Thursday, May 21, 2015

What Happens In Kim Booth's Recount?

(NB added May 26: The recount will take place on June 9.)

Tasmanian Greens leader Kim Booth suddenly announced his resignation from parliament on Wednesday. This decision follows the very recent death of Booth's father, but with family time and party regeneration cited as motivating factors.  (Some have suggested he was jumping before he was pushed as leader, but if so he did well at acting happy about having decided to go.) As with Christine Milne there are now many effusive tributes to Booth's career flowing, as well as trolling from the odd party-pooper, and a few joyful if slightly unhinged celebrations of the party's (supposed) impending demise.

Booth's headkicking style has contributed many memorable moments to Tasmanian politics, most notably the "shredder" affair in which he brandished a reconstructed shredded document in Parliament shortly after its existence was denied, removing a Deputy Premier.  His bad-boy purist-rebel image was such that Greens' advertising at the last election showed Booth (who was often at risk of losing his seat) grinning with the slogan "There's only one thing worse than having Kim Booth in Parliament, and that's not having Kim Booth in Parliament".  And now, we'll find out just what that is like.  I suspect there will be many of my readers who greatly admire his contribution, and others who cannot stand him and will be pleased to see him go. For my own part, I've enjoyed the few times I've talked with Booth in person during his 13-year career in state politics.

Unlike with Milne, at the moment I don't have the time to write a long analysis of Booth's tenure as leader, and there's not that much to say anyway.  Booth is the first Tasmanian Greens leader not to actually take the party to a state election - they only contested one Upper House seat and local councils on his watch - and their popularity seems to have stayed at about the level of the 2014 state election, or maybe recovered very slightly.  (More on this from EMRS very soon).

The situation with the Greens suddenly perhaps losing two of their three MHAs and taking two lowish-profile incumbents into the next election is being painted as a crisis for the party, but with almost three years til the next state election the timing could have been much worse.  I think their biggest challenge in Bass is not the loss of Booth's profile, but the fact that they were lucky to win in 2014 with the vote that they had at all, and a big swing from Liberal to Labor could cost them that seat unless their own vote goes up quite a lot.

The crisis might also be an opportunity - a chance for the party to put the purist/pragmatist divisions of the previous term (between Booth and the rest) a bit further behind them.  (There has been some speculation that Nick McKim might reconsider his Senate bid, but at this stage I am unaware of any substance to it.)

What has had my phone running unusually hot is the question of Booth's countback.  Hare-Clark recounts (unfortunately the legislation uses the ambiguous word "recount" instead of the much better "countback") are quite complex and this one is unusually difficult to call, especially by the standards of recounts involving the Greens.  Booth's seat will be filled by one of his fellow Greens (assuming they contest the countback) but it is not so clear which one, with the most likely chances Amy Tyler and Andrea Dawkins.

I warn that the following discussion of Hare-Clark recounts is quite complex (about Wonk Factor 4/5 in places).  There is just no way I'm aware of to spell this stuff out simply.  Or if that didn't scare you off, perhaps this will:

The basics of Hare-Clark recounts

(See also Greens Change Leaders and replacing Milne as Senator.)

Before I get into the nitty-gritty of the Booth recount I just want to explain some general things about how casual vacancy recounts in Hare-Clark work.  Hare-Clark is a proportional representation system and so it wouldn't be fair to have by-elections whenever someone resigns, since that would often result in a different party winning, affecting the proportional nature of the House.  Also, a minor party MHA would probably not quit their seat if it was going to go to another party, even if there were very good reasons for that MHA to quit.

The basic principle of the system is that the people whose votes elected the retiring member decide who should replace that member.   It is not about how many primary votes the unsuccessful candidates got at the previous election, nor is it about how close they came or did not come to getting elected.  The votes that elected the retiring member are what matter.

The votes of the retiring member are allocated to candidates who are contesting the recount (which can be anyone who wasn't elected at the original election who agrees to contest, even if they've left the state).  Often one candidate will have more than half the recount votes and win immediately.  Otherwise there is a simple pattern of exclusion of the candidate with the least votes, and passing their votes along to the next person on them, until someone wins (like a single seat election for federal politics).

What Votes Will Be Used In The Recount?

Hare-Clark recounts are governed by Schedule 6 of the Tasmanian Electoral Act 2004.

