Friday, February 12, 2016

Can Required Multiple Preferencing Solve The Senate Reform Mess?

Senate reform has been a major focus of this site since the farcical outcomes of the 2013 Senate election.  The massive gaming of the Senate system by preference-harvesting micro-parties resulted in candidates being elected from very low primary votes, in a candidate being elected because of confusion about party names, in one state's election having to be cancelled and rerun because of the loss of a relatively small number of votes, and countless other absurd things.  Nearly two and a half years after that election, and almost two years after the Joint Standing Committee on Electoral Matters delivered a unanimous report in favour of an alternative, we were, until very recently, still to see any serious commitment to fixing the disaster from either major party.

The JSCEM proposal was to scrap group ticket voting and instead allow voters to distribute their own preferences above or below the line, although if they did so below the line a minimum of six numbers would be required.  Voters could continue to just vote 1 above the line, but if they did so then once all candidates from their chosen party were elected or excluded, their vote would exhaust from the count.  This is very similar to the system used in the NSW Upper House.

The JSCEM proposal has come under a number of attacks, claiming it would unfairly exclude minor parties (wrong), that it would advantage the Coalition (wrong and wrong again), that it would permanently stop "progressive" control of the Senate (wrong), that the existing system is sound (wrong and wrong) and so on.  Beyond the reality that there is no perfect solution, none of the attacks have had any merit at all, but there is still a lot of wariness of the JSCEM proposal, especially in Labor ranks.

In defending the JSCEM-proposed system against these many groundless objections, my own experience has been that there is still something stark about it.  It is so strategically different to what has gone before that while it would certainly be a vast improvement, there is still the risk of strange consequences while parties adjusted their tactics.  One undesirable aspect of it is that coalition partners would have to think harder than they currently do about whether it is strategically wise to run a joint ticket or two separate tickets. I can understand some of the desire to look for other solutions, even if all the arguments made against the proposal in public have been rubbish.

There may, and not a moment too soon, be some progress.  Last week Phillip Coorey in the AFR ("Government moves on Senate reform as Turnbull threatens early poll") related that the Government is actively exploring reform and has discussed it with Labor and the Greens, as a possible prelude to a double dissolution (if the latter is desired).  The model discussed would allow a voter to number a minimum of six squares either above or below the line. Group ticket voting would be scrapped.  David Crowe and Rosie Lewis ("Greens lead charge to curb micro-party Senate preferene [sic] deals") then reported that the Greens' Lee Rhiannon may introduce a private member's bill for a similar system.  Nick Xenophon has been supporting a vaguely similar proposal for some time, as have the Greens. This week Lenore Taylor reported that a deal has been struck to allow for numbering of six boxes above the line or twelve below.

The Importance Of Savings Provisions

As discussed by both Antony Green and Damon Muller, the key issue with any proposal of this kind is formality.  It is all very well to instruct voters to number a required number of boxes above or below the line to make their vote count, but some voters will be so accustomed to the existing system that they will just vote 1 above the line, while others will stop at four or five, or attempt to number at least six boxes but make mistakes.

My home state in its Lower House has particularly strict rules on formality.  A vote in each of Tasmania's five five-member electorates must include each of the numbers one to five beside candidate names once and once only, or the vote is informal.  (Mistakes beyond five are allowed, as is stopping at five.)  Around 2% of voters invalidate their vote unintentionally by failing to follow the instructions.  That, however, is in a system that hasn't changed much in a century.  In a transition to a completely new system the rate would be much higher.

Savings provisions are the sections in legislation that deal with what happens if someone clearly intends to vote a certain way, but doesn't follow the instructions on the ballot paper.  If the system being broadly considered (which I will call here "required multiple preferencing") has no savings provisions then the rate of informal voting will be enormous and the election will be another farce, and possibly one that skews badly against the ALP.

However, there are a several savings provisions options that would eliminate the problem.  The simplest one is to just adopt the ACT model - if a vote contains fewer than six clear and valid preferences, just exhaust it after the last clear and valid preference.  In the case that the vote is for a candidate who has over a quota on first preferences, consider that vote wholly set aside as part of the quota.

In my debate with Electoral Reform Australia, I didn't like their idea of applying the ACT model to a system that abolished group party boxes altogether and didn't even try to make voters vote for more than one candidate.  I didn't like it because it was possible for parties to be disadvantaged if they chose charismatic candidates who might attract a lot of #1-only votes that could then exhaust (or amplify leakage to other parties in the case of such candidates going over quota.)

However in the required multiple preferencing proposal, I expect the great majority of just-vote-1 cases will be above the line.  These will pass through all candidates of a party then exhaust, creating no leakage issue. Very few voters indeed will vote for a single candidate below the line in contravention both of instructions and the way the old system worked.  If a special savings provision is required for such cases at all it could be that a vote solely for individual candidates from a party runs through the remaining candidates from that party, and then exhausts. (Edit: This could well be unconstitutional - see first comment - unless it was made very clear to voters that if they voted in this way it would be counted a certain way, which would have the drawback of making them aware they could do so.  So perhaps it is better to just exhaust such votes, but there's a risk that this will hurt some parties more than others.) I see no need (and some danger, albeit small) in going for the South Australian method in which a full group ticket is the saving provision for incomplete ballots.

