Posts by Steve Todd
Last ←Newer Page 1 2 3 4 5 Older→ First
-
Hard News: Footpaths, not manifest destiny, in reply to Keir Leslie,
I have some comments to make in response to your post, Keir, but it will have to be tomorrow evening now. Other things to be getting on with this evening.
-
Hard News: Footpaths, not manifest destiny, in reply to Rich of Observationz,
I forgot to mention that while UKIP should have got one seat (in my view), there was an arbitrary 5% threshold that the party did not overcome. Very unfair, considering the natural threshold was 4%.
-
Hard News: Footpaths, not manifest destiny, in reply to Rich of Observationz,
Not quite, Rich. Under my scheme, a group getting 40% of the vote in a 7-seat ward would be guaranteed 3 seats (3 x 12.5% = 37.5%).
In New Zealand local elections, we have either either FPP or STV. As I've said to you before, if you want something else, you've got a heck of a lot of work ahead of you - a good 10 years of hard slog, which would almost certainly be in vain. I believe all local councils using STV should be elected at-large, but I realise that is a step too far for Auckland at this time. Each successful (AK) candidate would need only 4.76% of the city-wide vote to be elected (100 / 21 = 4.76%) - not bad.
What would your "genuinely proportional system" look like? Straightout Party List, elected city-wide? Each group would gain an extra seat with each additional 5% of the votes attained (no doubt having to overcome a 4% or 5% threshold), but electors would have no say over which individual candidates would be elected. At the local level, that would not be a goer.
London Assembly-style MMP, perhaps? In those elections in 2012 (next one on 5 May), 25 members were elected - 14 from single-seat FPP constituencies; 11 city-wide from party lists. Labour and Conservative candidates are the only ones ever to win the FPP seats. In the current Assembly, Labour has 12 seats on 41.14% of the vote (PR, only 10); the Conservatives have 9 seats (31.99% - PR, only 8), the Greens have 2 seats (8.54% - PR, 2) and the Lib Dems have 2 seats (6.79% - PR, 2). UKIP got 4.52% of the vote, but did not get a seat, not even one of the 2 remaining true PR seats - which they were entitled to. Labour and the Conservatives ended up over-represented. Is that any better? I think not.
So, why not let us have a reasonably comprehensive description of your preferred PR system for local government elections, so we can compare it with properly-implemented STV (not the inadequate 3-seat wards (and one 2-seater) system we have in Wellington).
-
Hard News: Footpaths, not manifest destiny, in reply to Phil Lyth,
Yes, Phil, you’ve nailed it in one. The adoption of STV in multimember wards is clearly the way to go in Auckland. I would be very disappointed if Simon Wilson has discussed the problem of effective local representation in Auckland City without mentioning STV as (at least part of) the solution.
From 2019, the Auckland Council could be elected from two 7-seat wards and one 6-seat ward, using the local board population figures set out at https://en.wikipedia.org/wiki/Auckland_Council. The population of Auckland City was estimated to be 1,570,660 as at 30 June 2015. Each councillor therefore represents an average of 78,533 residents. That being the case, the three wards could be something like this—
North-western
Rodney/Albany/ North Shore/Waitakere – population 539,000 / 78,533 = 6.86 councillors
Isthmus
Whau/Waitemata and Gulf/Albert-Eden-Roskill/Orakei/Maungakiekie-Tamaki – population 512,660 / 78,533 = 6.53 councillors
South-eastern
Howick/Manukau/Manurewa-Papakura/Franklin – population 519,000 / 78,533 = 6.61 councillors.
While this would give 7, 6 and 7 councillors, respectively, it is obvious that fairer representation would be achieved by each ward electing 7 councillors.
Three 7-seat wards would enable all significant political factions in Auckland, particularly C&R, Auckland Future, Labour and the Greens, to obtain fair and equal representation on council. Each successful candidate would only need to attain 12.5% of the votes to be elected, which would ensure that each ward would be represented by councillors from at least three, but most likely all four, of the above-mentioned factions. (Under this system, there would be no need for C&R or Auckland Future to “make way” for the other.) In addition, at least one independent candidate would almost certainly be elected from each ward.
Also, under this system, there would be no safe seats for individual councillors. If a councillor representing one of the four factions mentioned disappoints voters, that councillor could be replaced with someone else from the same faction. There would be no need to vote against the faction to get rid of the errant councillor, as is necessary under FPP.
I’ll leave it at that (for now); I just wanted readers to see what a proper electoral system in Auckland could look like.
-
Hard News: Art with a job to do, in reply to izogi,
"could have *been* flown together".
-
Hard News: Art with a job to do, in reply to izogi,
"Goal achieved, I guess."
I suspect you're right, izogi.
This process could have been carried out so much better than it has been. A proper panel of vexillologists / designers, charged with coming up with seven (in my view) completely different designs, could have been assembled, for starters. Then, as George Darroch says, the alternative designs could have flown together – I say, with the current flag – all round the country (for the next two years).
Then, at the 2017 general election, a one-off referendum (by PV) of all eight flags could have been held. First to 50%, wins.
As you say, too late now.