In the 2014 Bass election, Kim Booth was fifth elected with 9330 votes, but never reached the quota of 10744.  The reason he did not reach quota was that a situation was reached in which Michelle O'Byrne (9584 votes). Booth (9330) and Brian Wightman (8296) were the only remaining candidates.  At that point Wightman was sixth and was excluded, leaving O'Byrne and Booth as winners but short of the quota.

It may seem to those who have grasped the basics that it is just Booth's 9330 votes that will determine the recount, but it is actually a little bit trickier than that, and the reason can be found in the fine print of section 4.(2) of Schedule 6.  Because Booth was elected without a full quota, Wightman's votes are thrown to attempt to get Booth up to a quota for the recount.

In this throw of Wightman's votes, up to 1414 votes worth could in theory end up in the Booth recount.  The only other choices for them are O'Byrne and exhaust.   If this happened at a normal election count, they would go heavily to O'Byrne and put her over quota, creating a surplus (and causing even votes that had Booth last to in some cases reach him!) but I can find no indication that this happens.  On my reading (and I'm not absolutely sure on this, so if it's wrong I will correct it), only those votes from Wightman that would have gone next to Booth (instead of O'Byrne or exhaust) play any role - and there won't be too many of those; probably a lot less than 1414 anyway.

So the votes used in the recount will be:

* 6661 primary votes for Kim Booth. (At least 62% of total)
* 2669  votes worth of preferences that reached Kim Booth from the 16 exclusions and three surpluses that happened prior to him getting elected. (At least 24.8% of total)
* Up to 1414 (but probably much less) votes worth of preferences from Brian Wightman thrown to attempt to get Booth up to quota. (At most 13.2% of total).

The expression "votes worth of preferences" refers to the fact that there will be some ballot papers at less than full value - I am giving the total value in whole vote terms of the ballot papers at each stage.

What do we know about where these votes will go in the recount?

In the case of the at-least-62% of 1 Booth votes, we know nothing from the cutup about where they will go in the recount.  As these votes were never distributed, there is no official information on how they flow. The Greens may have done some scrutineering sampling for the chance this would someday occur, but it would also be a challenging thing to sample.  If I obtain any such data I will post it.

There is some official information on where all the other votes came from, from the cutup.  As well as the up to 1414 votes Booth may gain from Wightman to bring him to quota, he gained 800 from fellow Green Andrea Dawkins, 283 from fellow Green Amy Tyler, 184 from fellow Green Anna Povey and 75 from fellow Green Ann Layton-Bennett.  346 votes came from the Liberal ticket when their last candidates were excluded and elected, and the remainder comes in small trickles from various unsuccessful Labor, PUP, Australian Christians, Liberal and Ungrouped candidates, and from the surpluses of Michael Ferguson and Peter Gutwein.

Each of these votes then goes back to the highest-listed candidate on the original vote who is standing in the recount.  But that isn't always the candidate they came to Booth from, so the numbers just given can mislead.  Suppose someone voted 1 Povey 2 Dawkins 3 Booth.  Then their vote would be one of the 800 that went to Booth when Dawkins was excluded, but it would actually go to Povey first.  Since around half of Dawkins' votes came to her from other sources, she wouldn't actually get anything like all those 800 back right away.

A more useful way of looking at it is that of the Greens' candidates primaries (Dawkins 467 Tyler 369 Leyton-Bennett 334 and Povey 339), each candidate gets back whichever of those primary votes ended up with Booth rather than leaking to Sarah Courtney, Michelle O'Byrne or exhaust.  There is some evidence that Dawkins' votes might be a bit more likely to leak than the others, so Dawkins' advantage over the other Greens on this part of the count is very small.

Wightman could also get back hundreds of votes because of the complication mentioned above, but it won't make any difference if he does.

What can we infer about the #1 Booth votes and where they go?  
The at-least-62% of votes that are 1 Booth will largely flow to other Greens.  We have three reasonably recent examples of the preference flows of #1 Green candidates on a ticket:

* In 2010 Nick McKim (Franklin) achieved a surplus and 83.2% of it flowed to a fellow Green
* In 2006 Peg Putt (Denison) achieved a surplus and 89.7% of it flowed to a fellow Green
* In 2002 Peg Putt (Denison) achieved a surplus and 83.2% of it flowed to a fellow Green

So we can be confident right away that at least, say, three-quarters of this tally (probably more like 80-90%) is going to another Green. Also, when unsuccessful Greens are excluded during the recount cutup, these votes will tend to stay within the ticket.  The proportion of Booth voters who have voted for all five Greens before anyone else should alone be almost enough to ensure a Green wins the recount, with the votes that came to Booth from the other Greens making it certain.