An issue that does come up with savings provisions (of any kind) is that even though the ballot paper will tell voters how many squares to number, it may be in the interests of some party to advocate that a voter deliberately numbers fewer.  (Issuing a simple how-to-vote card and avoiding preferencing other parties could be two reasons here.)  So we often see it made illegal for a party to promote a vote that is contrary to the instructions, even though that vote will be counted.

Whatever the arrangements it is utterly crucial that savings provisions exist so that if voters just vote 1, or mistakenly vote 1-2-3-4-5-5, their vote isn't discarded.  Savings provisions are also important to discourage parties from needlessly running ridiculous numbers of candidates.  So far the only comment in favour of savings provisions seen in discussion of the proposal has been by Senator Rhiannon.

What Would We See Under The Proposed System?

Firstly, the proposed system, if it had adequate savings provisions, would work.  It would achieve the same goal as the JSCEM proposal by making it more or less impossible for parties to get elected from very small primary votes.  It would make it slightly easier (compared to the JSCEM proposal) for minor parties with reasonable support (say, 7-8%) to be successful in some cases if they had very good preference flows from other parties.  Because micro-parties struggle to co-ordinate How-To-Vote Card distribution (since they have few supporters at booths), even if micro-parties attempted to preference-swap and benefit from preference harvesting, it would break down.

Compared to the JSCEM proposal it is, ironically, the Labor Party that might suffer compared to the Coalition and Greens.  Micro-parties tend to be either left or right aligned, and their preferences may well funnel more to the Liberals and Greens to Labor (as indeed they do under the current system).  Labor stood to gain from the preferences of micro-parties exhausting but for whatever reason has been scared of a proposal it could have done well out of.  On the other hand, Labor may in some cases overhaul the Liberals on Green preferences, but this will be a fairly  rare event since the Greens tend to either be in the mix for the final seat or else not over quota by much.  

However, overall, as with the JSCEM system, the only real losers out of a system designed to disadvantage preference-harvesters will be preference-harvesters.  Their time will be up.

A small diversion at this point: in a recent diatribe against the JSCEM proposal Malcolm Mackerras defended the election of Bob Day under the current system and said "To give my answers to these questions I imagine the election having been conducted under the Hare-Clark system which, as Canberrans well know, is the very best form of the single transferable vote. I can see no reason to doubt that [Sarah] Hanson-Young and Day would have been elected under Hare-Clark."

I can see every reason that Day would not have.  Bob Day was elected off the Family First ticket vote of 3.76% (or 0.263 quotas).  He was eighth in line for six seats and overtook Stirling Griff (with .74 of a quota to spare) and Don Farrell (with .59 of a spare quota) by preference-harvesting.

Tasmania has had Hare-Clark in its lower house since 1909 and in 165 Hare-Clark races in that time, nothing remotely resembling a win from a group vote as low as Day's has been seen.  The lowest vote with which any party group has won was the 7.36% recorded by Bruce Goodluck's group in Franklin in 1996, but that was still more than twice Day's vote in quota terms (since Franklin was electing seven seats rather than six).   Moreover, while Tasmanian Hare-Clark weakens preference flow by banning how-to-vote cards, there was no natural flow to Bob Day anyway since very few other micro-parties were likeminded to him.  The idea that Day even could have won under Hare-Clark is prepostrous.

The proposed new system is likely to see stronger cross-party preference flows and lower exhaust rates than Hare-Clark, because voters who follow the instructions will probably be unable to vote for just one party and stop (unless parties are really going to run twelve candidates and advocate a below-the-line 1 to 12 vote!).  All up though, because of the difficulty of directing micro-party preferences, I expect it to have a Hare-Clark-like flavour in that parties with very small primary votes won't win.

Under the proposed new system, as with the JSCEM model, minor parties with substantial support would still be potent forces at double-dissolutions.  With a quota of just 7.7% and relatively strong preferencing, it's entirely plausible parties in a state or two would win seats at double-dissolutions with, say, 2.5-4% primary votes.  A double-dissolution under the proposed system would return a crossbench with a more democratic mandate, but whether it would return a much smaller crossbench is not so clear - especially not with the prospect of a three or four seat Xenophon block.  What it would return is a crossbench elected under the new system and therefore less likely to be uncooperative with government legislation on account of the system change.

What's the catch?

Since this sounds like a very good system (provided it has adequate savings provisions) readers may wonder why I didn't overwhelmingly recommend it in the first place.  Mainly, I was concerned about the practicality and cost of any system in which every ballot paper is potentially unique.  Every paper will need to be either manually counted or else scanned and manually checked.

