The last remaining uncertainty about the composition of the New South Wales Legislative Assembly was resolved on Saturday (pending a recount), with a full preference distribution delivering Balmain to Greens candidate Jamie Parker.
Parker succeeded in edging Labor member Verity Firth into third place by a margin of 201 votes, before going on to a comfortable 2477 vote win over Liberal candidate James Falk with the distribution of Firth’s preferences.
The election of a Green to a single-member lower house seat is noteworthy enough in its own right, having only previously been achieved by Michael Organ at the federal byelection for Cunningham in 2002, Adele Carles in the Western Australian state byelection for Fremantle in 2009 and Adam Bandt in Melbourne at the 2010 federal election. Parker also achieved a first in securing his win at the expense of a sitting member.
Beyond that, the manner of the result’s determination — through a two-horse race for second place, which itself hinged on the preferences of the candidate who came fourth — is the kind of thing that keeps so many Australian psephologists in business.
Most of all, a tight three-cornered contest of this kind nicely illustrates what a difference an electoral system makes.
Britons will go to the polls on May 5 to decide whether to adopt a New South Wales-style optional preferential voting system — known in academic parlance and ordinary British usage as the “alternative vote” — in place of the first-past-the-post voting system the country has known since time immemorial.
The result in Balmain demonstrates precisely why they should vote yes. In a part of Sydney that has acquired an alliterative association with basket-weaving, preferential voting produced the contest that electoral justice demanded: one between the purist and pragmatic forces of the left.
By contrast, first-past-the-post would have seen Falk skate to victory Steven Bradbury-style, a result that would have failed the most basic democratic test by appalling most of his constituents.
However, the result is equally useful in illustrating an anomaly in the logic of our preferential voting systems.
South of the border, down Mexico way, a tellingly different saga unfolded in November when Labor and the Greens grappled for the inner city seats of Melbourne, Richmond, Brunswick and Northcote.
Here the electoral system was compulsory rather than optional preferential, which creates an expectation that parties will offer guidance to confused voters in the complete ordering of their preferences. Decisively, the Liberals broke with tradition by recommending that the Greens rather than the traditional enemy be placed last.
No such decision was made by the Liberals in New South Wales, because the electoral system didn’t demand it and tactical considerations encouraged an approach that kept the party’s message as straightforward as possible.
However, even if it were otherwise the result in Balmain would have been the same: the Liberal candidate would have remained in first place as the minor candidates’ preferences were excluded, before going down to defeat at the hands of the Greens with the distribution of Labor preferences.
Why were the Liberal votes that were decisive in determining the results in four Melbourne seats of no consequence in Balmain? For the simple reason that there were more of them — with the result that the Liberal candidate remained in the count until the very end, and his preferences were never allocated.
If it seems illogical that more votes for the Liberals in Balmain meant less influence than for the fewer such votes in the Victorian seats, that’s because it is.
Theorists have long been conscious of this, resulting in alternative approaches to the task of determining winners in systems where voters rank candidates in order of preference.
These are collectively known as Condorcet methods, so named after their 18th century French inventor, which also carry the more instructive name of “pairwise” vote counting.
Such systems seek to identify the candidate who would defeat each rival in a two-candidate preference distribution, rather than just the two who happen to make it to the finish line under the staged exclusion of defeated candidates familiar from Australian practice.
Under a Condorcet method, it would have been noted that the Labor and Greens candidates were preferred by the majority to the Liberal, and the winner between them would accordingly have been determined by Liberal preferences.
The Labor-versus-Greeens count of polling booth votes, which was conducted on election night, suggests the Greens would have won a very narrow victory just the same — albeit that such a system would perhaps have given the Liberals more incentive to flex their muscles, which they might well have done in the same way as their Victorian counterparts.
It should be noted that Condorcet methods, like anything else in life, have problems of their own. While it might be thought illogical for an individual to prefer A to B and B to C but nonetheless prefer C to A, it’s entirely possible for an electorate to do so when all its votes are added together. Various backstops have been proposed in the event of such a result, all of which undermine the methods’ conceptual elegance. It also raises the theoretical possibility of a winner emerging from an extremely low primary vote, no doubt explaining its failure to be adopted in government elections anywhere in the world.
Nonetheless, any Balmain Liberals out there who feel cheated at their essentially wasted vote might consider reminding the powers that be that something could conceivably be done about it. Perhaps they could start by writing to their newly minted member of parliament — while remembering that it’s a rare politician who favours reform to the system that gets him elected.