weren't recorded just before electronically casting their ballots. So they re-voted. ''The lady right next to us had the same problem. So many people said the same thing,'' said Malkin, a Democrat who sits on a SunTrust bank board with Buchanan. "It's like 2000 all over again.''
The Republican in this race, Vernon Buchanan, has declared victory on the basis of a 368-vote margin and stated that it's "time to move on." Yeah, I bet he'd like to move on and forget about the 18,000+ missing votes, but his opponent, Christine Jennings, is going to fight it out. Right now, the official plan is to conduct a "recount," but of course a legitimate recount is impossible to conduct for touchscreen machines with no paper trail, where a vote was never recorded in the first place. So the recount idea is going nowhere. There's certain to be litigation here, but there's just no way to recover the intent of the voters with these machines.
The only thing that will work here if these votes aren't somehow recovered from the machines is a new round of voting in Sarasota County.
Supervisor of Elections Kathy Dent has now done a 180 and will comply with voters' demands to scrap the electronic machines entirely and return to paper ballots. Dent had previously accused proponents of a voter-verified paper audit trail (VVPAT) of trying to obstruct the vote and "disenfranchise" voters. But in light of a newly passed referendum demanding a return to paper ballots (at least those referendum votes were counted), and in light of the embarrassingly large number of undervotes in Sarasota County, Dent has thankfully changed her tune.
Conclusions: the meltdown that wasn't?
I watched the election night coverage on the major news networks, coverage that Matt Tabbi has brilliantly (if vulgarly) lampooned at Rolling Stone, and amid all the sports analogies and talking heads I don't recall hearing much (if anything) about the electronic voting problems that I knew had gone on all day long. If such problems were referenced, it was in passing-I certainly didn't see a segment dedicated to them as we waited for the returns to come in, but perhaps I just missed that part of the coverage.
After the election, as many of use feared and expected, election officials and e-voting vendors rushed to assure the public that everything had gone smoothly, and that any reports of problems were either "conspiracy theory," "urban legend," or user error. For the most part, the media has bought this narrative. The party that came out ahead on Tuesday, many members of which were worried about electronic voting problems prior to the election, have moved on to bigger and more important issues, like who will be the new Senate majority leader.
The problems in Florida, however, threaten the "move along, nothing to see here" narrative, and you can be certain that this issue will take center stage again in 2007, as official, detailed reports of what happened last week start to surface and the 2008 presidential campaign gets under way.
By way of conclusion, I want to link this truly excellent post by VotersUnite.org, entitled "What, Exactly, IS an Election Meltdown?" The post first cites pundits and commentators stating that things went smoothly, or better than expected, and that the forecast meltdown didn't materialize. It then runs down a long litany of major problems, some of which resulted in large-scale disenfranchisement, and others that cast doubt on the outcome of critically important races. Here are just two of the items in the list:
What if lots of electronic ballot boxes (memory cards) were missing in a major city, and only 23 had been found after an extensive search, and the election director said she loses them all the time and normally no one pays any attention, but this time four local races hung in the balance? Is this "smooth" to the people whose ballots were lost in Indianapolis? And then ... what if partisan control of the United States Senate depended on one race in one state, where the reported margin of victory was three-tenths of a percent, and a recount was impossible because there was no way to recover voter intent from the electronic tallies? In what world is this "better than expected"?
Be sure and read the whole thing, and don't let this issue die. There's a whole lot to do before 2008 gets here, and relatively little time to do it.
On election day, a major piece of the mystery of "how did we let things get this bad?" became very clear to me. People that used the touchscreen voting machines, including my wife, who'd read my report and was duly skeptical of the DREs, raved about the experience. The touchscreen machines make fantastic demo units that really sell you on the idea of e-voting. So it's no wonder that states and counties across the country went gaga over these machines and just opened up their wallets when a vendor rep showed them a product demo.
This positive user appraisal accords with the exit polls from the Cuyahoga primaries, where over 90% of voters reported being pleased with the experience of voting on a DRE. The elderly were among most satisfied with the new machines.
So if you watched an e-voting documentary like "Hacking Democracy," or you've been following my coverage of e-voting, and you're wondering how all of this happened, then here's your answer: DREs are awfully pleasant to use, in spite of the fact that they're not worth much as actual voting machines. Ultimately, the story of the triumph of the touchscreen DRE is really a story of the triumph of style over substance, a story that participants in our democracy are all too familiar with.