Posts by Steve Todd
Last ←Newer Page 1 2 3 4 5 Older→ First
-
Again, agreed. This is what Andrew Geddes suggested people might want to do, but I've heard that he actually confused people. Also, he talked about making sure you don't help the ones (plural) you don't like, whereas the "numbering all candidates" ploy only ensures you don't help elect your bottom-ranked canidate.
For example, if a voter's 42nd preference (out of 43) is activated, that means the previous 41 candidates are either elected or excluded, and preferences 42 and 43 are vying for the last vacancy. In that case, preference 42 is the last candidate to be elected, or the runner-up, and preference 43 is the runner-up candidate, or the last to be elected.
-
Legal Beagle: Voting in an STV election, in reply to Rich of Observationz,
Rich, if you were to read the papers I linked to (at the dunedinstadium website) near the top of page 2 of this thread, particularly the first four, I doubt you would come back and say you still don't know how STV works.
For those who don't want to know (or who don't particularly care about) the workings of STV, all they have to do is what I said to Paul a few minutes ago, just rank-order those candidates they are able to – job done.
-
Legal Beagle: Voting in an STV election, in reply to Paul Campbell,
I agree with your summation, Paul, but I can't help reiterating that it was not necessary for people to rank-order all 43 candidates in order to have an effective say in the outcome. Associate-Professor Janine Hayward recommeded people just rank-order the candidates they want to see elected. I go a bit further and say people should rank-order only those candidates they are able to. If your vote runs out and there are still three or four seats (out of 14) to be filled, what does it matter? You've affected the outcome in the way you wanted; others can determine who fills the last vacancies. Everyone happy.
-
Legal Beagle: Voting in an STV election, in reply to izogi,
The sad fact of the matter is, izogi, “media” workers have only ever had a very basic understanding of how STV works. That is very clear from the hopelessly incomplete and inaccurate descriptions of STV we see in our local papers, particularly around local election time.
I’m sure that newspaper editors, in this democracy of ours, adhere to the principle of ‘free speech’, which (simplistically), I take to mean we are free to communicate with each other, even when what we espouse is actually quite wrong. Follow-up letters from readers (often unpublished – I have personal experience of this) will give (hopefully) correct information.
However, I want to take the opportunity given by your response, to actually show readers exactly how later preferences can never harm earlier preferences, using the 2007 Kapiti Coast District (at-large) election, that I referred to upthread.
This was an election to fill five seats. For the purposes of this explanation, the description begins at iteration 14 – Anne Molineux is already elected.
As you read through it, keep in mind that, at the conclusion of each iteration, the current keep values of the elected candidates are re-calculated, downwards, to take account of the increasing number of votes that have dropped out of the election; and that the votes remaining in the election are recounted, using their new (lower) keep values.
Iteration 14 – Quota 2832.64
Total surplus is 279.30
David Scott is elected with 2892.45 votes (surplus 59.81). Norm Pashby has 1650.48 votes and is now excluded (1650.48 plus 279.30 = 1929.78, which is less than the next lowest candidate, Hilary Wooding, on 2058.21).
Mansell 2201.31
Wooding 2058.21The gap between Barry Mansell and Wooding is 143.10 votes. Pashby's 1650 votes and the total surplus of 279.30 votes are now redistributed.
Iteration 15 – Quota 2761.23
Total surplus is 1048.76
Peter Ellis is elected with 3091.05 votes (surplus 329.81)
Mansell 2371.47
Wooding 2273.72Following the distribution of the total surplus of 1048.76 votes, the gap between Mansell and Wooding is now 97.75 votes. This indicates that Pashby's supporters tended to favour Wooding over Mansell.
Iteration 16 – Quota 2709.49
Total surplus is 475.25 votes
Tony Jack is elected with 2757.43 votes (surplus 47.93 votes)
Mansell 2487.54
Wooding 2456.18Following the distribution of the total surplus of 475.25 votes, the gap between Mansell and Wooding is 31.36 votes. Jack's surplus, plus the surplus votes of the other three elected candidates, are now redistributed.
Iteration 17 – Quota 2685.58
Total surplus is 287.43
Mansell 2544.51
Wooding 2539.22Following the distribution of the total surplus of 287.43 votes, the gap between Mansell and Wooding is now only 5.29 votes. It is now clear that the supporters of the four elected candidates are tending towards Wooding, rather than Mansell.
Given that both candidates are still about 140 votes short of the quota, and that there are no more candidates to be excluded from the count (adding new votes to the mix), it can now be said that Mansell is doomed.
Iteration 18 – Quota 2671.27
Total surplus is 170.40
Mansell 2587.49
Wooding 2584.64 (gap 2.85)Iteration 19 – Quota 2662.58
Total surplus is now down to 100.66
Mansell 2612.06
Wooding 2612.43Wooding has now overtaken Mansell. The remaining iterations are performed in order to redistribute the ever-decreasing total surplus, and to put Wooding up over the final quota.
