The 9th U.S. Circuit Court of Appeals' decision to delay the California recall election because of fears that voting systems used in six counties would disenfranchise large numbers of voters failed to consider less disruptive remedies or the fact that its order might lead to more problems than it solves.

Despite criticism this week of the court and of a UC Berkeley study that was the basis for its ruling, it is clear that punch-card systems produce higher "error rates" than other ways of voting. The question is what to do about it.

The error rate includes overvotes, when people vote for more than one candidate, and undervotes, when a voter fails to choose any of the candidates for a particular office.
Some of these undervotes are deliberate, the equivalent of "none of the above," and they should not be counted as errors. But those instances should be the same across all voting systems, so the higher number of undervotes in systems using punch cards almost certainly has something to do with the technology.

The Berkeley study, which was funded in part by two companies that sell electronic voting systems, found that in the 2000 presidential election, 2.64 percent of ballots cast on punch-card systems had either an overvote or an undervote on them. This compared to 1.99 percent for paper ballots, 1.68 percent for electronic voting machines and 1.37 percent for optical scan machines.

For whatever reason, the rate seems to be related to the relative sophistication of the voters. As the proportion of high school graduates or the median family income increases in a county, the number of errors or undervotes declines. As the proportion of minorities in the population increases, so does the extent of these problems.

The court concluded that 40,000 Californians would be disenfranchised if the punch-card systems were not replaced by newer and more foolproof ways of recording the vote. But that number ignored the potential for reducing problems through voter education.

One remedy the court could consider as a compromise would be to order an intensive voter training program in those counties using punch cards, along with prominent signs in each polling booth reminding voters to check for hanging chads -- the little pieces of cardboard which, when not fully punched from the ballot, invalidate a citizen's vote.

The court also failed to consider the possibility that just as many voters might lose their franchise through a rush to new technology.

A July 2001 study by the CalTech/MIT Voting Technology Project found high error rates in the electronic voting machines.

"The mechanics of voting on these machines are often confusing," the report said. "It is often not obvious how to undo a selection, how to check that all races have been voted, how to distinguish between the office and how to register the votes."

Some systems are so touchy that a voter can push a button to move to the next page and more than one page will pass by without the voter noticing it, the report said. It can also sometimes be difficult to tell where the list of candidates for one office ends and the next one begins.

Even optical scanning, which is in widespread use and is considered fairly reliable, has problems.

"This system also loses a significant number of ballots, though less on average than other systems," the report said. "Election officials complain of paper jams, the cost of printing and ballot management."

Combining the special recall election, and its 135 candidates, with the already packed presidential primary ballot could also prove to be a disaster. Conny McCormack, the Los Angeles County registrar, told the Los Angeles Times that the new system to which that county is converting cannot handle a ballot that large.

The InkaVote system, similar to the optical scanning technology used for standardized testing in the schools, can accommodate 312 ovals, the equivalent of 12 pages of questions and candidates, the Times reported. But the last presidential primary took 10 pages, and the recall is expected to require eight by itself.

"It's more than a wrinkle," McCormack told the Times. "No one even asked the largest county in the state if we had the capacity to run it in March. The answer is no."

The only possible way out might be to run the two elections separately on the same day, with different ballots and different machines for the recall and the regular primary. This would obviously lead to long lines, confusion and frustration -- and to some voters giving up without having made their choice.

There is one other alternative. All the studies suggest that old-fashioned paper ballots are still a very acceptable technology, easy to understand and use, especially in an election with just two questions on the recall and two ballot propositions. Their only drawback is that they take much longer to count. But delaying announcement of the result for a few hours would be better than delaying the vote itself for five months and leaving Gov. Gray Davis, and all Californians, under a cloud of uncertainty.

Perhaps the court, if it is unhappy with punch cards, should consider ordering those counties to go back to using paper and pencil.