The question is what we can know about where these votes will go between the four Greens.  The 2002 Denison case is not useful because in that case the Greens did not recommend a ticket order for their minor candidates.  But in the 2006 Denison and 2010 Franklin cases, the Greens did recommend a ticket order.

2006 Denison was an example of what happens when the Greens have a very obvious #2 candidate.  Cassy O'Connor was high-profile mainly because of her role in the Save Ralphs Bay campaign, and also had some profile as a former TV journalist.  Of the votes from Peg Putt that went to another Green, 65.8% flowed to O'Connor, meaning that O'Connor would easily win Putt's recount (and in fact did).

2010 Franklin was at the other end of the scale.  Activist Adam Burling was not a very high profile name (although he was a local councillor at the time) and was a risky choice as endorsed #2 because of his hardline reputation.  Indeed, he was outpolled by ticketmate Wendy Heatley.  Of the votes from Nick McKim that stayed within the ticket just 36.4% went to Burling.  Had McKim resigned in the 2010-2014 term, it is not even completely clear that Burling would have beaten Wendy Heatley in the recount, although he would have started it with a large lead.  We'll never know.

Because there are no how-to-vote cards in Hare-Clark, it is very difficult to make party voters follow a certain order.  A specific order can be publicised in fliers or online, but voters often will not follow it.  The Greens ticket order, as shown by the screenshot below (click for larger version), was 1 Booth 2 Tyler 3 Dawkins 4 Povey 5 Layton-Bennett.

The questions are really:

* what proportion of Greens voters in Bass just followed the ticket as publicised?
* are there any reasons to think any of the other Greens candidates will do especially well on Booth's votes?

As for the first, it would be surprising if the flow was much weaker than the Franklin 2010 case, but Tyler's low profile at the election causes me to suspect it would at least not be much stronger.

As for the second, it seems from her higher primary vote and better preference performance that Dawkins had a slightly higher profile than Tyler or the others at the election.  However, the red herring here is that Dawkins is now a Launceston councillor so some have assumed that would help her in this recount.  It doesn't, because she was elected to Council after the state election.  Dawkins' slightly higher profile comes mainly from her running a Launceston cafe.

Overall I doubt Dawkins' advantage on profile is massive, otherwise she would have got more primaries compared to the other three.  I think it's most likely that although most Booth voters won't follow the ticket, enough will do so to give Tyler a solid advantage on the #1 Booth votes.  There is also a link between Booth and Tyler through Tyler being a staffer for Booth, that may cause Booth voters to preference her.

So who will win?

One of the four Greens will win the countback assuming some of them contest it, and I think it's most likely between Dawkins (slightly the higher profile) and Tyler (the #2 ticket candidate).  It's not impossible one of the other two Greens could get it, but I think they are less likely chances as there is no obvious reason why either might win.

At the moment I slightly prefer Tyler's chances (if she contests), but this is not a very confident assessment, and it would be really useful to hear of any scrutineering info.  I'll be waiting with interest to see who comes out ahead and by how much when the recount is done.

I understand that it will take at least ten days from Booth's resignation being formalised for the recount to be held.

Update (Thursday 21st): Brian Wightman has been reported by The Examiner as saying he won't bother nominating.  This doesn't affect the throwing of his votes as discussed above, but it does mean instead of going back to Wightman they would go back to whichever contesting candidate was highest on them.

ABC TV News reported they hadn't been able to contact Tyler and were not sure if she was going to contest.

And What Happens If Dawkins Wins? (May 22)

With mystery still surrounding whether Amy Tyler will claim her potentially winning golden ticket, but with Dawkins clearly a starter, I've had a lot of questions about what happens to Dawkins' Launceston City Council seat if she gets the job.

This gives me an opportunity for a sequel ("Fifty Votes Darker"?) to the above "#politas nerd erotica" and I promise readers this one will have you just as tied up in plot twists as the first.

In the event that Dawkins wins the Booth recount and takes the state seat, she then has one year to resign from Council, otherwise her Council seat becomes vacant.  Whether she resigns or it becomes automatically vacant, the outcome is the same: a countback similar to Booth's to determine her replacement, based on the votes she had when she was elected.

