The Sunshine State is once again the scene of a messy election controversy, and residents of Sarasota, an affluent beachside community, aren't the least bit amused.
More than a month after polls closed, the certified loser for the congressional seat is refusing to concede, given an extraordinary wave of ballots with no vote cast in that race and a margin of victory as skimpy as a bikini.
But determining just what went wrong, if anything, has proved difficult given the voting machines involved: touch-screen computers with no printout for voters to confirm. The problems roiling Florida's 13th Congressional District may be one reason that a federal advisory board on Tuesday recommended that the next generation of electronic voting machines be "software independent." In essence, that means creating an independent auditing trail.
Thus, six years after a messy presidential election forced Florida and many states to spend millions of dollars to bring in electronic voting systems, an influential elections panel is urging better-designed systems that may bring back an element of paper. The recommendations will inform new guidelines drawn up in 2007 by the Elections Assistance Commission, of which the panel is a part. Nearly 40 states require that their voting systems meet those guidelines.
"What needs to be figured out [now] is what can we cobble together for a medium-term solution that is not too expensive," says Steven Hertzberg of the Election Science Institute, an election research group based in San Francisco. Then, "we need to innovate in this industry."
So far in Florida's 13th District, there's no evidence that the machines malfunctioned. Sarasota County has already done a machine and a manual recount. Neither significantly budged Republican Vern Buchanan's 369-vote lead over Democrat Christine Jennings. The Sarasota Herald-Tribune has also reviewed every ballot cast.
But suspicions remain because 18,000 ballots, or 13 percent of the votes cast, recorded no votes in the congressional race.
"It's a huge embarrassment," says Ron Basescu, a local resident who had trouble voting. After filling out his ballot on the county's touch-screen machines, which were installed in 2002, a review screen showed that he had either skipped the congressional race or his vote was not recorded. He went back and filled it in. The same thing happened to his wife. Now, he's not sure if their votes really counted.
"If 18,000 people don't cast a [complete] ballot, there must be something wrong," he says. "There should be a revote."
In the manual recount, officials printed out the records of each vote cast, says county elections supervisor Kathy Dent. Undervotes were found in all county precincts, she adds.
But the paper trail is suspect because it's not independent of the software. So if the computer system malfunctioned, this paper trail might have an error, and the voter is no longer there to catch it.
The state has tested for any problems 10 computers, five of which were used in the election. Officials are also testing the software to make sure it is identical to the program certified by the state. Once these tests are finished, the state should be able to say with certainty if the computers malfunctioned, Ms. Dent says.
But some election experts disagree. A comprehensive testing regime would be too complicated to be done quickly, says Mr. Hertzberg, and even then, testers may never know.
"The entire result is 100 percent reliant on the software itself, and there's no way to validate the vote total," he says. The bottom line: "You can't trust the result."
Last week, the National Institute of Standards and Technology came to the same conclusion about all electronic voting machines that do not have an independent voter-verified audit trail. Officials ultimately must trust that the software has remained error-free. "Verifying that this is the case is so complex as to be infeasible; current testing methods could not guarantee this," the report said.
Ken Fields, a spokesman for Election Systems & Software, the company that makes the machines used in Sarasota, says computer malfunctions can be ruled out through testing. The machines "have been proven time and time again to accurately record votes," he says.
The equipment has safeguards, including two different digital records of the votes that can be compared. Voters are also given a chance on-screen to verify their selections.
However, Mr. Fields says, "the security and accuracy of an election depend not only on the equipment, but on the processes used" by the jurisdictions.
In its review, the Sarasota Herald-Tribune reported Tuesday that it found only one significant pattern in the undervotes: A disproportionate number came from straight-ticket voters in both parties. The report surmises that such voters were probably moving quickly through the ballot and, because of poor ballot design, missed the House race. The newspaper notes that the undervote distributions suggest no computer problems, but "machine error can't be completely ruled out."
Some voters will probably use the same touch-screen machines in 2008 and beyond. Election officials are wary of current paper-trail systems on the market because printers have been known to jam, printed records can be difficult to interpret, and few voters actually look at the printouts, says Doug Lewis, executive director of the National Association of State Election Directors.
Voters in Sarasota, however, decided last month to replace their touch screens with optical scanners, an electronic system that counts paper ballots.
But there's still this year's House race to sort out. The losing candidate, Ms. Jennings, is contesting the election in the hopes of getting a judge to order another. The case is scheduled for Dec. 15. Jennings could also appeal to the House for a new election.
Some residents feel that Jennings's efforts are just prolonging a bitter race. "It was a very nasty campaign," says Ruth Bellaire, a retiree. "I think a lot of people didn't want to vote, to tell the truth."