An analysis of New South Wales electronic vote counting

TL;DR

This paper re-examines the 2012 NSW local elections' electronic vote count, uncovers software errors affecting results, and discusses improvements for fairness and accuracy in electronic vote counting systems.

Contribution

It identifies critical software errors in official election counts and proposes methods to ensure fair randomness and accuracy in electronic vote counting.

Findings

Discovered a software error affecting candidate Rina Mercuri's outcome

Corrected the software error after notification and official correction

Found two additional errors in the 2016 election count

Abstract

We re-examine the 2012 local government elections in New South Wales, Australia. The count was conducted electronically using a randomised form of the Single Transferable Vote (STV). It was already well known that randomness does make a difference to outcomes in some seats. We describe how the process could be amended to include a demonstration that the randomness was chosen fairly. Second, and more significantly, we found an error in the official counting software, which caused a mistake in the count in the council of Griffith, where candidate Rina Mercuri narrowly missed out on a seat. We believe the software error incorrectly decreased Mercuri's winning probability to about 10%---according to our count she should have won with 91% probability. The NSW Electoral Commission (NSWEC) corrected their code when we pointed out the error, and made their own announcement. We have since investigated the 2016 local government election (held after correcting the error above) and found two new errors. We notified the NSWEC about these errors a few days after they posted the results.