Posts by Steve Todd
Last ←Newer Page 1 2 3 4 5 Older→ First
-
Hard News: Art with a job to do, in reply to nzlemming,
I know what you're saying, nzlemming, but I can't help myself. I will advocate, explain, and defend PV / STV at every opportunity.
-
Hard News: Art with a job to do, in reply to Russell Brown,
Thanks for that, Russell. I appreciate you taking the time to respond.
-
Hard News: Art with a job to do, in reply to Jason Kemp,
Jason, as this blogpost is about art and design, rather than the workings of the PV voting system, I will make this response, then leave it at that.
Firstly, I apologise to you for my terseness last night. I had spent most of Saturday explaining and defending PV on Kiwiblog, and I was pretty fed up (even though it was nearly 24 hours later).
Your reply indicates to me that you think preference-rankings work the way multiple-FPP works, each ranking – 1, 2, 3, etc. – is a vote, each vote having the value of unity. That is not the case. Your single vote was your first-preference for Red Peak. Any other preference-rankings you gave were not votes, they were contingency choices only, which would only come into play, in successive order, should, firstly, Red Peak be excluded from the count.
If your preference-rankings included both the Black and Red Lockwoods (say, 3 and 4, respectively), then, when Red Peak was excluded, your vote would have transferred to Black Lockwood at Count 4 (your second preference having already been excluded from the count), to become, in effect, an additional first-preference vote for that option.
While 1,376,544 people cast a valid vote (as 7 p.m. on Friday) in the referendum, only 1,310,455 people indicated a preference between the two Lockwoods. In the contest between what turned out to be those two remaining options, 662,160 people voted for Black Lockwood (including yourself, as per the preceding paragraph) and 648,295 people voted for Red Lockwood.
The second and later preference-rankings behind the 1,127,191 first-preference votes given for the two Lockwoods were never looked at. It is completely irrelevant that most of those contingency choices might have been 2nd and 3rd preferences for the other Lockwood and the Silver Fern (Black and White). To be clear, those voters did not block-vote the two Lockwoods and Silver Fern (B and W) to shut out Red Peak.
I agree with Sacha, Lilith, Simon Pound – thanks, Sacha – and George Darroch that the process adopted to choose the four alternative options was, shall we say, less than optimal, for the reasons they give. (Call me naïve, but I refuse to believe a group of people, comprising “the great and the good”, would participate in a politically-orchestrated set-up.) In addition, just for the record, in my view it was a disgrace that Red Peak was added to the panel’s four choices, which is the main reason why I ranked it at No. 5. (Another reason is, it simply does not say ‘New Zealand’ to me.)
I think there should have been seven completely different designs to choose from, including no more than two silver fern designs, and perhaps as many as three Māori-specific designs. (I share Michael Meyers’ concern about “the very low voter turnout in the Māori electorates”. Clearly, most Māori felt they did not have “skin in the game”. If we do this again, that needs to be sorted, including there being Māori designers on any panel of designers.) Had that been the case at this referendum, however, it is quite possible that almost all of the 1,204,793 people (87.52%) who gave their first-preference vote to one of the three fern options, would have simply split their votes between the two completely different fern options, with one of them still winning.
You say, “and the 2 designs which were not ferns might have got more preferences.” You meant more first-preferences. (There could have been as many as one million second or later preferences for each of Red Peak, Koru (and Silver Fern (B and W)) tucked in behind those 1.1 million first-preference votes for the two Lockwoods. I guess we’ll never know (sigh).) Maybe, but it is clear that the overwhelming majority of voters wanted a fern design.
If there had been one fern design (presumably one of the Lockwoods), then one of the *four*, not two, non-fern designs (including Red Peak) would have had to be pretty amazing for the support for Silver Fern to collapse sufficiently to enable it to come through and attain 50% of the votes. Clearly, in this case, the people decided the two non-fern options we got to consider, weren’t.
Okay. That’s it! Let the discussion of art and design continue.
-
Hard News: Art with a job to do, in reply to Jason Kemp,
"I voted for Red Peak but the way preferences work it was not a surprise to be beaten."
