I just got back from being the Democratic observer at a "Risk Limiting Audit."
Here in Georgia, the Dominion scanners read either the mail-in ballot or the machine-printed ballot that is created when you use the machine to select your votes in person. The scanners read a QR-style barcode, but there's also a plain-text summary of the person's vote for every race printed on it.
But how do we know the scanners are reading the mail-in ballots correctly, or that the QR code actually matches the voter's picks? That's what this audit is for. The state generates random numbers that are then used for every county in Georgia to pick a few batches - the in-person votes from a specific precinct, or a specific 100-vote group of the mail-ins, or one of the bundles of early votes. While it doesn't show in the public results, the state keeps a record of the results from EACH batch as it was reported by the machines on or after Election Day.
So on Audit Day, each county pulls out those randomly selected batches, puts each batch on a table with two paid election workers, and they hand-count them. They use their human eyes to read the printed text, ignoring the barcode, or they visually count the check marks on the mail-in ballots. Those human-generated results are then checked against the machine-generated results, and we see if they match.
So next time you hear someone talking about the machines changing the vote, or wondering about those QR codes, or whatever, let 'em know - there really are humans going back and double-checking this stuff. And not just one human - there's two at each table, and election supervisors organizing it, and partisan observers sitting there keeping an eye on the process. And this is just ONE of the many checks and precautions. Other states have different procedures, but they're all careful and thorough. Election fraud is HARD, y'all - and that's why there's so little of it.
