Thinking about Elections

HOW DO YOU KNOW SCANNERS COUNT VOTES ACCURATELY? On this subject, my last post was about “Logic and Accuracy testing” of every piece of election equipment prior to an election. This post is about the audits done after an election, comparing the tabulated result to the actual ballots. If you think that the machines flipped votes, these audits will detect that. If you hear that someone’s computer model suggests the possibility of hanky-panky in vote counting, auditing the actual ballots will check this. Here’s how it works.

Scanners that tabulate votes are specialized computers that read paper ballots. The tally is maintained in memory, typically on removable media. How can you confirm the tally is accurate, that the machine hasn’t malfunctioned or been hacked? You do it through a risk-limiting audit (RLA.)

This special type of audit involves (1) storing paper ballots securely until they can be checked, (2) manually comparing a statistical sample of paper ballots to the computers' records for those same ballots, (3) checking whether all the computer records of ballots in the election were totaled correctly. RLA provides statistical assurance that election outcomes are correct by manually examining portions of the audit trail—paper ballots or voter-verifiable paper records. The RLA ensures that at the end of counting ballots, the hardware, software, and procedures used to tally votes found the real winners.

I’m posting two videos that explain RLAs. The first is a 3-minute video overview. The second is longer and describes the process in more detail using Colorado to illustrate. I found the second one really helped me understand this.

Note that RLAs require paper ballots. Counties without paper ballots cannot do RLAs. They cannot do recounts either.

https://www.youtube.com/watch?v=V4WI4otRp3Y

https://www.youtube.com/watch?v=x_dw6GpCC6Q