What on earth is that supposed to mean, Jason? Do you actually understand how the Preferential Vote works? Have you actually had a look at the preliminary result at www.electionresults.org.nz?
The reason why Red Peak was unsuccessful, was simply because it did not receive enough votes to win. Even if every vote given for Koru (51,879) and for Silver Fern (Black and White) (77,802) had transferred to Red Peak (119,672), upon their exclusion from the count, Red Peak would still only have received a total of 249,353 votes, well short of the 552,827 first-preference votes received by Silver Fern (Black, White and Blue), not to mention the 574,364 first-preference votes received by Silver Fern (Red, White and Blue).
In other words, Red Peak was always going to be the third option to be excluded from the count. Your comment reproduced, above, implies that the vote-counting system worked against Red Peak. Nothing could be further from the truth.
-
@mpledger
"You only need rank 4 flags – the 5th rank never comes into play."
Correct. That is because when the most successful losing (in this case) option is excluded, the votes given for that option will not transferred to the next preferences shown on them, because there is no need to do so - a winner has been found.
I'm perfectly happy to say, though, that I'm going to go the whole hog and put a 5 beside Red Peak, as my protest at that design being included in the ballot. (I'm doing that even though I know Graeme is right - it won't be seen as a protest). That inclusion was a slap in the face to the panel (who, I must also say, did a pretty ordinary job of coming up with some decent options for us to consider - which should all have been completely different).
-
Hard News: #GE2015: Proper Mad, in reply to izogi,
But assuming they can comfortably do so, is there some reason why the show ponies wouldn’t simply resign, and let in the next council candidate, if they weren’t also elected mayor?
Show ponies who weren’t also elected mayor could indeed “simply resign”, but that would not “let in the next council candidate”. Under the current local electoral law, it would result in a by-election. The council and ratepayers would not be happy about that. (Remember, under my proposal, the councillor vote-count is carried out first. Currently, it is the other way round.) In addition, the show pony will be seen as having duped the voters, thereby almost certainly damaging his or her future political ambitions.
Furthermore, candidates for council campaign for votes by stating what they intend to do for their ward / city or district. If they declare that they will promptly resign from council if they do not also win the mayoralty, they will be seen not only as holding the voters to ransom, but also as not being genuinely committed to “serving the people” / the best interests of the city or district. They would rightly be seen as just wanting to “trough” it; of merely wanting to further their own selfish ambitions. Under such circumstances, they would, in all likelihood, not be elected to council and therefore would be withdrawn from the subsequent mayoral vote-count.
So, yes, show ponies might well act in the way you suggest, first or second time round, under the new system, but perceived adverse political consequences would prevent such actions, or result in a decision not to put oneself forward for election. I’m quite sure that, in time, the system would settle down and operate in the manner suggested in my previous post – well, that would be my expectation anyway.
You have also obliquely alluded to the issue of how councillors elected by STV should be replaced when vacancies occur.
Under STV, there is actually no need for by-elections; vacancies should be filled by a recount of all the votes cast, with the other candidates who were elected at the original election being guarded (against defeat). (The STV calculator is not currently programmed to perform such recounts, but could easily be so.) This is essentially what you are suggesting when you say “Especially if STV makes it easy for voters to have indicated their next preferred council candidate anyway.” (It is not sufficient simply to “let in the next council candidate”; to provide for the runner-up candidate at the original election to be appointed to the vacancy, because under the transferable voting system that STV is, the runner-up candidate may not be the one the voters would have elected instead, had the vacating member not stood at the original election.)
In this regard, look at what is happening in Palmerston North right now. A sitting councillor (Grant Smith) was elected mayor in a by-election to replace Jono Naylor (now a List MP). However, the election of the 15 councillors in October 2013 was city-wide (at-large), not from several wards. That meant, under STV, the quota for election that each successful candidate had to attain, was 6.25% of the votes cast (votes / 16). (For the record, well done Palmerston North.)
