One of many filed by Donna Curling….

CIVIL ACTION FILE NO.: 1:17-cv-2989-AT

PHILIP B. STARK hereby declares as follows:

1.This statement supplements my declarations of September 9, 2018;September 30, 2018;October22, 2019;December 16, 2019;August23,2020; and August 31, 2020. I stand by everything in the previous declarations.

Excerpts:

Risk-limiting audits of a trustworthy paper trail are a powerful and efficient tool to ensure that the reported winners really won. Risk-limiting audits of an untrustworthy paper trail are a distraction from fundamental problems in election integrity, not a cure. RLAs are not magic

Arlo is based on a method published in 2012and does not completely implement that method. The method is a “lowest common denominator” method expressly designed to be simple and to demand little of the voting system, rather than to be efficient. There were more efficient auditing methods before 2012, and there have been many advances in RLAs in the last eight years. Arlo does not implement the most efficient methods for conducting RLAs, nor methods for auditing many kinds of contests, including super-majority contests. Arlo only supports “ballot-polling” audits, one of the least efficient methods. Arlo does not support the most efficient method, ballot-level comparison auditing. Arlo supports only one way of drawing a sample: unstratified random sampling of individual ballot cards, with replacement. Other sampling methods are more efficient and give jurisdictions more logistical flexibility

https://www.stat.berkeley.edu/~stark/Preprints/cgg-rept-7.pdf