Palm Beach County elections officials could have averted a software glitch that erroneously awarded two Wellington Village Council seats to losing candidates if they had followed the instruction manual, the manufacturer has told state election officials. Supervisor of Elections Susan Bucher vehemently denied the claim. “I read the reference guide three times yesterday,” she said. “Nowhere does it tell you to check for this, ever.” Even as the question of who is to blame grew murkier, a clearer picture of the error itself emerged Monday, with the company, Dominion Voting Systems, sending out a national advisory warning election officials how to avoid a similar mistake. In the advisory, Dominion also suggests that the mistake could have been caught before the election had one key test been performed differently. The so-called “synchronization” error was caused when Bucher’s central vote-counting software was lined up to accept races in a different order than they appeared on the Wellington ballot. As a result, election-night totals on Wellington’s three races were shifted in a circle – with village council Seat 4 votes going to the mayor’s race, votes for mayor going to council Seat 1, and votes for Seat 1 going to Seat 4. On Monday, Bucher said the error occurred when her staff entered the titles on the ballot of each position, such as “Seat 1.”
Dominion’s release says that whenever information is changed, the software must be checked to make sure the ballot and the central vote-counting software match. Bucher says her staff was never told of the potential for an error. Before the mistake, they never had seen the page Dominion’s notice instructs them to check, she said. Bucher initially blamed a technical glitch with Dominion’s equipment. In a letter to the state last week, Dominion said its software didn’t cause the error but made it hard to catch.
Dominion’s advisory Monday was “in line with what the company has told us,” Florida Department of State spokesman Chris Cate said in an email. “The advisory is not new information,” he added. “It essentially highlights information that could be overlooked in the reference guide.” Bucher sharply disagreed. “What it’s showing us in the advisory notice is something new, something that we’ve not ever known how to do before, because we didn’t know it was a problem,” she said. The feud pits the makers of the equipment and the state officials who certified that it works against the local elected official responsible for operating it during elections.
Aside from warning elections officials to double-check the order of races after making changes, Dominion’s advisory says that the mistake also can be caught during the required pre-election “logic and accuracy” tests, in which a small number of sample ballots are fed into the system to make sure they are correctly counted. But Bucher’s office tested all three Wellington races with the same result – four votes for the first candidate, eight votes for the second. Because there were only two candidates in all three races, the results appeared to be correct, even though they could have been wrong, with each identical total being shifted to a different race.
This technique – using the same pattern of votes in every race – is common, Bucher said, and was originally recommended by Dominion. Cate confirmed the technique is used across the state.
It is “imperative,” Dominion wrote, to conduct a “sufficiently robust” pre-election test.