search

ron-rivest / 2017-bayes-audit

Repository for paper "Bayesian Tabulation Audits Explained and Extended"
MIT License
0 stars 0 forks source link

Reported outcomes file should list more than just winning outcome. #9

Closed ron-rivest closed 7 years ago

ron-rivest commented 7 years ago

The reported outcomes file 2-reported/23-reported-outcomes.csv should not just list the overall winner of the contest.

For a plurality contest, it should list all candidates, and the vote totals they received.

For other sorts of contests, it should list all outcomes (within reason) and their tally or score (when this has a meaningful interpretation). Here "within reason" might mean all outcomes with at least 1% of the score/tally, or having at least 1% of the score/tally of the leading outcome, or something like that.

ron-rivest commented 7 years ago

Upon further thought, I disagree with this issue. For the purposes of the audit, we only need the reported winner. (At least for Bayesian audits.) The README has been revised to reflect this.