Friday, May 24, 2019

Gladstone Rises Up: An Error In The 2013 Tasmanian Senate Count

There's apparently not all that much going on in the 2019 election postcount, where the only major dramas left at present appear to be which (probably) left party loses in the Queensland Senate and whether anyone can possibly avoid a recount in Macquarie.  When I compare it to 2016, I'm quite surprised at how busy I'm not.

This means I have time to post something curious I've been meaning to post for some time.  As is well known, the 2013 Senate count was not the Australian Electoral Commission's finest hour.  In Western Australia, the original count had a tipping point between two candidates, neither of whom could win, but the resolution of which determined the final two seats. The loss of 1370 ballot papers meant that it could not be determined who had won, and as a result the entire 2013 WA Senate election had to be voided and rerun in 2014.  This resulted in the resignations of the Electoral Commissioner and the Electoral Officer for Western Australia and major changes to the way ballot papers are handled.  The farce also contributed to the death of Group Ticket Voting at federal level. Under the system we have now the tipping point would have been irrelevant and the lost ballots may well not have affected the outcome.  Many other issues with the AEC's culture were identified in a review and many positive changes have been made.

I can now reveal that there was also a mistake in the count in Tasmania that year, and it got through every possible keeper and into the final results.  Earlier this year I was compiling some booth swings and I noticed that something weird happened in the Gladstone booth in Bass in the 2013 Senate count.  Gladstone is a small rural mining town in the state's far north-east corner that at least used to have a somewhat wild-west reputation.  The Labor-Green state government and its federally-backed forestry peace deal were not well taken in the area, or in the rest of Bass for that matter, and Andrew Nikolic (remember him?) recorded a 12-point 2PP swing in Gladstone and a primary vote of 57%.

But oddly, in the Senate in the Gladstone booth, the Liberals recorded only 7%.  And the Palmer United Party, who recorded over 8% in the Reps in Gladstone, could only manage two below-the-line votes (rather weird when BTLs were much rarer than ATLs in that election.)  Instead the spoils went to the far-right Rise Up Australia Party, who despite polling only 110 ATL and 19 BTLs in the entire rest of Bass, had somehow managed 82 ATLs and zero BTLs of the 166 formal Senate votes cast in Gladstone.

So perhaps something was driving this, a home-town candidate perhaps?  Well firstly, in that case, why were the Green and Labor votes more or less on a par with their House result and only the Liberal and PUP votes affected?  And also, how to explain that nobody out of these 82 voters voted below the line, a 1 in 472,831 chance based on the behaviour of the remaining Bass Rise Up Australia voters?

As it turns out, there wasn't any home-town effect.  The table below, showing the 2013 Tasmanian Senate groups and also their 2013 Reps and 2010 Senate percentages, indicates what the real problem was (click for larger clearer version)

Clearly, Rise Up Australia has been given what should be the Liberal above the line total.  The Liberals have been given Palmer United's above the line total.  Perhaps Palmer United have been given Rise Up Australia's above the line total, or perhaps there was some other error (two zeros instead of one, or two ones instead of one) in the parties between PUP and the Greens - by the time the list reaches the Greens, it's back on course.

The old Senate system only required the 1-only ATLs to be counted and the numbers of ATLs checked and entered, whereas under the new system every ballot has to be computer-scanned and then confirmed as correct by two human operators.  If the human operators disagree or raise an issue it is raised to a higher level.  A mistake of this kind therefore wouldn't occur.  But this mistake could have happened in 2013 simply if there was some kind of error in data recording and checking processes.  And we know that in 2013 there were a lot of mistakes.

This error then made it all the way to the final count without being caught, and Rise Up Australia are still credited on the AEC website with 82 votes they did not get, costing the Liberal Party about $200 in public funding.

What difference did this error make?

The 2013 Senate count for Tasmania was in fact very close, because of tipping points created by Group Ticket Voting (though it was ultimately not quite as close as the 2016 final seat between the Greens and One Nation).

Jacqui Lambie (then Palmer United Party) won the final seat.  But her victory depended on three tipping points:

1.  The Sex Party's Robbie Swan outlasted Family First's Peter Madden by 821 votes.  Had this not happened, Madden would have won the seat, fuelled by a perfect preference snowball thanks to foolish preferencing decisions by Labor and the Greens especially.

2. Labor's third candidate outlasted the Sex Party's Robbie Swan by 244 votes.  Had this not happened, Swan would have won the seat, even though he doesn't live here.

3. Lambie outlasted the Liberal Democrats' Clinton Mead by 1276 votes (he doesn't live here either).  Had this not happened Lambie would have been excluded, but the LDP would also have lost, and the Liberals' Sally Chandler would have won the seat.

The main error was the Liberals being short 82 votes that were instead credited to Rise Up Australia.  These 82 votes then helped Rise Up undeservedly outlast the Country Alliance, but that made no difference since in either case both would be eliminated.

The 82 misplaced votes then flowed to Democratic Labour.  On their exclusion, they flowed to Australian Independents. On their exclusion, they flowed to Family First at preference 11.  Finally, on Family First's exclusion, they flowed to the Liberals at preference 23, and now everything was back on track.

So ultimately, although these votes shrank the margin at the second-closest exclusion point in the count, they weren't enough to determine the outcome.  And presumably had they been enough, somebody preparing a challenge would have noticed it somewhere (but who knows).

I've devoted a lot of time to arguing the case for the new Senate system in terms of the way it gives voters freedom to effectively direct their own preferences, and ends the election of parties from tiny vote shares through dodgy preference deals.  But here's another very tiny argument in its favour - it forces the checking of every specific ballot, rather than just the adding and recording of totals in which an error of this kind might happen.

PS Nick Casmirri has another such case - 368 Labor votes to Group X in NSW Senate 2010.

No comments:

Post a Comment