The state audit of FL-13 began Tuesday and has been marred by conflicting reports over what is causing deviations in their “mock election.” The mock election was conducted on five spare machines that were prepared for Election Day but never deployed. In the tightly controlled test, state elections workers were videotaped as they cast "ballots" following scripts based on actual results from machines used on Election Day.
The state focused on three possible scenarios for the discrepancies: the script instructing voters on which selections to enter did not match the selections made in the real election, the vote simulators didn't follow the script exactly or machine error. Two of the 10 discrepancies were blamed on errors in the script. But the mock election results showed Jennings beating Buchanan 100 votes to 62 votes, not 95-62 as the scripts had called for. There also were five discrepancies in other races. Despite reviewing scripts, rechecking their math and watching videotape of the simulated election all day Wednesday, auditors resolved only one - a script error in a local race. Unanswered was what caused the deviations in the congressional race. After establishing the scripts were accurate, officials reviewed 24 videotaped ballots - each of which followed the script perfectly - and planned to review as many as another 227 today and Friday, if necessary.
State officials again downplayed the discrepancies, saying they likely resulted from errors in following the scripts. Buchanan's campaign also called the differences minor, saying the few that occurred show there was no widespread machine failure as Jennings and others claim. But Jennings' campaign said any variations in a limited mock election highlight why it's pushing for a comprehensive examination by outside experts.
The auditors were hampered by machine software issues as well. The machines kept no evidence of voters' actions other than a final ballot. Such machines are considered "software dependent," meaning the only way to verify vote tallies is to depend on the software that produced the results in the first place. But those machines are "more vulnerable to undetected programming errors or malicious code" and susceptible to election-rigging by a single person, the National Institute of Standards and Technology said in a recent white paper.
The next step in the state audit begins tomorrow, with state officials conducting a mock election on touchscreen voting machines actually used on Election Day, November 7.
This update was prepared by Debra Milberg, Class of 2009, Moritz College of Law.