In the councillor by-election to replace Smith that is currently under way (and which concludes next Saturday, the 16th), the successful candidate must attain 50% of votes cast, plus 1. In other words, a candidate elected by a majority will replace a candidate elected by proportional representation. That could well mean that a candidate is elected next Saturday who is politically opposed to Smith, thereby frustrating the will of the people at the original election in 2013.
A recount of the votes cast in the original election would ensure that a candidate elected by PR was replaced by a candidate who is elevated to council under the same PR conditions. Should that candidate be politically opposed to the vacating member, then so be it. After all, the replacement councillor is highly likely to be the candidate who would have been elected at the original election had the vacating member not also stood.
-
Hard News: #GE2015: Proper Mad, in reply to Moz,
Preference distributions are one of those apparently simple things that leads to chaotic behaviour (in the mathematical sense). The problem is that in the senate exactly which votes get allocated to a candidate and which get distributed matters, and some systems allocate fractional votes and some don’t, resulting in a different outcome for the election.
Moz, the various STV systems used in mainland upper house elections in Australia, are all very poor examples of STV; unworthy of the name, in fact. Were STV to be adopted to elect the UK House of Commons, it is very unlikely it would resemble Australian Senate STV.
Any true STV method would not include above-the-line voting, which turns STV into little more than Party List PR. I would envisage a Tasmanian lower house or ACT-style ballot paper being used, with fully optional preferential voting, as in Ireland, Northern Ireland and Scotland.
I would hope they would use Meek's method to count the votes (as in NZ STV). That would eliminate all the horrors of Australian-style STV that you obviously know so well, and would give results exactly in accordance with the voters' wishes.
-
Hard News: #GE2015: Proper Mad, in reply to Rich of Observationz,
We should elect our councils by an MMP type to elect councillors followed by the indirect election of a mayor by a majority in council (as the PM is “elected” by a majority in parliament).
Rich, MMP is entirely unsuitable for local elections in New Zealand, for the reasons given 20 months ago.
At that time, I agreed with you that mayors should be elected by the councillors, but I have since had a re-think. While I cannot tolerate mayoralties being filled by show ponies and national politicians looking for something to do post their Parliamentary careers (and using their high(er) profiles to practically walk into the job), I still believe mayors should be elected by the people.
A compromise, that is only suitable in respect of councils where STV is used, is to require all mayoral candidates also to stand for council. Once the councillor results are known, those mayoral candidates who were not elected to council, would be withdrawn from the mayoral election and their first preference votes transferred to the second preference candidate on each voting document. The mayoral count then takes place in the normal manner.
Under this proposal, councils would not consist of the mayor and x councillors; they would consist of x councillors, one of whom was also elected by the people to be the mayor. This proposal would likely result in show ponies and ex-MPs (who only want to be mayor) baulking at standing for mayor, because, in most cases, they will have no interest in what they would regard as the consolation prize of being a councillor, should they be unsuccessful in their tilt at the mayoralty.
It would also ensure that, in almost all cases, mayoral candidates will be sitting councillors who will likely have spent a triennium or two on council and will know the ropes before deciding to contest the mayoralty.
The reason why this proposal would not be suitable for FPP-elected councils, is that councillor votes given for mayoral candidates who were not elected to council would be wasted, in that they could not be transferred to another mayoral candidate. This would mean that those voters in mayoral elections who voted for a candidate not elected to council, would not have their mayoral vote counted. That, of course, would be grossly unfair, not to mention undemocratic.
-
Further to my immediately preceding post, I thought I would back up the assertions I expressed there with some figures. I have now had a good look at the final results for the 8-seat Henderson-Massey Local Board and 7-seat Waitemata District Health Board elections. My comparison of these elections follows.
(Warning: This post is far longer than I imagined would be necessary to present a meaningful comparison, but, as something like this has not been done before, it might nevertheless be of interest to those interested in electoral systems generally, and STV in particular.)
In the Henderson-Massey board election, there were 72,461 electors, of whom 22,599 (31.19%) voted. Of those who voted, 197 cast an informal vote, and 653 left this issue blank, leaving 21,749 who cast valid votes.
