So far, it looks as though Utah's new touch-screen voting machines worked properly.

But Utahns cannot really know for sure whether the machines accurately recorded everyone's vote or whether high-tech tampering occurred.

Granted, there will be the occasional recount based on a few very close races, and local election boards will perform cursory canvasses of all the results. But there will be no scientific sampling of machine results for this election to verify the accuracy of the count statewide.

Utah is among half the states in the nation that do not audit electronic voting machine results as a matter of policy or law. Critics of the machines warn that regular audits are necessary to ensure that computer glitches do not alter the results of elections, or that computer hackers have not somehow thrown an election.

They are right.

Paper ballot records are essential, of course, and when a recount comes up with a discrepancy between paper and electronic records, the paper record should rule. To its credit, Utah election officials insisted that Diebold, the company providing Utah's voting machines, equip each one with a paper printout so that each voter may verify that the ballot is accurate before the final vote is cast. The machines store the printouts in locked boxes in case they are needed for a recount.

But while Utah does not now have an audit requirement, Lt. Gov. Gary Herbert plans to implement one by the November elections, according to spokesman Joe Demma. This is good news, because it's the only way to verify. Herbert's plan is to seek expert advice on the matter this summer and add the requirement administratively.

"We could just randomly take 10 or 15 machines and audit the paper trail, but what we want are professional guidelines," Demma said. "We're going to look for professionals to set some guidelines for what would qualify as an accurate statistical model."

The League of Women Voters has also called for audits.

In contrast to states that do nothing, California audits 1 percent of it electronic voting machines in each election to ensure ongoing accuracy and integrity. Utah would do well to follow suit with at least that percentage.

In the past, such procedure was less critical. A punch-card ballot requires no calibration; its operations do not depend on software. It would be immensely more difficult for an election hacker to make a physical alteration in the punch-card trays or card readers that would change the result of the vote.

Computers, by contrast, count based on invisible criteria -- codes and algorithms.

Election officials and Diebold representatives maintain that the machines are secure enough to render the chance of fraud extremely remote. The machines are not networked, and voting information is transferred on encrypted memory cards. Other than possible tampering with some future wireless transfer of election data, sabotage would have to be an inside job.

To hear Diebold describe it, their Titanic is unsinkable. We admit it's pretty strong, but we are also mindful of the present condition of the RMS Titanic -- it's a rusting hulk at the bottom of the Atlantic, a monument to overconfidence.

Regular audits would reassure voters. There are some who remain highly skeptical. It doesn't help that Diebold's CEO joked that he would ensure that George W. Bush got Ohio's vote in the 2004 election.

Everything from gas pumps to police radar guns to grocery store scales require careful calibration. Elections, it seems to us, are more important than any of those.

As Ronald Reagan said, trust but verify.

This story appeared in The Daily Herald on page A5.

Do you agree?