This repository supports the possible auditing of the 2016 Australian Senate elections.
This work is joint with Vanessa Teague (U. Melbourne), Philip Stark (UC Berkeley), Zara Perumal (MIT), Berj K Chilingirian (MIT), Ian Gordon (U. Melbourne), Damjan Vukcevic, and Ronald L. Rivest (MIT).
One approach we plan to explore is the "Bayesian audit" method of Rivest and Shen.
Another family of approaches we may explore are the "black-box auditing" methods
developed by Stark and Rivest (unpublished, although mentioned in
this talk).
These are ballot-polling methods (not comparison audits).
These approaches can use a variety of methods for computing "variant samples",
since as a Polya's Urn approach, or the
method developed by Rivest and Yu.
We use with appreciation python code and materials developed by Grahame Bowland for computing the outcome of such an election. Some output from his program can be found here.
We also appreciate and use the code and materials developed by Berj K Chilingirian, Zara Perumal, and Eric C Huppert as a framework for election auditing.
Further materials related to post-election audits can be found on the voting-related web page of Philip B. Stark.
Requires python 3
To clone repo with submodules:
git clone --recursive https://github.com/ron-rivest/2016-aus-senate-audit.git
To pull the latest working versions of submodules use:
git submodule foreach git pull origin master