Those 21,749 voters cast a total of 146,582 votes for the 28 candidates, an average of 6.74 votes each. The eight successful candidates received a combined total of 60,449 votes (41.24%), and the 20 unsuccessful candidates received a combined total of 86,133 votes (58.76%).
The highest polling successful candidate (Linda Cooper, who was elected to the governing body of the Auckland Council and who therefore cannot take up her seat on the board – section 88A(2) of the Local Electoral Act, via section 8(1) of the Local Government (Auckland Council) Act 2009) received 9,490 votes (6.47%), and the lowest polling successful candidate received 6,558 votes (4.47%). The highest polling unsuccessful candidate (the subsequently-elected Chris Cooper) received 6,450 votes (4.40%), closely followed by incumbent board member Tracy Kirkley, with 6,430 votes (4.39%).
(Note: I am including Linda Cooper’s votes in this analysis, on the basis that those electors who gave her one of their votes, did so without knowing their vote would not count.)
The combined majorities of the eight successful candidates (being the total of votes for each successful candidate, less 6,451 in each case) summed to a total of 8,841 votes (6.03%). Therefore, a total of 94,974 votes (64.79%) – 66.22% in 2010 – were ineffective in helping to elect any of the candidates.
In addition, it will be noted that, as is very common with FPP elections (and multiple-FPP elections, such as this one, are no exception), the eight successful candidates were elected on considerably less than 50% of the total of votes cast; on just 41.24%, in fact (46.41% in 2010). (For the record, if the figures are adjusted to take account of the fact that Linda Cooper is treated as having vacated office as a member of the board, the eight candidates actually elected received 41.88% of the total of votes cast – 57,409 / 137,092.)
So, not only do the people of the Henderson-Massey board area continue to be represented on their local board by members enjoying minority support, but they are also unequally represented.
In this regard, the successful candidates received varying numbers of votes, ranging from 9,490 down to 6,558, a difference of 2,932. In addition, some electors will be represented by several members for whom they voted (who could well be in political opposition to each other), some by one or two, some by six, seven, or even all eight, and, of course, many will not be represented by a single candidate they voted for.
This unstable state of affairs is brought about, of course, by electors having too many votes (each of equal value, the value of unity) and using them, to the tune of an average of 6.74 each. It is this random nature of multiple-FPP elections that will have contributed to Tracy Kirkley’s narrow defeat in the lottery that was this particular election.
Compare all this with the election for the Waitemata DHB, conducted by single transferable voting (STV).In this election a total of 111,590 valid votes were cast for the 35 candidates. The initial quota for election, therefore, was 13,948.750000001 (111,590 / 8, plus one one-billionth of a vote), being 12.50% of the total of valid votes cast.
The seven successful candidates received a combined total of 69,880 first preference votes (62.62%), and the 28 unsuccessful candidates received a combined total of 41,710 first preference votes (37.38%).
(The figure of 62.62% is a little low, because of the large number of candidates – more than for any other DHB election this year. The corresponding figure in respect of the Capital & Coast DHB election, for example, where there was only 23 candidates, was 69.24%.)
First preference votes received by the successful candidates, and for the runner-up candidate, were as follows—
Max ABBOTT 14,990 (13.43%)
Christine RANKIN 12,297 (11.02%)
Pat BOOTH 11,976 (10.73%)
Sandra CONEY 9,010 (8.07%)
Warren FLAUNTY (elected 6th) 7,602 (6.81%)
James LE FEVRE (elected 5th) 7,269 (6.51%)
Allison ROE 6,736 (6.04%)John TAMIHERE (Runner-up) 6,201 (5.56%)
The next three most popular candidates (on the basis of first preference votes received) were Mary-Anne BENSON-COOPER 3,587 (3.21%), Tracey ADAMS 3,472 (3.11% – 25th to be excluded), and Brian NEESON 3,414 (3.06% – 26th to be excluded).
It will immediately be noted that, even before Abbot’s surplus of 1,041.25 votes is transferred, and then, subsequently, the votes of the lowest polling candidates, six of the seven successful candidates have already received a higher percentage of the votes cast than was received by Linda Cooper in the Henderson-Massey board election.
