Not long after the missing ballots fiasco during last year’s Western Australian Senate election delivered a hammer blow to the Australian Electoral Commission’s public credibility, new questions are emerging as to whether further errors have been leading to botched results in other counts. The answer to such questions, it so happens, is very clearly no. But it’s more than a little curious in the current environment that the AEC should be behaving in such a way as to give reasonable people cause to ask them.
The issue arises from the software the AEC uses to conduct its monstrously intricate Senate counts, a process spanning hundreds of phases through which candidates are progressively elected and excluded and their preferences distributed at varying values.
Details that must be accounted for along the way include the exclusion of small numbers of votes where minor errors are made in numbering of below-the-line preferences, and a peculiarity in the handling of rounding, which can cause fractions of votes to disappear and then re-emerge later in the count.
Get Crikey FREE to your inbox every weekday morning with the Crikey Worm.
Contentiously, this is conducted using what the AEC calls its “proprietary software”, EasyCount — which, unlike similar software used by electoral agencies in Victoria and the Australian Capital Territory, is hidden from public view.
Among those unhappy with this state of affairs are Hobart lawyer Michael Cordover, who sought access to the source code through a freedom of information application shortly after last year’s election. This was refused by the AEC with reference to exemptions that apply to material that would compromise trade secrets or diminish the commercial value of the disclosed information.
After a request for a review of the determination and some related email correspondence, the AEC has told Cordover it will seek to have him declared vexatious on the grounds that he plans to “harass” the commission with further applications. By way of contrast, Guardian Australia notes that it took 750 FoI requests before the Department of Climate Change took similar action in 2011 against Tim Wilson, then the Institute of Public Affairs’ director of climate change policy.
The reason this is a sensitive matter for the AEC is that it has a profitable sideline in providing “fee-for-service” elections to civic and corporate groups, in which it uses software adapted from EasyCount. In this it faces competition from commercial operators, as well as the state and territory electoral agencies.
However, the AEC’s remarkable heavy-handedness in dealing with Cordover has inevitably given rise to more fevered speculation as to what it might be trying to hide.
Putting some meat on these bones is the precedent established by the open-source software used for the Australian Capital Territory’s comparably more complex Hare-Clark elections. When the Australian National University’s Logic and Computation Group put this under the microscope, it was able to identify three bugs that had been in place during the counting of votes on five occasions going back to 2001, albeit that none were decisive with respect to the result.
According to two academic authorities on electronic vote counting, Rajeev Gore of the ANU and Vanessa Teague of the University of Melbourne, this demonstrates that quality assurance of the kind the AEC has relied on with respect to EasyCount is “meaningless”, and that it is “probable” that making it open source would lead to the detection of similar errors.
With all that said, the potential for democracy to be derailed by flaws in the AEC’s secret software needs to be kept in perspective.
The precise mechanism the AEC uses to count the Senate vote may be under wraps, but its publication of the data file containing the preference order of all below-the-line votes makes it possible to conduct a count independently and verify it against the official result.
Following the contested 2013 result in Western Australia, Perth-based coding enthusiast Grahame Bowland did just that, and was able to report that his result was precisely identical to the AEC’s.
Even so, what the AEC desperately needs right now is to restore the public’s confidence in the procedures through which it discharges the most vital of its functions.
Displaying acute sensitivity towards individuals showing a very healthy concern with the transparency of its processes seems a funny way of going about it.