Monday, February 22, 2016

Senate Reform: It's Finally On!

Today the Treasurer, Scott Morrison, introduced the Commonwealth Electoral Amendment Bill 2016 (see explanatory memorandum) to Parliament.  This bill primarily reforms Senate voting to remove Group Ticket Voting and eliminate the broken preference-dealing system that led to many farcical outcomes at the 2013 election.  The defects of that system have been covered exhaustively on here before (click the "senate reform" tab) and I will not discuss them further here.  This article concerns the system in the new Bill, how it will work and whether it is any good.  I expect to update this article over coming days as news and comments come to hand.

The Bill was read in the parliament today and debate was immediately adjourned.  The Bill will now be scrutinised by the Joint Standing Committee on Electoral Matters (JSCEM) with a reporting date of 2 March.  Possibly the JSCEM scrutiny will still result in minor changes.  After that it can be sent to the Senate, where barring anything unexpected it will pass with support from the Coalition, the Greens, Nick Xenophon and possibly others.  Labor's stance on the Bill is not clear at this stage.  The Australian Electoral Commission has said it will need three months to implement the changes, which in theory keeps a July double dissolution in play.

Proposed Changes: Above The Line

The core recommendation of the JSCEM inquiry into the 2013 count was to scrap Group Ticket Voting and replace it with optional preferential voting.  But whereas the original JSCEM system would have openly allowed voters to just vote 1 (which would have led to a quite high exhaust rate), the Bill proposes that above-the-line voters be directed to vote "By numbering at least 6 of these boxes in the order of your choice (with number 1 as your first choice)".

To prevent an increase in informal voting, savings provisions will allow for above-the-line voters who just vote 1, vote for fewer than six parties, or make a mistake before their vote reaches number 6, to still have their vote counted.  Such votes will exhaust once they have passed through all parties they can be assigned to, so if a voter just votes 1 their vote will count for that party only.  

Although major reform of below-the-line voting was also canvassed, in the end the only reform included for BTL voting has been an increase in the number of mistakes permitted in a formal vote from three to five.  

At this stage there is no "Langer provision" banning voters from advocating a just-vote-1 style vote. However the purpose of "Langer voting" (eg 1-2-3-3-3 in the House of Representatives) was to allow voters to exhaust their preference without it reaching either major party.  In the Senate a voter wishing to do this can deliberately vote for six micro-parties with no chance of election, so it isn't clear why anyone would bother encouraging voters to exhaust their Senate preferences.

Minimal Change Below The Line

Very little has changed for voters wishing to vote for candidates below the line in order to vote across party lines or against preselected orderings.  They still have to number at least 90% of squares for a valid vote, however the number of errors permitted for a valid vote has increased from three to five.  

It is unclear why a proposal to allow BTL voters to number only 12 squares for a formal vote was rejected but some possible reasons could be:

* Since a BTL voter would fill in more boxes than an ATL voter, encouraging ATL voting in preference to BTL voting makes things simpler and hence cheaper and faster for data entry.

* Parties may have been concerned about voters using increased BTL voting to buck their own party's preselection order or vote across party lines.  It was hard to tell whether enough voters would do this to make a difference.

* Parties may have been worried about the risk of votes leaking out of their ticket or exhausting before reaching all listed candidates.

* Parties may have felt obliged to nominate long lists of candidates to ensure that a voter voting below the line could cast a vote which complied with the instructions yet was just for that party.

Of these reasons, the second at least should be considered invalid as it is a purely political reason.

Other Changes

Senate papers will no longer be counted at booths but will instead be transferred to central counting.  (I am not sure yet if they will be counted by direct data entry or if there will be a hand-count for primaries first.)  At this stage it is not clear how soon reliable information on likely Senate results might be available.  See Antony Green's post on this.

Parties will be allowed to display logos on ballot papers to eliminate voter confusion with parties with similar names.


The proposed system ticks the most important boxes for Senate reform.  When I first wrote about Senate reform in detail here I outlined that any Senate system trades off three objectives: high formality, full preferencing and voter control over preferences. 

While Sam Dastyari has been scaremongering about high informal votes, his claims must have been based on a hypothetical version of the system with no savings provisions.  In fact any vote that was formal under the old system will be formal under the new system.  If anything, informal voting might even decrease slightly.  

As with the JSCEM proposal this option values formality and voter control very highly (though the latter not quite as highly given the BTL differences) and sacrifices full preferencing.  However it does not sacrifice full preferencing to anything like the same degree as the JSCEM proposal.

In my view the direction to voters to number up to six party boxes represents an improvement on the JSCEM proposal, the question being how practical it will be.  Expect a bumpy ride here and for it to be an expensive election with a relatively slow Senate count process.

The lack of adequate BTL reform is on the surface disappointing (see more below.)


As with any system descended from the JSCEM model, this system is designed to disadvantage preference-harvesters and that is exactly who it disadvantages.  Aside from that there will probably not be any detectable difference in its impacts on the relative chances of Labor, the Coalition and the Greens.  I expect to post some more detailed modelling of the system when time permits, but the results will be broadly similar to my modelling of the original JSCEM recommendation.  Analysis based on results of previous elections is always a little shaky because if the larger micro-parties have even the slightest sense they are already negotiating mergers to increase the chance of fourth-party victories.

The system will completely prevent the election of parties with ridiculously low primary vote shares.  However it is possible in some circumstances that minor party or independent candidates will get up with something like 6% of the primary vote (3% for a double dissolution).  This wouldn't be likely to happen in many states at once, and in most cases a fourth-party candidate will need more like 8% (4.5% for a double-dissolution) to win.  

