http://www.statesman.com/news/content/news/stories/nation/10/21/21votingmachines.html

 

Researchers blast voting machine strategy

Devices are bought, placed haphazardly, Ohio professor says.

By Elliot Jaspin

WASHINGTON BUREAU

Saturday, October 21, 2006

 

WASHINGTON — A new concern is surfacing in the debate over switching to electronic voting machines.

 

Although most of the attention has focused on paper trails and how vulnerable the machines are to hackers, a professor at Ohio State University and one of his former students are questioning the haphazard way these machines are bought and distributed.

 

During the Nov. 2, 2004, election, lines for polls stretched outside buildings and down sidewalks. Some of these Columbus, Ohio, voters had to wait two hours to cast a ballot. The federal government has not set voter-to-voting machine ratios, so states and counties have had to guess.

 

Theodore Allen, who teaches industrial engineering, and Mikhail Bernshteyn have found that many voters were either unable to vote or faced long lines during the 2004 election because officials failed to accurately judge how many voting machines they needed.

 

Compounding this problem, over the past few years the federal government has given state and local governments millions of dollars to buy new voting machines, although no one seems to have any clear idea of how many machines are actually needed.

 

Because there are no national standards, each county or state decides how many machines it needs.

 

Election officials interviewed for this story all said they bought machines based on the number of registered voters in their county. But each county had a different idea of what the proper ratio of machines to voters should be.

 

In Travis County, election officials bought one machine for every 278 voters. Ohio, on the other hand, passed a law saying every county had to have at least one machine for every 175 voters. In Houston, the government bought one machine for every 250 voters, while Georgia decided it needed to buy more than 19,000 machines to ensure there was one machine for every 190 voters.

 

With the cost of each voting machine running between $2,000 and $3,000, the bill can quickly add up. Georgia, for example, paid more than $40 million for its electronic voting machines in 2003.

 

The way government buys and deploys these machines is seriously flawed, according to Allen and Bernshteyn. Allen experienced the problem firsthand in 2004 when some voters in Franklin County, Ohio, where he lives, had to wait five hours to cast a ballot.

 

The problem Allen saw in his hometown was repeated around the country during that election. Newspapers reported, for example, that in New York City some lines snaked outside polling places for three blocks, and in St. Louis the wait was 2 1/2 hours in one precinct.

 

Allen reasoned that the voting problem was similar to the problems that he and his colleagues solve for businesses, such as banks that need to know how many teller machines to install. In both cases, scientists have developed mathematical formulas to describe how waiting lines form and what can be done to prevent them.

 

Allen, who also runs Sagata software with Bernshteyn and another partner, has developed a computer program that simulates elections and analyzes the number of voting machines needed in each precinct.

 

Using his software, Allen says he found that the wait time in Franklin County could have been reduced from five hours to one hour in 2004 just by changing the location of some voting machines.

 

Allen and his colleagues also found that the length of a ballot is as important as the number of voters who are likely to use a machine.

 

"Deciding how many machines you need and where you need to put them independent of the issue about the ballot is a little sub-optimal," Allen said. "Clearly, you need more machines if the ballots are long because it has to do with how long it takes to vote."

 

In an election where voters took no longer on average than two minutes to vote, Allen said only one machine was needed for every 246 voters. But if the voting time were increased by 90 seconds, the number of machines would have to increase by 38 percent.

 

Such calculations are seldom used by election officials.

 

The way Palm Beach County decided to buy 4,900 machines is typical of how elections officials around the nation have justified their decisions.

 

Arthur Andersen, Palm Beach County election supervisor, said, "There was set in law in Florida a ratio of one to 125. But with the institution of the touch screens, that went out the window. It was left for each community to determine the ratio of machines to voters based on what the traffic would dictate. Most counties were working on a ratio of one to 200. So my staff tried to get this up to be more expeditious, and what they arrived at was one to 175."

 

Ohio state Sen. Jeff Jacobson used almost identical logic when he had the ratio of one machine for every 175 voters written into law.

 

"I don't know that anything was scientific," Jacobson said. "We are all sort of guessing around at this stuff. And we sort of eyeballed it, and we tried to eyeball it in a way that left us some wiggle room. I tried to go higher than what people thought was needed. One to 175 seemed doable, and that's how I arrived at that number and stuck it in."

 

Ohio bought more than 25,000 machines and spent $135 million to upgrade its election machinery.

 

The money to pay for these new machines is coming primarily through the federal government as part of the Help America Vote Act, passed after the 2000 presidential election debacle. Nearly one-third of the nation's voters will be using new equipment this year because of the act, according to a report by Election Data Services.

 

Michael McDonald, an assistant professor of government and politics at George Mason University, said an unintended consequence of these new machines may be to slow voting in some places as people grapple with the unfamiliar technology.

 

"The only saving grace is that we are not in a really tight presidential election," McDonald said. "We won't see turnout reach 60 percent. We are probably closer to 40 percent. It could be worse."