-
Hard News: Art with a job to do, in reply to tussock,
Wow, tussock. Straight back at me, eh?
"There is no golden rule, you’re making that up. There’s a set of axioms that seem desirable for any particular vote counting method to have, and none of them can cover all such desires at the same time."
I guess you're right. It's *my* golden rule. The rest I agree with.
I covered the question of compromise candidates in my responses to izogi. We'll just have to agree to disagree on that one. We're talking public referendums / elections here, not (say) elections to elect a committee of a mathematics society.
I would have preferred to see the current flag in the mix, too. But, again, we're talking public referendums here. Apparently, the argument was, that people would not want to change the flag unless they knew what flag they were changing to.
To me, the question should always have been something like, "Which of these flags [including the current flag] should be New Zealand's flag?" Those who didn't want change would have voted (with their single vote, which is transferable) for the current flag. First flag to 50%, wins.
I'm not troubled either way. Those people who want to keep the current flag, will have the opportunity to say so in March. It's all good. In the great scheme of things, the cost is neither here nor there really.
-
Hard News: Art with a job to do, in reply to izogi,
Izogi, I implied I’d get back to you today. Here goes—
(I’ve overdone it again, people. Sorry.)
“From your tone, does that mean there’s already precedent for 50(A) to take priority over the OIA if tested in court?”
Extremely unlikely. The Act has only been in force since 15 August. I was just expressing my opinion.
I’ve had a quick look through the ‘Offenses’ provisions of the Electoral Act 1993 and the Local Electoral Act 2001, but could see no equivalent provision to that of section 50 “Unauthorised use or disclosure of referendum information” of the New Zealand Flag Referendums Act 2015.I’m no lawyer, but, as I say, the wording is very clear to me – preference data will not be released, regardless of (say) section 5 of the Official Information Act 1982. That section says, in part, “information shall be made available unless there is good reason for withholding it.” Perhaps the presence of section 50 is good reason to withhold the release of preference data. Given Eric Crampton’s OIA request, perhaps time will tell.
In this day and age, such data should automatically be made available online. It should also be in a format such that other people can take the information, run it through their own ‘STV Calculators’, and confirm the official result for themselves. If they want to use the data to carry out pairwise comparisons (which, in this case, might have revealed, amazingly, that the Condorcet winner was Red Peak), so what? Condorcet Pairwise Counting is not PV / STV. Condorcet counting can violate later-no-harm (and later-no-help); PV / STV cannot. And, as both tussock and I have pointed out, PV will almost always elect the Condorcet winner, if there is one, in any event (as it almost certainly did at this referendum).
Given the importance of the flag referendums process to the government, it is possible that section 50 was inserted specifically to ensure that no-one could take the preference data and use it to undermine the official result of the referendum, by showing that PV produced the “wrong” winner (we know it produced the right winner, as it always, without exception, does) / embarrass the government. It was in the original bill, so it was not inserted as a result of submissions received.
I realise you have conceded that Condorcet counting “would not be feasible in practice”, but I’ll give you my thoughts, anyway.“Option Z is the second choice of every single voter, and so is eliminated immediately.”
And rightly so, in my view. Why should option Z win the election when no-one, using their *single* vote (which is transferable) voted for it? No other electoral system in use in public elections anywhere in the world sets out to find, and elect, a compromise candidate (if there is one). FPP certainly doesn’t. Why should PV / STV be required to do so?
The purpose of elections is to enable people to elect their representatives, not to arbitrarily discard candidates who “stand for something” in favour of *compromise* candidates. Imagine if your scenario were able to be played out.
On the six o’clock news next October, a po-faced Wendy Petrie informs the nation that candidate Z, who received not one single vote, out of 60,000 votes cast, nevertheless romped home to win the Wellington mayoral election.
When asked if s/he was embarrassed by his / her win, s/he replied, “Not in the least. Why should I be?”
“Um, well, no-one actually voted for you.”
“Ha! Votes, shmotes. Fortunately, our voting system is so robust, so flexible, it was able to reveal and give expression to the collective view of the good people of Wellington, that I was the only candidate qualified to take this great little city into the amazing future that awaits it.”
(NB. No suggestion, and certainly no admission, that s/he was merely the “compromise” candidate.)
Our shocked – and now formerly intrepid – reporter shuffles off, shaking his / her head in stunned amazement.
In 2018, the council would vote 14–1 to return to FPP for the 2019 and 2022 elections.
“My main gripe is with people declaring how other voters think without actually having real data to justify what they’re saying.”Speaking for myself, it is just my opinion that around 80% of the Lockwood voters would have given their second preferences to the other Lockwood. It just seems entirely logical to me that that would be the case. I’m certainly not “making a wide-ranging assertion”, about how voters think.
-
Hard News: Art with a job to do, in reply to izogi,
Izogi, have a look at my exchange with Eric Crampton at his blog, about 5 weeks ago, following his op-ed in the Dom Post of Friday 6 November, and my letter in response, to the editor, of Monday 9 November, here: http://offsettingbehaviour.blogspot.co.nz/2015/11/condorcet-conundrums.html#disqus_thread.
