Skip navigation
Electronic Design

E-Voting—For Want Of A Paper Trail

Did you vote in your state’s primary this year? Are you sure? Remember “hanging chads,” those sloppily punched—and therefore uncountable—ballot cards used in Florida during the 2000 election?

Concern about the reliability of electronic voting systems is old news. Electronic voting machines were around long before the Florida fiasco. But for some reason, the problem and the machines seem unfixable, which is why they’re getting so much attention in this presidential election year.

Hardly a week goes by without someone calling for banning the use of electronic or e-voting equipment, announcing plans to test them, or proposing legislation that will require the use of voter-verified paper trails along with the high-tech voting systems. The problem just won’t go away and has led to a run of mostly independent studies of the state of e-voting in the United States.

For the most part, the news isn’t good. Counties in several states have reported problems with their e-voting machines. With so much negative publicity about e-voting, a report by Common Cause and the Verified Voting Foundation says that only 15 of the 24 states holding presidential primaries or caucuses on February 5 used e-voting machines to select candidates.

The same report rated six of the 15 states that held presidential primaries on Super Tuesday as being at “high” risk for having election results affected by electronic voting machine malfunction or tampering. It also found that 17 states were at “medium” risk for having election results affected by voting machine failure, a classification given to states using voting systems that deploy paper ballots or produce a voter-verifiable paper record of each voter’s vote but do not require audits.

Another study by the Evaluation & Validation of Election-Related Equipment, Standards & Testing (EVEREST) found that Ohio’s e-voting systems have “critical security failures” that could impact the integrity of elections in that state. Funded by the federal government, the $1.9 million EVEREST study looked at three voting systems—Election Systems & Software (ES&S), Hart InterCivic, and Premier Election Solutions (formerly Diebold). What’s the problem?

As one state official put it following the publication of the study, “To put it in everyday terms, the tools needed to compromise an accurate vote count could be as simple as tampering with the paper audit trail connector or using a magnet and a personal digital assistant.”

The Brookings Institute, a centrist think tank, came up with slightly different results after looking at the same three e-voting systems. According to Brookings, voters generally prefer e-voting to paper-based alternatives, even though the Brookings study detected a 3% error rate or higher in some e-voting machines. That’s not a bad number for polling in general, but it might not work for voters, though Brookings placed some of the blame for the errors on poorly designed ballots.

Yet another study by, a Pew Center Web site that tracks election issues, noted that California, Colorado, Florida, New Mexico, and Ohio have chosen or are considering statewide paper-based optical scan systems, which were used by the highest number of voters in 2006 (Table 1).

Connecticut’s primary in February was the first in which the entire state voted on optical-scan instead of lever machines, following a national trend in the rapid decline in the use of lever machines (Table 2). Florida is now the biggest state to reject touchscreen systems, setting a July 1, 2008 deadline to remove them in favor of optical-scan systems.

Approximately 80% of the voting public’s votes in the U.S. were tabulated by computer and 38% used direct recording electronic (DRE) voting machines in some of the early primaries. Votes tabulated by computer or direct recording electronic (DRE) voting machines do not provide a “paper trail” to make recounts or audits if the electronic voting machines fail or are hacked.


More than six years and millions of dollars into a major overhaul of the U.S. election system, a number of states are contemplating returning to paper-based voting systems after failed or troubled experiments with newer voting technology. Voters in all or part of 20 states now cast ballots electronically without backup paper verification.

DREs are essentially a computer, which can work very well if properly activated and if there are no software problems—and if they aren’t hacked, an issue reported as a potential threat by some academics that have independently tested these systems. Critics of DRE equipment believe that programmers can alter the electronic record of ballots cast of these paperless systems because there is no record or link to voters for ballot verification. In some cases, errors have been reported in the counting of ballots when the numbers recorded on the cartridge printouts did not match the paper-tape backup inside the machines.

Optical scanning has its own problems, from incomplete markings that could be misinterpreted when tabulated to printers that run out of toner, resulting in cards with incomplete or unreadable marks.

“While the national enthusiasm for a return to paper is undoubtedly a reaction to lingering concerns about electronic voting, it also mirrors the same push into electronic voting we saw in the wake of HAVA,” says Doug Chapin, director of, referring to the Help America Vote Act of 2002, which required all states and localities to upgrade certain aspects of their election procedures. “What remains to be seen is whether states see paper-based voting as the right decision or just the best decision right now.”


DRE vendors have long argued that many of the errors found in the operation of their equipment have been the result of poll workers pushing the wrong buttons on their control panels.