In the 2013 Senate election only around half a million ballot papers had to be manually entered.  Under the proposed system this will rise to about thirteen and a half million.  In most cases the ballots will be easier to enter because they will have fewer numbers, but they will still have to be very carefully checked to ensure all numbers have been entered.   For computer calculation, programs accustomed to extracting data from fewer than 100,000 ballots at the time will need to be rewritten to deal with over four million.  Some of these problems affect only the immediate election and may result in an extremely slow manual count.  Others will increase the cost of running elections substantially (though that is still better than having to rerun them, a la WA).

So far we have not seen any technical discussion of just how expensive this scheme will be to implement, how it would be implemented as a first attempt with anything from a few to maybe seven months of leeway, and just how long it would take to count the election.

I expect to post updates as more news on the proposed system comes to hand, and probably as I think of more aspects of the currently proposed changes that I have not discussed yet.  See also coverage at Tally Room, and Antony's latest (which goes further on the potential logistic issues with implementing the new system quickly - see especially the comment by Michael Maley, who has commented further in comments here).

Senate reform prior to the 2016 election (in the worst case applying to elections from 2019 onwards) is essential. Should the proposed new system fall over under closer scrutiny then the parties have a duty to agree on reform of some other effective kind, and no excuse for not so doing is acceptable.


  1. Michael Maley has sent the following excellent comments:

    First, re the possible savings provision you float, under which “a vote solely for individual candidates from a party runs through the remaining candidates from that party, and then exhausts”. This, I think, would be very risky: any process which deems voters so to have expressed preferences would be in danger of a constitutional challenge based on the requirement that senators be “directly chosen by the people”. (That’s not a risk with ticket voting, because it’s put to the voters as a clear alternative to numbering of individual preferences, with the tickets available at the polling places (see McKenzie’s Case), in contrast to the South Australian model, where the tickets are basically hidden from view.)

    Secondly, I completely agree that without savings provisions informality would skyrocket. But it would be likely to affect in particular those 20% plus of voters who have been inclined to vote for the non-major or micro-parties, and who wouldn’t have the guidance of how-to-vote cards. Frankly, for most voters, a ballot paper is just another form to be filled out, and how many people read the instructions before filling out a form? And more to the point, are there really many forms other than ballot papers where the most pedantic rules imaginable are be applied to try to have them invalidated? Wills are perhaps the only ones that spring to mind. With paper forms, people generally expect substantial compliance to suffice. It’s a bit different with computerised forms, where people expect instant error messages. (Years ago, when Colin Hughes was Electoral Commissioner, I asked him to approve some leave for me, and then pointed out to him that I had ticked boxes which the form said I should have crossed: he took the point.)


  2. (Michael Maley comment part 2)

    Thirdly, on counting: almost nobody in the public debate has come to terms with the implications of scaling up Senate counts. Suppose that all the Senate ballot papers in New South Wales have to be brought to one place for processing. That will in all likelihood be the largest assembly of ballot papers ever seen in one place in Australia, and will increase the scale of the central operation almost 50 fold compared with 2013. Intuitively, the difficulty of managing such operations goes up exponentially with their scale, not linearly. (Don’t ask me to prove it, but I’m sure I’m right on that.) And to avoid it becoming a giant SNAFU, you need not just more counters, but more table supervisors, shift supervisors, shift managers, … in other words, a whole management structure to attempt against all the odds to ensure that counter number 467 out of the 1000 you have in the aircraft hangar, who really didn’t understand the training so well, doesn’t bring the whole thing down in a smoking wreck. And you need a lot more time, and a huge increase in space.

    Too many people outside the business think that counting is easy, that you just open the box, and anyone can count what’s in a box. In fact, around the world, far more elections go sick during the counting than during the polling. I always used to tell my staff that a counting centre should be as clean and organised as an operating theatre, that a box of ballots is like a box of angry mice on benzedrine, and that if you don’t have a plan for how to control them when the lid comes off, you will be in bad stuff up to your eyeballs. I’ve seen two major meltdowns of counts where there were too many ballots being dealt with by too few people who knew what they were doing: one in Johannesburg in 1994, and one in Dili in 2007. In each case, they just found themselves overwhelmed almost before they knew what was happening to them. In the South African case, their problem was that they had planned for the counting, but not for the receipt of materials at the counting centre (shades of WA in 2013), and eventually 170 truckloads of ballot boxes and materials were simply dumped out at the front of the building by tired polling officials. It was like watching the Challenger explode.



    12 February 2016

  3. Centrebet is now running a market on whether Senate voting reform (of any sort) will pass before the next election. Current odds imply an 81% chance that it will.


The comment system is unreliable. If you cannot submit comments you can email me a comment (via email link in profile) - email must be entitled: Comment for publication, followed by the name of the article you wish to comment on. Comments are accepted in full or not at all. Comments will be published under the name the email is sent from unless an alias is clearly requested and stated. If you submit a comment which is not accepted within a few days you can also email me and I will check if it has been received.