I lay out my case against Condorcet counting, but he never actually engaged with me, finally asking me a question which was nothing other than pretentious nonsense. I realised I was wasting my time, and let it slide.
I'll think about your other questions / observations tomorrow.
-
Despite saying to Jason Kemp, I'd "leave it that", there has been little or no discussion of art and design since my last post on Monday, so I’ve decided to respond to some of the comments that have been made over the last 48 hours. But first—
To mpledger—
I agree the process concerning the choice of options was not good, but, whether good or bad, whenever you have a number of options to rank-order, it would be extremely unlikely that they would all attract a roughly equal number of backers. It can happen when a political party puts up several candidates in a party column in a multi-seat STV election (e.g. Tasmania), and the names are randomised/rotated, thereby helping to even out the allocation of preferences, but unlikely here. In this case, no matter how good the various designs might have been, some would have appealed, some wouldn’t have.
Also, it is doubtful that one flag would “run away with it.” That implies that one flag would be “everyone’s second preference”. Although that concept is beloved of opponents of PV / STV, and social choice theorists (particularly those in US academia), it is just as unlikely in practice, in my view, as a particular flag being everyone’s first choice, or everyone’s third choice, etc.
To izogi and Rich of Observationz—
The votes were pre-processed (by being scanned) as they were received throughout the three-week voting period (see section 42 of the Referenda (Postal Voting) Act 2000). After 7 p.m. on the 11th, they were sent to the STV Calculator, which produced the preliminary result. So, the electronic preference data is there, and could be extracted. Everything is kept for 6 months, then destroyed (section 40(2) of the New Zealand Flag Referendums Act 2015).
I reckon, of those who voted for Black Lockwood, 80% or thereabouts would have given their second preference for Red Lockwood, and vice versa. I would love to know the real percentages. Eric Crampton, Head of Research at The New Zealand Initiative in Wellington, is putting in an OIA request, to test whether the Electoral Commission can lawfully invoke section 50(1) of the Act, to withhold electronic preference data. The wording seems pretty clear to me – we are never going to know anything more than what we know now.
To Hilary Stace—
How you choose to vote in local elections is entirely your business, of course, but I always recommend people rank as many candidates (in their true order of preference) as they genuinely can. If, for example, there is one seat left to fill, and two candidates remaining, both of whom I dislike, I will still indicate a preference for one of them over the other, if I can (and, except in DHB elections, I always can), to ensure my vote has as much say as to the make-up of council as possible. If, by ranking as many candidates as I can, my vote can help make the council, say, not-too-bad rather than bad, then I’ve done myself a favour.
To Rich of Observationz—
As I have said previously, I ranked Red Peak at No. 5. I could have left it blank (which is indeed equivalent), but actually putting the 5 under that flag was my protest at it being included in the list of options.
I know you will know this, but I’ll give my reasoning anyway. Basically, a vote cannot transfer to a last-ranked option / candidate, when in single-winner or multi-winner elections / referendums, because the count stops as soon as the winner(s) is / are found. Take the case of those voters who ranked Red Lockwood 4 and Black Lockwood (the eventual winner) 5. When Red Peak’s votes were distributed 57,631 to BL and 48,396 to RL, putting BL past 50%, the count terminated. While RL was then formally excluded from the count and BL was declared the winner, the fifth preferences for BL on RL’s voting papers were not transferred to BL, because there was no need to do so – BL already had the required 50%+ of the votes.
To izogi—
“So Lockwood/Black might have won by combining second-third-forth preferences of the alternatives, but theoretically another of the alternatives might have overtaken it if second prefs of Lockwood/Red had been attributed to its first prefs.”
But no circumstances could possibly arise whereby Red Lockwood could be excluded from the count before the three that were. If the counting system had incorporated pairwise counting, to ensure the Condorcet candidate (if there was one, which there was – Black Lockwood) was found, it could have shown another flag (say, Red Peak) beating Black Lockwood, but that would break the golden rule of preferential voting – later preferences can never harm earlier preferences. The Red Lockwood voters would not have indicated a second (or later) preference for Red Peak if those preferences were going to be used to defeat Red Lockwood. And, as I’ve said elsewhere, we would then be back to FPP.
You’re right, there is no perfect voting system, but there are none better than PV/STV.
“My own view is that I might as well rank all options according to the order I'd actually prefer them, unless I genuinely don't have a preference between them.”
That’s exactly my view, too.
To tussock—
“Obviously we should run a count that finds the Condorcet winner if there is one, but STV will find the Condorcet winner in almost all real single-winner elections anyway.”
Absolutely not. See my reply to izogi, immediately above.
Of course, because we will never know the second and later preferences sitting in behind the first-preference votes for Red Lockwood, we cannot be 100% certain that Black Lockwood is the Condorcet winner, but we can be 99.99% certain – well, I can. As I say, above, it is almost certain that the overwhelming majority of Lockwood first-preference voters would have given their second-preference for the other Lockwood. Therefore, Black Lockwood (which defeated Red Lockwood in a pairwise count) would almost certainly have beaten all the other options in a pairwise count, as well.