11/19/00- Updated 09:38 PM ET

Both machine, man needed to tally votes

By Toni Locy, USA TODAY

What's the most accurate way to count votes in a tight election?

Is Texas Gov. George W. Bush right that it's with machines? Or is Vice President Gore correct that it's with human hand counts?

Experts say that neither is exactly right. Nor exactly wrong. They say hand counts and machines are both needed to do the job right.

Many legislatures have reached the same conclusion. Bush signed such a process into law in Texas in 1997.

"Machine and hand counts are a good check and balance on each other," says Todd Urosevich, a vice president at Election Systems & Software Inc. of Omaha.

In close races, machines are only as good as the ballots they're reading. "What you are looking at is that voters will not, cannot or just plain do not follow instructions," says Doug Lewis, executive director of the Election Center, a non-partisan, non-profit group in Houston that trains elections officials.

Machines also are not foolproof. They can and do overlook flaws, such as incompletely punched holes on ballots.

People aren't the perfect alternative either. They can read only so many ballots without wearing down. That's why, Lewis says, the prayer of election officials everywhere is: "God, please let the winner win in a landslide."

Mechanical lever machines were first used in 1892. The punch-card-reading machines made their debut in 1964. William Rouverol, the mechanical engineer who built the prototype punch card reader, says the country went to this system because it wanted faster results.

Machines also reduced the manpower needed to count the increasing number of ballots cast by a growing country. By then, allegations were prevalent that elections had been rigged and stolen by corrupt people doing the counting.

Converting to mechanized vote counting brought "integrity" to elections in the USA, says Urosevich, whose company makes punch-card-reading machines and other voting systems. But the machines are only as good as the ballots they are asked to read. Relying on people alone to do the counting also can be problematic, especially when they are counting hundreds of thousands of ballots.

"The mind gets tired, the eyes get tired, the body gets tired," Lewis says.

In a new USA TODAY/CNN/Gallup Poll of 870 adults Sunday, the public perceives machines as more accurate than humans, 58%-35%. The poll had a margin of error of +/- 4 percentage points.

Urosevich says machines that read punch cards are 99.9% accurate, and maybe as much as 99.999% right.

In a close contest, that's not as good as it sounds. In Florida, where 6 million votes were cast, machines with 99.9% accuracy could have mistakenly rejected 6,000 votes. Machines with 99.99% could have kicked out 600. And those operating at 99.999% accuracy could have disregarded 60 ballots.

"That's a significant number of votes, and in a close election, a hand count is not only necessary, it's essential," says Rebecca Mercuri, a computer science instructor at Bryn Mawr College in Bryn Mawr, Pa.

In Florida, the state Supreme Court is being asked to choose between man and machine.

"I think the real question is: Who gets to decide, and who do we accept as the final decision-maker?" says Paul Rothstein, a law professor at Georgetown University in Washington. "Is it the Florida Supreme Court? The 11th Circuit Court of Appeals? Or the U.S. Supreme Court? It's whose word do we finally accept?"

Contributing: Bruce Rosenstein

(Requires: Real Player.)
Having trouble? Click here.