There are several DRE manufacturers, including Sequoia Voting Systems, ES&S, Hart InterCivic, and Premier Election Solutions, which was Diebold Election Systems before changing its name after a year-long attempt to sell its e-voting subsidiary. United Technologies Corp. made a public unsolicited $3 billion bid for Diebold in early March after first approaching Diebold two years ago, a move investors viewed as an opportunity to expand its electronic security business. Diebold rejected the offer.

Sequoia Voting Systems has aggressively fought attempts to have its systems tested independently, arguing that such tests violate confidentiality agreements and risk exposing its trade secrets. Like other voting machine manufacturers, Sequoia tests or farms out the testing of its voting machines. In March, it announced that it had contracted with Wyle Laboratories to review and test voting equipment source code and accuracy for any of its equipment used in New Jersey. Wyle Labs is an Election Assistance Commission (EAC) accredited Voting System Test Lab.

Sequoia says it signed on Wyle when some of its DRE machines “appeared to show a discrepancy between the number of votes cast and the reported number for each party.” A subsequent review by the state and Sequoia determined that the data on all cartridges were correct and that all votes were accurately recorded. Sequoia said its engineers have identified the case of the discrepancy and have upgraded the source code, which is currently undergoing certification testing to eliminate the potential for future discrepancies.

Despite ongoing concerns about electronic systems in voting, nearly one-third of the nation’s registered voters will face new voting equipment in the presidential election in November. Jurisdictions with 63% of the nation’s registered voters have changed their voting systems, marking the largest shift in voting equipment in the nation’s history.

The result is that the use of DRE systems will be down in the low 30% region nationally, while voting with optically scanned paper ballots will be up over 55% nationally, according to Kimball Brace, president of Election Data Services, which conducts voting equipment studies and tracks their use. “We’ll see the use of more optical scanning in time, but the larger voting jurisdictions will still rely mainly on DRE systems,” Brace notes.

Brace says this is true, even though DRE equipment vendors are rapidly upgrading and recertifying their systems. He also says this upgrading has become an elaborate and expensive process under standards published by the Federal Election Commission in 2005 and the Elections Assistance Commission, which was born out of the enactment of HAVA.

Recertification is becoming more costly. “It’s skyrocketing,” notes Brace, to more than a million dollars per unit model in many cases. “Who’s going to pay for this?” Brace suspects that it’s only a matter of time before DRE vendors start passing along certification costs to their customers—the local, county, and state governments that purchase them.


In 2000, the Federal Voting Assistance Program (FVAP) began a pilot program to determine if absentee votes could be reliably and securely cast over the Internet.

Fewer than 100 volunteers in about 20 states were involved in the experiment, casting absentee ballots in the November 2000 election. Not knowing how the test would turn out, the volunteers were allowed to cast a traditional paper-based absentee ballot in case the Internet experiment failed.

Each volunteer received a CD with a browser plug-in that displayed and transmitted ballots to FVAP servers. Volunteers used Netscape Navigator 4.05 or higher with encryption. Encrypted ballots were sent to the FVAP server over the Internet. Voter identity was authenticated under a certification program developed by the Department of Defense (DoD). Once their vote was cast, the DoD deactivated the voters’ certification to prevent them from voting again.

The FVAP declared the experiment a success, but admitted that expanding the program for thousands of voters would require additional planning and experimentation. Emboldened by the initial effort, Congress asked the DoD to conduct a larger Internet-based voting program.

In 2001, the Pentagon began designing what it called the Secure Electronic Registration and Voting Experiment, or SERVE. Initially, the DoD planned to involve at least 100,000 people in its first SERVE experiment. If it worked, Internet voting would be available to the hundreds of thousands of military personnel stationed overseas and eventually others.

But the program was cancelled in 2004, even before its implantation, due to security concerns. As a result, most consultants and other experts in the election process believe it will be years before the Internet is used for absentee or any other kind of voting—if ever.

It is not clear if any of this is going to be resolved before the November election. A state Supreme Court judge in New Jersey agreed in March to rule on whether or not DREs are scientifically reliable, but no decision is expected before September. This means New Jersey voters will vote electronically without a paper trail in November, unless the governor or attorney general intervenes.

Several bills are being pushed through Congress aimed at tweaking the election system, with the language in some of these proposed bills specifically targeting the DRE machines. The only one that has made any progress was introduced in the House in January—HR5036, the Emergency Assistance for Secure Elections Ace, which offers $600 million to voting districts across the country that convert to paper ballots or install audit systems before the upcoming national election.

Both political parties unanimously supported HR5036 until the White House decided it didn’t particularly like the bill. It passed the house 239-178 but fell short of a required super-majority. Given those results, it’s unlikely the Senate will take up the bill this year.

Curiously, despite all of the reported problems and all the ink devoted to them by newspapers and magazines over the years, there are no documented cases of election tampering involving electronic voting machines anywhere in the U.S.

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.