The three-month study of Cuyahoga County's May primary election, done by the San Francisco-based Election Science Institute, found vote counts in the machines' electronic memory in many cases did not match results on the paper-receipt backups, often by 25 votes or more. ESI says such discrepancies could be disastrous in a close election.

Utah's Diebold Elections Systems machines use the same technology as Cuyahoga County, including printing a backup to a paper spool similar to a cash register tape. That paper backup was the source of most of Cuyahoga's problems when it was torn or failed to print.

ESI's report this month warned that the election system in Cuyahoga, which includes Cleveland, "exhibits shortcomings with extremely serious consequences, especially in the event of a close election. These shortcomings merit urgent attention. Relying on this system in its present state should be viewed as a calculated risk."

"What we've really found is errors in how all this stuff works together in the hands of regular election people," said ESI founder Steven Hertzberg. "There are some things that are not matching. That's disconcerting. We don't know yet exactly why that is occurring."

Utah election officials say while Cuyahoga County and Utah use the same technology, Utah's training and election procedures - which ESI found key to accurate voting and recounts - are superior.

"It's important to note that Cuyahoga County has been screwing up elections for decades, and this is nothing different," said Joe Demma, Lt. Gov. Gary Herbert's chief of staff. "Here, we are doing all we can to train folks and support the counties to train them."

Cuyahoga Elections Board Director Michael Vu said it is misleading to compare Cuyahoga, which has more than 1 million voters, to any Utah county. There are 1.28 million voters in Utah.

"You have a lot of factors in Cuyahoga County that you don't have in Utah," says Vu, who was Salt Lake County's elections manager before moving to Ohio. "In urban counties, such as Chicago, Cuyahoga County, St. Louis, and cities in California, this is a huge social-change management issue. Those urban counties have all had problems."

Vu's old boss, Salt Lake County Clerk Sherrie Swensen, says her office observed Ohio's primary to prepare for Utah's June 27 primary. "From seeing the problems they had, we may have even over-prepared," Swensen says of Salt Lake County's intensified training and procedures.

Davis County Clerk Steve Rawlings conducted an audit shortly after his primary. The ballot-by-ballot paper results from a sample of machines paralleled electronic tallies. "They matched perfectly," he said. Utah County also checked its electronic vote with the same flawless result.

"The [Cuyahoga] errors seemed to be related to their poll workers," says Rawlings, who has studied the ESI report. "There were a bunch of inconsistencies in the way they set up their policies and procedures that allowed the counts not to match up."

A committee of election officials and voting experts appointed by Herbert to develop statewide recounts and audit procedures will begin meeting Sept. 6. ESI pulled together a team of 17 researchers, including representatives from the University of Utah Policy Center and Brigham Young University's Center for the Study of Elections and Democracy.

The researchers found that almost a third of Cuyahoga's polling place "incidents" were related to voter registration, such as misspelled names and incorrect addresses. About 16 percent of the incidents involved the machines.

"The key to making everything work well is having good procedures," says the Policy Center's Thad Hall, who worked on the study. "That's one of the things you see lacking in Cuyahoga County that is not lacking here. [Utah's] counties have put together better procedures."

The most troubling errors involved the printed backups, ESI said, because Ohio's election law - and Utah's - designate the paper backup as the "official" ballot. In the manual audit, ESI could not find paper evidence of some of the electronic ballots - nearly 10 percent of the tapes were "either destroyed, blank, illegible, missing, taped together or otherwise compromised."

Much of the data was lost because election workers failed to load or operate the backup printers properly, Hertzberg said.

Many of the canisters used to protect the paper tapes were not properly labeled or signed by election workers, the study found. "Such failures of custody increase the risk of a legal challenge," it said.

ESI recommended Cuyahoga extensively retrain election workers and overhaul its procedures. If the system is not improved, ESI warns, "One likely result is diminished public confidence in a close election."

"We are not seeing anything that is unfixable or insurmountable - this stuff is not a showstopper," said Hertzberg. "All these issues can be addressed as long as the partners involved are willing to address it."

Hertzberg says independent audits, done by researchers who are not part of the election process, are key to public confidence in the machines. "A sophisticated team is a whole other level of security," he said. "To avoid a public crisis of confidence, we can't go too far."

Swensen agrees and would like to see a regular independent audit of at least a small sample of machines.

"We are going to look into getting resources for an independent audit, but that isn't 100 percent," said Demma.

Another Ohio county has taken a do-it-yourself approach to checking voting results. Franklin County, which encompasses Columbus, is posting its ballot-by-ballot results online for anyone to analyze.

Franklin County Elections Board Director Matthew Damschroeder says a conversation with his mother gave him the idea. "Last Thanksgiving, my mom looked at me and said, 'How do you really know when you put your vote in the machine, it comes out the right way?' '' he said. "Mom, apple pie and questions about electronic ballots - that was the moment of truth for me."

Public confidence in voting depends on openness, Damschroeder said. "Posting that data online is our first step toward turning the auditing role over to the citizenry. Anyone in the world can print out individual ballots and count them if they wish."

Beyond issues of keeping individual voters' ballots secret, Swensen is intrigued by the idea. "Anything we can do to build confidence in the machines is good."

Ohio's Cuyahoga County election mess by the numbers:

89 percent of precincts reported at least one voting problem.

30 percent of all problems involved registration errors, such as misspelled names or wrong addresses.

23 percent were administrative problems, such as not being able to get through to the Board of Elections help desk.

16 percent involved problems with voting equipment.

9 percent of problems involved poll workers, usually a worker failing to show up.

4 percent were printing problems, but those were the most significant because the paper backup is the "official" ballot.