It would be easy (as with Rosalie Woodruff on Huon Council should Woodruff take Nick McKim's state seat) to just look at who the other Green candidates are and assume that one of them would get it.  But in Dawkins' case, the first complication is that although she ran as a Green in the 2014 state election and is now seeking to win a Green seat on a countback, in between those two things she was elected to local council as an independent.  This has no impact on her ability to run for Booth's recount and she does not even need to still be a Tasmanian Greens member to do so, nor to even serve as a Green if elected - it seems that just as the Upper House used to be stacked with closet Liberals, so also Green-turned-independent is a more successful path to a council seat than straight Green.

The second complication is that even if we assume Dawkins' Green background would rub off in the recount, there are no Greens for it to rub off on; the sole endorsed Green for LCC, Emma Williams, was elected.

There are thirteen candidates who could nominate for Dawkins' vacancy, and the votes distributed at the recount would be:

* 796 (46.4%) votes of Andrea Dawkins primaries.
* 839.73 (48.9%) votes worth from 11 of the 13 unsuccessful candidates and the surpluses of seven elected candidates, at the value they had when first distributed.
* 81.27 (4.7%) votes worth of votes from the candidate whose votes put Dawkins over quota, Karina Stojansek.  These votes were worth 304 votes, but are reduced in value in Dawkins' recount because most of their value went somewhere else in her surplus.

This disadvantages Stojansek in this recount, but that's arguably fair since her votes have been in part used to elect others.  The even less lucky candidate is Tony Peck, who was not only voted off after decades on Council but is also the only recontesting candidate who would not get any of his primary votes back (because he was still in the race when Dawkins was elected).  A #1 vote for any of the other twelve that was 2 for Dawkins makes it into the recount, but a 1 Peck 2 Dawkins doesn't.

(Unfair?  Yes.  Easy to find a better alternative to? Not at all. You're welcome to try - but please don't suggest redoing the whole count, since that can dis-elect people who won first time around, or result in a candidate being replaced by an opposing party who the candidate's party narrowly beat).

Looking at the cut-up the biggest source of Dawkins' votes was actually Kyle Barrett (200.41 votes), but Barrett is a prominent Launceston UTAS Young Liberal whose campaign was endorsed by Robin Gray!  Most likely a lot of these preferences are not Barrett's #1 votes but just votes that pooled with him through the cutup.  The same point applies to Barrett as applies to Dawkins in Booth's recount - if a vote is 1 Ellison 2 Barrett 3 Dawkins then while it appears to have come from Barrett, it actually goes back to Ellison in the recount.

None of the candidates seem to have a big advantage on the 921 votes that are not Dawkins primaries.  Someone (maybe Mandie Austin) might get slightly over 100 of them, but no-one would get that many more.  The advantage would not be large enough to stop someone with a large chunk of Dawkins primary votes from winning.  My suspicion is that despite the disadvantage of having her primary votes in the recount reduced in value, Stojansek could still get the seat.  This is based on her being one of the highest-polling defeated candidates, being female (female-to-female preference flows in local government are strong) and having a similar Green-linked-independent appeal, including as widow of the late Jeremy Ball.  If Stojansek didn't get it, the next most likely would probably be Austin, but I wouldn't completely write off Peck, Barrett (as politically absurd as that would be) or even others.

It's all academic if Dawkins doesn't win the recount, but there it is, anyway!


Amy Tyler May Not Nominate (June 1)

Georgie Burgess at The Examiner today reports that "The Examiner understands Ms Tyler won't nominate".  If Tyler doesn't run and Dawkins does then Dawkins should win the recount easily as she is both higher-profile than her ticketmates and next on the Greens' ticket after Tyler.  I wouldn't totally rule out one of the other two Greens, but think it's very unlikely anyone but Tyler would beat Dawkins.

Amy Tyler Confirms Not Nominating (June 5)

Amy Tyler has now publicly stated she is not running and considers Dawkins best placed to take over from Booth.

Other Greens Also Not Running (June 5)

The way is being paved for Dawkins to win the seat without real opposition with the other Greens being reported by the ABC as also not contesting.  If Dawkins is the only Green candidate then she will certainly win.

