Go to the mobile version of this Web site.

Login | Contact Us | Site Map | Paid archives | Electronic edition | Subscription Questions | Extras

Getting it right

By Myung Oak Kim, Rocky Mountain News (Contact)

Published January 12, 2008 at 12:30 a.m.

Text size  
Kurt Anderson, a transportation manager for Pitney Bowes, pulls a cart of mail-in ballots off a truck and into the bulk-mail area of the Valmont Station post office in Boulder this week.

Photo by Paul Aiken / The Camera

Kurt Anderson, a transportation manager for Pitney Bowes, pulls a cart of mail-in ballots off a truck and into the bulk-mail area of the Valmont Station post office in Boulder this week.

Mesa County officials Martina Pena, left, and Jennifer Manzanares test the county's touchscreen voting machines this week at the Mesa County Courthouse as Grand Junction Mayor Jim Doody watches in the background.

Photo by Christopher Tomlinson / Daily Sentinel

Mesa County officials Martina Pena, left, and Jennifer Manzanares test the county's touchscreen voting machines this week at the Mesa County Courthouse as Grand Junction Mayor Jim Doody watches in the background.

Related Stories

More Elections

More stories »

Map my news

Activists pushing for more integrity in Colorado's voting system want an election where paper ballots are counted by hand.

But hand counts have nearly twice the error rate of tallying paper ballots by machine, experts say, and there's good reason why the old-fashioned system was largely abandoned years ago.

"The human error inherent in hand counting overwhelms the mechanical problems of scanning," said Charles Stewart, chair of the political science department at MIT and an expert on election forensics. "There is a reason why businesses use adding machines. Human beings are really bad at doing tedious tasks. That's why automation is a premium in elections."

Machines aren't fail-safe either. Optical scan devices that tally paper ballots have lower error rates than hand counting, but they still can misread irregular marks or folded ballots.

The bottom line, experts say, is that no voting method is perfect.

That's why Colorado lawmakers face a daunting challenge as they attempt in the next few weeks to pull Colorado out of voting-system chaos and set a course for conducting the presidential election.

With the primary seven months away, officials across Colorado don't know how they will conduct the election because the secretary of state last month decertified many machines currently used to cast votes or count ballots.

"All these systems have various trade-offs and it's up to local election officials and the citizens they represent . . . to figure out what is the best balance of trade-offs," said Jonathan Katz, a voting-systems researcher and political science professor at the California Institute of Technology.

Experts warn that Colorado faces major risks by overhauling voting systems so close to the presidential election.

"It sounds like Colorado and some other states are just panicking needlessly," Stewart said. "The big enemy of running an election well is confusion and chaos. Any state that's changing things on the fly is inviting confusion and chaos."

Stewart and other researchers said they are seeing the most support for one voting model - paper ballots counted by optical-scan devices.

That system will be the most widely used in the nation's 2008 elections, according to Election Data Services, a Washington, D.C.-based consulting firm.

"Electronic systems are coming down now from the peak in 2006 and it's flipping over to the optical scanners," said Kimball Brace, Election Data Services president.

Opposition to electronics

The current chaos over voting systems in Colorado was triggered by a group of activists, which includes computer scientists and engineers, who are wary of government, upset about presidential election scandals since 2000 and opposed to electronic-voting equipment with no audit trail and prone to hacking.

The group sued former Secretary of State Gigi Dennis in 2006 to stop the use of electronic-voting systems. A judge determined that Dennis, a Republican, violated law by not adequately checking whether electronic systems were secure and ordered her to recertify machines.

On Dec. 17, Secretary of State Mike Coffman, a Republican, announced the results of that review. He said thousands of machines used in all but 12 counties were too flawed for use in the 2008 elections. He decertified optical- scan equipment used in 49 counties and electronic-voting machines used in six counties, including Denver, Arapahoe and Jefferson.

That announcement angered county clerks who say the electronic-voting equipment works fine and has passed numerous tests before and after elections.

Clerks and equipment manufacturers are preparing to appeal Coffman's decisions and will try to get their machines approved for use this year. Coffman is asking the legislature to help him move through the appeals process more quickly.

Meanwhile, lawmakers will consider what voting model they want.