It will also be noted that the seven candidates leading on the count of first preferences, were the seven candidates who were eventually elected (with one change of position), the reason for this being pointed out in my immediately preceding post.
The surplus votes of the successively elected candidates, and the votes of the successively excluded candidates, were transferred to continuing (still hopeful) candidates in accordance with the preferences indicated on the relevant votes, through 45 more iterations, until the final outcome was produced.
At the completion of the count, a total of 104,400.509359888 votes remained in the election, 7,189.490640112 having become non-transferable. The votes that had accumulated upon the seven successful candidates throughout the count amounted to 95,426.617111869 (91.40%). The remaining 8,973.892248019 votes (8.60%) had accumulated upon the runner-up candidate.
The votes for the seven successful candidates at the final iteration all exceeded the final quota (13,050.063669987 – 104,400.509359888 / 8), which, of course, they must do. The votes, plus those for the runner-up candidate, were as follows—
Max ABBOTT 13,445.249362985
Christine RANKIN 13,659.674789627
Pat BOOTH 13,596.379644577
Sandra CONEY 13,744.199079694
James LE FEVRE 13,740.341465062
Warren FLAUNTY 13,663.635514684
Allison ROE 13,577.137255240John TAMIHERE 8,973.892248019
The final total of surplus votes, being the final votes of the seven successful candidates less the final quota, in each case, amounted to 4,076.171421960. As there was no point in transferring this surplus, because the seven winners had been found, it cannot be said that the surplus votes of each of the successful candidates were ineffective in helping to elect them.
Hence, the percentage of *effective* votes in this election is the aforementioned 91.40 (rather than 87.50 (12.50 × 7)), compared to 35.21 in the Henderson-Massey board election (100 – 64.79). *Ineffective* votes were those that accumulated upon the runner-up candidate, amounting to 8.60%, compared to 64.79% in the H-M board election.
In addition, the people in the Waitemata DHB area are *equally* represented on their DHB, because the successful candidates all received approximately an equal number of votes. Furthermore, the successful candidates are all of *equal* status, because they each attained the required (final) quota of votes.
In my view, the superiority of STV over FPP is clearly demonstrated by the comparison I have made of these two elections.
Finally, while the number of first preference votes received by each candidate is certainly a measure of their popularity, the more significant measure of each successful candidate’s popularity is their final keep value.
Therefore, for the record, the final keep values for the successful candidates, and for the runner-up candidate, in descending order, are as follows—
Max ABBOTT 0.710716544
Pat BOOTH (elected 3rd) 0.737846231
Christine RANKIN (elected 2nd) 0.751025633
Sandra CONEY 0.860048524
James LE FEVRE 0.927156656
Warren FLAUNTY 1.000000000
Allison ROE 1.000000000John TAMIHERE 1.000000000
The final keep value for Warren Flaunty and Allison Roe, both of whom were elected at the final iteration, and for John Tamihere, the runner-up candidate, is 1.0 in each case, because their votes were not transferred.
What these figures mean is that Max Abbott, for example, needed to keep only 71.07% of all the votes he received during the count, in order to be elected, whereas Warren Flaunty and Allison Roe needed to keep 100% of the votes they received.
While Christine Rankin was elected second (at iteration 30) and Pat Booth was elected third (at iteration 32) (and indeed received 321 more first preference votes than he did), Booth ended up with a slightly lower final keep value, indicating that he had slightly broader support than her across the entire DHB area.
-
Legal Beagle: Fact check: Q+A on mayoral…, in reply to Andrew Geddis,
Thanks for that, Andrew.
I see that Dave Macpherson’s 2953 votes were not regarded as void (confirming that there is no longer any provision to void FPP votes given for candidates who retire from the election after the close of nominations but before the close of voting).
I note, too, that candidates who retire from an election by STV are no longer regarded as having withdrawn (see clause 22 (as amended) of Schedule 1A of the Regulations), which means they are excluded from the count (if and when that becomes necessary), and their (reduced number of) votes transferred in the normal manner (as Bryon Burke’s votes *incorrectly* were in the 2004 Wellington mayoral election).