Recount Result (June 9)

Andrea Dawkins was duly elected today with an absolute majority against the one Labor, three PUP and one independent candidates bothering to contest the recount.  Dawkins' tally of 72.6% against that lot may seem a bit underwhelming, but the share of Booth's recount that came from Wightman was somewhat higher (1136 votes) than I expected.  In total, 23.5% of the Booth recount votes came to Booth from a non-Green source, and some of the votes coming from minor Greens candidates would have originated with non-Green candidates too.  Also, 20% of the votes that didn't go to Dawkins exhausted.

Dawkins Resigning From LCC Soon (June 20)

The Examiner on Thursday reported Andrea Dawkins will resign from Launceston City Council within a fortnight, triggering a recount for her LCC seat to be probably held in July.

Dawkins Recount Coming (July 3)

Andrea Dawkins has resigned from Launceston City Council and her recount, likely to be won by Karina Stojansek (see above) who has nominated.  I do not know at this stage which other candidates are running.  The button will be pressed shortly after midday Monday.  It will (unless the TEC decides otherwise on account of some last-minute issue) be a computerised recount using the electronic ballot papers from the 2014 data entry, and it won't take all that long at all.

I have been involved in scrutinising the recount system to make sure it can do its job, and while I haven't encountered any issues with the Computer Count implementation, there are certainly some issues with the Local Government Act electoral provisions that could do with some attention!  These aren't things likely to affect the recount outcome in this case as they are typically issues only capable of affecting fractions of a vote, or maybe the odd whole vote here and there.  Recommendations have been made for attention to these in the next review.  My report may receive a PS dealing with the Dawkins LCC recount and after I've finalised it I may post a link to it here.

All this said I have no special knowledge of the contents of the recount so I don't know anything special about who will win based on the work I've been involved in.

Dawkins Recount Done (July 6)

The button was pressed on the Launceston recount today and I witnessed this spectacle.  As expected despite the disadvantage of having a large pile of her votes devalued for the recount, Karina Stojansek was the winner.  The method of victory (the following is all subject to official confirmation) was slightly surprising though with Krista Preece, who had been fourth excluded with just 1.1% of primaries in the original election, putting in a big challenge.

After the electronic throw of all Dawkins' votes Stojansek 406.06 led Austin 252.08 Preece 248.24 Barrett 190.25, Jones 177.67,  Brady 143.50, Elliott 138.42, Harris 109.43.  The other four eligible candidates didn't contest.  Peck, who was most disadvantaged by the "Hare-Clark recount bug" (votes that were 1 Peck 2 Dawkins or similar were not included in the recount because Peck had not been excluded at that stage) was actually first out with just 51.36; even without the recount bug it seems his support among Stojansek voters was very low and he wouldn't have been competitive.

After the exclusions of Peck, Harris, Elliott, Brady, Barrett and Jones, Preece had passed Austin and Stojansek still led Preece by over 100, but the preferences of Austin allowed Preece to close to within a final margin of just 11.89 votes.  So voters who voted 1 Austin 2 Dawkins, 1 Dawkins 2 Austin or similar strongly preferred Preece over Stojansek.

In the old manual recount system this would have been a day-long count and possibly scrutineers would have called it for Stojansek early and then had some nervous times late in the count as Preece caught up.  Today with the press of the button there was no such drama and it was all over in about ten minutes.


  1. Fantastic! The countback process makes sense, but it makes me wonder how on earth they count votes at less than full value in a general election, as they scatter and recombine over a dozen or more eliminations and surpluses. I guess it involves calculators, lots of sticky notes and some very long tables. Kudos to the TEC. Sweet democracy.

    Thanks for the very helpful analysis, will be interesting to see how it plays out and if there are any insights to be learned about Green voting behaviour, particularly since we might have a second countback on the way. An Amy Tyler victory would be a massive coup for the Greens, people like Amy are well and truly the future of the party. Nothing against the other candidates, I would just be ecstatic to see Amy become an actual real-life MP. Fingers crossed.

    1. Apologies for the delay in publishing the above comment, the email notifying it wound up in the wrong place.

      The process for general elections is largely computerised except that the votes are still (at state elections) counted and transferred manually. In Hare-Clark there are actually not many transfer values - each surplus creates one new one. It is different to some systems where there are constantly new transfer values being multiplied by each other until you end up with hundreds of them.

  2. Great analysis Kevin! Thanks for sharing it. :)