Momentum is pointing toward some type of paper system.

"The election in 2008 is going to be primarily done on paper," predicted Senate Majority Leader Ken Gordon, D-Denver, who is drafting legislation on the issue.

Hand counting unreliable

Two Colorado counties - San Juan and Jackson - hand count ballots in certain elections, according to Coffman's office.

About 470 people usually cast ballots in San Juan County, said clerk Dorothy Zanoni. She said four election judges hand counted ballots in 2004 and completed the tally by 10 p.m. on election night.

In recent elections, she's used optical-scan machines to count ballots. Those machines were decertified by Coffman.

Zanoni said she's frustrated by the current turmoil.

"Nobody trusts us anymore," she said.

She said she's going to hand count ballots this year to avoid the certification problems.

"I'm allowed to do that. It's still legal," she said.

Zanoni said hand counting has worked fine for her county, but it wouldn't work in large counties.

"It's too much stress. You have to have too many judges and we all have to have CBI background checks."

El Paso County will conduct a hand-count test in the next few weeks, said clerk Bob Balink.

Denver Elections Director Michael Scarpello, who is trying to plan for elections with an expected turnout of 250,000 voters, said he's seen the errors of hand counting while running elections in Nebraska.

"Hand counting is the absolute most unreliable method possible," Scarpello said. "I've had my top people hand counting in Omaha. I don't think I've ever seen a hand count that did not have mistakes in it."

Two researchers - Stephen Ansolabehere of MIT and Andrew Reeves of Harvard - studied recounts of New Hampshire elections between 1988 and 2000. In their 2004 study, they found that optical-scan machines made at least a third fewer errors in recounts than hand counting.

"Historically there is about a 1 percent difference between initial counts and recounts when ballots are tabulated by hand," the study said. "The discrepancy between initial counts and recounts falls to about 0.5 percent with the optically scanned ballots."

Hand counting "is extremely problematic and it's not practical," said Ted Selker, an MIT computer scientist and co-director of the Caltech/MIT Voting Technology Project. "These people that think that hand counting is a panacea - it's wishful thinking."

With optical scanners, those used at neighborhood polling places are more dependable in counting than ones used in central offices - as is the case with mail and absentee ballots, studies show.

That's because precinct scanners will spit out a mismarked ballot back to the voter, and the voter can re-do the ballot. In central locations, voters are not present when ballots are scanned.

Experts say optical scanners can misread folded ballots, stray marks or get jammed.

"Machines can't figure out voter intent compared to humans," Katz said.

Hand counts defended

Al Kolwicz, a voting integrity activist leader, acknowledges the human error problem with hand counting.

Many hand counts in this country are done by election judges who call out a candidate chosen on a ballot and someone else who makes a stick mark on a board, he said.

He said a different hand-counting method, called "sort and stack," used in Switzerland but not in the U.S., is just as accurate as machines.

That system involves a separate ballot for each race. The ballots are color coded by race, stacked according to the voter's choice and then counted.

But even the stick-marking system is preferable to machines whose tallies cannot be independently checked, Kolwicz said.

Activists "like that better than using non-verifiable machines," he said.

Kolwicz and other activists said election officials aren't paying enough attention to making the process open and auditable in the current debate over voting systems.

The activists want security and accuracy. But they also want to be able to trust the election system. And machines - including optical scanners as they are designed today - don't allow for the public to check operations and tallies, Kolwicz said.

"If you come out to play baseball and I come out to play hockey - we get on the field and we don't agree what the game is, it's going to be a mess, and that's what's happening," he said.

Selker, the MIT scientist, said the debate has failed to focus on other, more important aspects of elections - ballot design, voter registration checks and polling-place operations. He said those issues create far more errors in election counts than whether you hand count paper ballots or vote on machines.

"Nobody's focusing on them," Selker said.

kimm@RockyMountainNews.com or 303-954-2361

Comments

  • January 14, 2008

    12:39 p.m.

    Suggest removal

    Scott writes:

    Gad! This is getting ridiculous! All of the counties should use a system similar to Adams County. You have a paper ballot that you fill in the circle, the ballot gets scanned and your vote is counted. If someone feels that the counting machine has miscounted, then there is a paper trail to perform a hand count.

    Scott