Thursday 28 July 2016

Questions regarding section 282 Senate re-count for Tasmania

Introduction

Section 282 of the Commonwealth Electoral Act provides for the AEC to re-count Senate votes after a double-dissolution, essentially as a run-off election between the twelve successful candidates to determine six of them that might have been elected at a normal Senate election and would therefore be assigned a six-year term if the Senate chooses to apply the section 282 procedure.  While the AEC has not yet published the results of such a re-count, both Grahame Bowland (@angrygoat) and myself (@deanashley004) have made attempts to run a complete section 282 count from the full preference data made available on the AEC website.   We initially got different answers (including who got into the “top six”) and in the process may have stumbled on an unintended outcome of the liberalisation of below-the-line (BTL) voting.    

The critical issue is whether the section 282 count procedure includes re-testing ballots for formality and therefore requires BTL ballots to have at least six preferences among the twelve candidates included in the section 282 count.   At stake is the possible exclusion of tens of thousands of BTL ballots from that count, despite initially meeting all formality requirements and having some preferences expressed for section 282 count candidates.

The below is my analysis only and is not necessarily endorsed by Grahame.

The competing methodologies

There are at least three possible ways this could work and between Grahame and I we’ve run the count for the three below.

My initial thinking:   That section 282(1) refers only to repeating sections 273(7) to 273(30).  This does not include section 273(4) which is the exclusion of informal ballots.  I therefore ran the section 282 count on the basis that all previously formal ballots were available for inclusion and a BTL ballot would remain included (as a BTL ballot), even if it preferenced less than six of the twelve candidates in the recount.  The only ballots I excluded were those that had no preference for any of the twelve candidates.  I excluded these not as being informal, but as lacking a first preference and therefore excluded from the quota calculation in section 273(8).  Only 5616 ballots were so excluded.

The result of my count was the same six Senators as first elected: the top two from each of the Liberal and ALP tickets, and then Jacqui Lambie taking the fifth seat and Peter Whish-Wilson comfortably beating Lisa Singh for the sixth long-term seat.

Grahame’s initial thinking:  That a BTL vote that did not have preferences for six of the twelve candidates was now informal, and would be excluded unless it could be saved as an above-the line (ATL) ballot.  This is the basis on which Grahame ran his initial count.   This resulted in the exclusion of an additional 33,330 BTL ballots (more than one-third of all BTL ballots) and gave a quite different result.

While losing a lot of BTL ballots would be expected to hurt Lisa Singh, that was more than made up for by the resulting reduction in the quota, with a larger surplus reaching the Labor #3, Carol Brown.   However Singh still had her covered and this surplus subsequently made its way to Singh and allowed her to take the fifth seat, with Whish-Wilson narrowly beating Jacqui Lambie for the sixth.

The third possibility:   That I was right about BTLs not becoming informal as a result of lacking six preferences for the 12 candidates, but that BTL ballots left with zero preferences can be rescued if they also happen to have a valid ATL vote.  This is the basis of Grahame’s latest version of the count.  This relies on section 282(4) of the Electoral Act, which I admit I’d failed to pay attention to.

A count on this basis rescues 95 ballots that I had excluded but gives the same result – Whish-Wilson comfortably beating Singh for the final seat.

Comment

I’m not really comfortable with any of the three scenarios.  The problem with the third option is that if section 282(4) is opening the door to revisiting formality, then my original argument for allowing BTLs with less than six preferences to remain as valid BTLs is in real peril.

This leads back to Grahame’s initial run, and the horrific possibility that a huge number of BTL votes that may have started with twelve or more preferences now have less than six and are excluded from the section 282 count unless the voter happens to have taken the “belt and braces” approach and also voted ATL.   This possibility is almost certainly a legislative accident, but it may be real.   It appears to be the only option that allows room for a somewhat natural operation of section 282(4) - the first option essentially ignores it and the third applies it but denies the premise for its operation.   My main remaining doubt comes from a perceived inconsistency with section 282(1).

I actually delved into the history of section 282(4).  It wasn’t part of the original predecessor  to section 282 as introduced in 1984, but was part of changes introduced in 1990.  The 1990 changes also re-wrote section 239 to which it refers, but I don’t think they changed its operation relevantly and so it doesn't appear to be the catalyst for the change.  Unfortunately I haven’t been able to track down the Explanatory Memorandum for the bill and while the JSCEM report on the 1987 double-dissolution briefly discussed section 282 and the Senate’s failure to rely on it, it didn’t suggest any changes to the section.

[Update 29-Jul-2016]  I still haven't tracked down the EM for the 1990 changes - it's not in any of the usual places.  But I did find the second reading speech which suggested that the bill was primarily to implement changes recommended by reports of the Joint Select Committee on Electoral Reform from the previous parliament.   I found the relevant report but like the later report linked above, it doesn't appear to discuss section 282 or provide any recommendations in relation to that section or section 239.  The catalyst for adding section 282(4) remains opaque.  [/Update].

It’s worth noting that until the 2016 changes, the possibility of a previously formal BTL vote becoming informal for the section 282 recount would have been very low.  With the previous system requiring all boxes to be numbered and the savings provision only reducing that to 90% of them (with three numbering errors allowed), a ballot would need to be unlucky to be initially formal but have the empty boxes so concentrated that it ends up informal for the twelve-candidate runoff.  And in any event this would only be ballots that had been rescued by a savings provision in the first place. 

In the absence of a better explanation, it is conceivable that section 282(4) was included to allow salvage of the handful of such votes that also had an ATL preference.  It’s possible that the AEC noted the potential to save such votes in the 1987 count and reported this. [Update 29-Jul-2016] See update above [/Update].

However the implication if correct is that BTL ballots are supposed to be tested for formality over the twelve runoff candidates, and following the 2016 changes this would make informal (for the section 282 count) not just a small number of ballots that had already relied on a savings provision, but many thousands of ballots that originally had twelve or more preferences (including some for the runoff candidates), of which only a tiny fraction will then be rescued by an ATL vote.

The stakes may be high.  The first and third scenarios appear to give a long-term seat to Lambie, while the second partially disenfranchises tens of thousands of voters and appears to give it to Singh.   If this is the legally correct option (and I’m not sure that it is), it is clearly an accidental one.  And something similar could happen in other states, with the major parties most likely to benefit from the reduced quotas. 

Of course this assumes that the Senate applies the section 282 outcome, and any controversy over the results would be fuel for those proposing an alternative method.   While it would be hoped that Singh would be personally above advocating to take advantage of an accidental outcome, her party may have potential (including some legitimate) benefits in other states from applying the section 282 methodology.   And with that methodology enshrined in legislation it would normally be those proposing alternatives that would be accused of realpolitik.   

If there is in fact an unintended interaction between the new formality provisions for BTL votes and the section 282 re-count, the fact that Senate needs to decide the issue at its first meeting would appear to rule out the possibility of a legislative fix to allow another re-count on a revised basis.  So the Senate may be left with a choice between adopting a flawed outcome or doing as it has done in the past and applying another methodology.

[Update 04-Aug-2016] I've since noticed some parallels between the wording of sections 282(1)(b) and 273(12).   Section 273(12) definitely doesn't result in new formality checks and this may support my initial thinking (and that of others who have commented below and elsewhere) that section 282(1) does not.  If so, it still leaves the small mystery of a rationale for the 1990 insertion of section 282(4).  [/Update].