[Image courtesy of CCGHouse]
A recent article in ArsTechnica takes a look at the math and procedure behind post-election audits. As I’ve mentioned elsewhere numerous times here on the blog, auditing is emerging as a powerful tool for election administrators – but it can be daunting (at least it is to me) to wrap your head around the concepts and statistics that make audits such a valuable idea.
The article follows a single risk-limiting audit in California’s Napa County conducted by a team led by Berkeley’s Philip Stark and goes, step by step, from the throw of 10-sided dice though scans of ballots to a hand count that ultimately confirms the accuracy of the Napa count.
Ars Technica’s Cyrus Farivar does an especially good job of thoroughly covering the issue – describing the process, distinguishing risk-limiting audits from other fixed-percentage audits and even identifying the challenges to more more widespread adoption of these audits (which right now includes the time to complete them).
I can’t really do the article justice with a simple summary; it’s well worth a read – and a bookmark in case you need a case study of how the rapidly-emerging idea of audits is transitioning from theory to practice.