While some faulty analysis has suggested that only the Xenophon Team and the "big three" would win, I think it's very likely that in a double dissolution some of the other current crossbenchers would survive. This is less clear for a half-Senate election, given the collapse of the Palmer United vote.

I think we should give the new proposed system a cautious tick of approval but be very wary of possible implementation stuffups given the tight timeline and the amount of reprogramming and retraining that may be necessary (see the article below and comments for some more on these concerns.)  

More comments will be added as time permits.

Addendum 1 (Monday): 

There is also a thread at Tally Room where there are lots of comments.  As always the comments from Michael Maley are especially interesting and worth reading.  Michael raises the point that the proposed system is inferior to the original JSCEM proposal because it discriminates against some opinions, which the original JSCEM proposal did not.  Those opinions are those of voters who want to vote across party lines for candidates.  Michael gives the example of a voter who wishes to vote only for female candidates.  Another example would be a Greens voter who wanted to preference left-wing ALP candidates but not right-wing ones.  I personally will often put some ALP candidates high on my ballot but the Shoppie types last or nearly so.

It's probable that a fairly high proportion of voters who vote across party lines are very knowledgeable voters who actually enjoy voting BTL - but not all.  Some voters also vote BTL to put particular candidates (like, say, Shoppies) last, a preference for which discrimination is hard to avoid. All this mitigates the discrimination, but doesn't explain it.

Sometimes discrimination in system design is unavoidable, but it's not immediately clear that this is the case here.  I agree with Michael that we really need to see a convincing explanation of why BTL voting was not further liberalised - and also one of why it was liberalised at all, given that the increased error tolerance will affect such a trivial number of votes.  The increased error tolerance smacks of utter tokenism.

Another point raised is that " identical preferences for candidates may produce a formal vote if expressed using the above the line mechanism, but an informal vote if expressed using the below the line mechanism."  This is indeed a new curious feature of the system that even the old system did not have, but I'm not sure it is indefensible.  The voter voting below the line in violation of the instructions when they could have cast the same vote formally above the line following instructions is creating unnecessary work for electoral authorities.  I'm reminded of an extreme example of a perhaps similar principle: a voter listed numbers down one side of the paper then drew a ridiculous nest of arrows leading to the boxes.  The returning officer took the view that even if following the arrows would produce a formal vote, there were limits to how much work a returning officer should go to for a voter who had chosen to make things too difficult.

It was reported on Lateline that Labor would be voting against the bill.

Addendum 2: (Tuesday) BTL Proposal And Explanation Is A Farce

It's now my view that the proposed BTL change should not be made.  It will save trivial numbers of votes, it is tokenism designed to create an appearance of appealling to people who like voting below the line, and it makes manual rulings on formality more difficult, as well as probably requiring unnecessary reprogramming.

What we really need to know is why this embarrassingly token change to BTL, and only that change, was included.  A Twitter exchange started by Peter Brent (who, let's remember, was let go from paid blogging by the Oz so they could continue to pay for all that other stuff) finds Mathias Cormann claiming the model was needed to get Senate support given that Labor were divided.  However the Coalition needs only the vote of the Greens to pass the Bill, and Richard di Natale says the Greens were open to broader BTL reform.

Given that Cormann denies blaming the Greens and states that consultations included those with Gary Gray, one possible explanation is that the lack of broader BTL reform was an attempt to convince the ALP to support the Bill so it would have cross-party support, a noble aim.  However with Labor unreasonably opposed to reform anyway, if that is the case then the Coalition and the Greens now have the option to pass more meaningful BTL reform.  So why don't they take it?

Addendum 3 (Wednesday): Bill Amended

On-the-night counting of primary votes has been restored.  There is also discussion about BTL reform.


  1. Hi Kevin, the Australian Unions have a 'lovely' poll here regarding the senate changes. The fishy smell is quite overpowering...

  2. Urgh. An opt-in for starters. There was also a union-commissioned Essential which seemed to have skewed the results by stressing that the change was a Government policy so as to evoke a paranoid response, but I haven't seen the exact wording of that one.

  3. Kevin. Surely there is one huge issue here that you should comment on. That is: is strategic "anti-nutter" voting now possible with granular options. For instance someone could say "I want to prevent party X from having a total majority in the senate and I want to prevent a Harridine person from having the balance of power". OK. So does that person go to the 4th Senate candidates of their choice only as "blocks" against a nutter final senate spot?

    Please write on all this. I am NOT telling you how to tell anyone how to vote. I AM asking if strategic voting is generically possible.

  4. The kind of strategic voting that will be possible in this system is just the same that is possible in Hare-Clark and other simple non-ticket STV systems. The aim is to keep your vote in the count at full value for as long as you can and preferably to have it still there on the table at full value at the final exclusion. Anyone interested in strategic voting will need to vote below the line and number just about all of the boxes, and start by voting 1 for someone who is definitely not going to get a quota either on the first count or on surpluses. A good way is to pick all the candidates you would be reasonably happy with and put those who clearly have absolutely no chance of election first. That increases the chance your vote stays alive while candidates who you may like but who don't need your vote cross the line. If they don't make it your vote goes to them later.

    As a simplified version someone who really wants to vote for, say, Labor but wants to preference tactically should vote for the #4 Labor candidate first.

    At some point your vote will have to make a choice between the party you dislike and the nutter though, just in case they are the last two standing.

    This system will make this sort of strategic voting much easier than before but it is always a bit hit-and-miss in terms of whether your vote threads the needle or gets trapped in someone's surplus at some stage.

  5. Thank you Kevin. That helps. Your reply is largely the way I was thinking about voting strategically.


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.