Washington

Pencil, paper and simple optical scanners are the most reliable proven technologies for avoiding another voting fiasco akin to last November's US presidential election, say some of the country's leading scientists.

David Baltimore, president of the California Institute of Technology, and Charles Vest, president of the Massachusetts Institute of Technology in Cambridge, jointly launched the Caltech–MIT Voting Technology Project a few weeks after the election.

The project's findings — issued on 16 July — are that high-technology solutions have a poor record in improving voting accuracy, and that authorities should abandon them. The report, Voting: What Is, What Could Be, says that paper ballots and optical scanners are the most accurate, with an error rate of around 1.5%, whereas punch-card machines and electronic devices register nearly twice as many errors.

The report points out that Internet voting and other advanced solutions could open the door to large-scale, systematic voter fraud in the United States. It argues, nonetheless, that “properly developed” high-technology options could eventually prove superior to what is currently available.

The widely publicized problems that dogged last year's vote in Florida, where George W. Bush beat Al Gore by a tiny margin, were rampant throughout the country, the project finds.

“The problem is more serious than we had imagined even last December,” says Baltimore. “The voting process has simply not been taken seriously as a central element of the democratic process.”

The group estimates that between 4 million and 6 million votes were lost because of ballot, registration and other problems. To slash that number by about 1.5 million in the next election, the project recommends abandoning existing electronic, punch-card and mechanical-lever systems in favour of hand counting and optical scanning of hand-marked ballots. It also calls for a new system for field-testing other alternatives.

According to the report, procedural changes such as making regional registration information available at polling sites could prevent the loss of another 2 million or so votes, but further reductions will require better voting technology.

It recommends federal funding to develop new voting technologies which could include ballots, either electronic or paper, that can be printed on-site, allow voters to confirm their selections, and be stored for later review. Although optical-scanner systems come closest to that, they have drawbacks, including high costs, inflexibility and significant scanning inaccuracy.

“We want every American's vote to count,” says Vest. To that end, he says, the next phase of the project will focus on the design of new equipment and processes. “Computer technology appropriately applied offers significant opportunity for cost-effective advances,” he says.

→ http://www.vote.caltech.edu