State law requires an audit of one race every two years, and Republican Secretary of State Brad Raffensperger decided to check his own win over Democrat Bee Nguyen. It was the largest margin of victory in any statewide race, at 9.3 percentage points.
He declined to audit the close U.S. Senate race between Democratic U.S. Sen. Raphael Warnock and Republican Herschel Walker, saying it would require election workers to review a huge number of ballots by hand while they’re also preparing for a runoff Dec. 6.
The roll of the 10-sided dice created a random 20-digit number that was fed into a computer to select ballot batches that will be pulled in all 159 Georgia counties.
Election workers started retrieving the ballot batches and sorting them into piles for each candidate on Thursday. When the audit is completed, the results will be compared with computer-generated tallies.
This is Georgia’s second statewide election audit, but it differs from the first ballot review in 2020. That one recounted all 5 million ballots cast in the presidential election to determine that Democrat Joe Biden had in fact defeated Republican Donald Trump by about 12,000 votes.
This time, as many as 300,000 ballots will be reviewed to ascertain with a 95% confidence level that the outcome of the race was correct. In all, there were nearly 4 million ballots cast.
Critics of Georgia’s voting machines, manufactured by Dominion Voting Systems, say audits wouldn’t necessarily detect whether voting computers that print out ballots had been manipulated, and few voters would have noticed whether the printed text had been altered.
Georgia officials say there’s no evidence that voting equipment has ever been manipulated during an election, and that machines are kept secure by local election officials.
“I’m not a fan of electronic machine voting for all voters,” said Joy Wasson, a DeKalb County voter who attended the start of the audit at the Capitol. “I believe in hand-marked paper ballots as being a more secure, efficient way to run elections.”
Sterling said the results of the audit might not exactly match computer results because of human counting errors, but the count should be close.
“So for all conspiracy theorists out there, when these don’t match exactly, that’s expected,” Sterling said. “That is not showing fraud. That is not showing anything.”
The audit is expected to be finished by Friday, and then the secretary of state’s office will post the results on its website. It will be conducted with software from VotingWorks, a nonprofit company that the state also used in its audit two years ago.