Iteration 20 – Quota 2657.41
Total surplus is 59.23
Mansell 2626.77
Wooding 2628.82 (gap has widened from 0.37 votes to 2.05 votes)Iteration 21 – Quota 2654.36
Total surplus is 34.78
Mansell 2635.42
Wooding 2638.50 (gap 3.08)Iteration 22 – Quota 2652.55
Total surplus is 20.40
Mansell 2640.51
Wooding 2644.19 (gap 3.68)Iteration 23 – Quota 2651.50
Total surplus is 11.95
Mansell 2643.50
Wooding 2647.53 (gap 4.03)Iteration 24 – Quota 2650.88
Total surplus is 7.00
Mansell 2645.25
Wooding 2649.49 (gap 4.24)Wooding has now almost attained the quota, falling 1.39 votes short at this iteration.
Iteration 25 – Quota 2650.51
Total surplus is 4.23
Mansell 2646.28
Wooding 2650.64 (gap 4.36)Wooding has now attained the quota and is elected. Mansell is excluded.
From the above, it can be seen that the later preferences of the voters who supported the already-elected candidates, at each iteration, favoured Wooding over Mansell, and that, therefore, the voters, collectively, preferred Wooding to Mansell, although it was a remarkably close-run thing.
Therefore, it is clear that Mansell was ranked lower than Wooding (or not ranked at all) on the relevant voting documents that played a part in deciding which candidate would fill the fifth and final seat.
-
Legal Beagle: Voting in an STV election, in reply to Sacha,
Wow! Not bad. I managed 13 of 20 candidates for the C&C DHB. I didn't have sufficiently adverse opinions of the remaining seven, to "suppress" a most-disliked candidate.
I suppose you could say the moral of my story, above, was, rank-order as many or as few of the candidates as you see fit. Whatever we all do, it's all good. In all STV elections, there's enough proportionality to satisfy pretty much everyone, particularly for those voting in the DHB elections, and especially for those lucky people who are voting in the Dunedin City and Palmerston North City elections (14, and 15, citywide, respectively).
-
I was casually playing around on the Internet the other day, and stumbled across the Otago University Magazine, Issue 19, February 2008, in which there was an article titled “Voting with your feet”, by Nicola Mutch. In this article, a Dr Chris Rudd (who had claimed to originally be a proponent of the single transferable vote (STV) system for local body elections, but had since changed his mind) was quoted as saying—
“I thought it [STV] would be fairer and more accurate. But, in every single case we’ve studied, the person who won the first-choice race won the election. Ranking your candidates hasn’t changed any results, but it has added a layer of complication to the whole system.”
I have seen several examples of candidates coming from behind to win / take a seat. It happened in Kapiti Coast District (at-large) in 2007 (Wooding overtaking Mansell, which Rudd and his fellow researchers must surely have been aware of); Celia Wade-Brown overtaking Kerry Prendergast in 2010; and Mark Peck overtaking Prendergast’s husband, Rex Nicholls (at the second iteration), to take the third seat in the 3-seat Lambton Ward in 2013, to name three.
In addition, although the researchers wouldn’t have known this in 2008, we must never overlook the impressive performance of Jinty MacTavish in Dunedin’s Central Ward in 2010, who was 11th on the count of first preferences, was elected 4th (out of 10), and who finished with the third-lowest final keep value – a very clear example of the power of second and subsequent preferences.
What people like Rudd would appear not to realise, is that when people vote in STV / PV (single-seat STV) elections, they rank the candidates in their true order of preference, because they know their later preferences cannot harm their earlier preferences. (Readers who have followed this thread will know that this particular property of STV is one that I consider to be extremely important – in fact, vital.)
That being the case, it stands to reason that the leading candidates at the first iteration will almost always be the most-preferred candidates in the election (sometimes with exceptions that become apparent as any particular count unfolds).
Consequently, those who argue that repeated counts and transfers are therefore unnecessary, overlook the fact that the purpose of second and later preferences is to correct the distortions that arise from allowing only first preferences to be cast (particularly in *single-seat* FPP elections), and that, as I say, the system allows voters to list their true preferences without fear of wasting their votes, thereby ensuring the correct candidate, or candidates (in the collective opinion of those voting), is / are chosen.
And, let’s not forget, under PV (in particular), the split vote, whereby, in FPP elections, electoral support for candidates A1 and A2 is divided, allowing candidate B to win (as will happen in Auckland this coming weekend), is completely eliminated.
Interestingly (at least to me), on 15 September [2016], a Kapiti Coast resident, in a letter to the editor of the Dominion Post, argued the opposite of Rudd. The writer warned voters not to rank-order too many candidates, because “If you choose to number your preferences then your one vote can be split, supporting all of your numbered preferences depending on how the vote counting proceeds, including those that you don’t want.” (He said this had happened in Kapiti Coast – 2007?) Putting that the other way around, he was saying that lower rankings could be used to defeat candidates you like! Apart from the fact that he was quite wrong, overlooking STV’s later-no-harm property, he actually didn’t know what he was talking about.
A candidate who comes from behind to win a seat does so because the voters, collectively, preferred that candidate to the one who was pipped at the post. In such a situation, the surplus votes of already-elected candidates, that are transferred to, say, candidate X, to help X overtake and defeat candidate Y, will, in every case, show a lower preference ranking (or no preference ranking at all) for the now-defeated Y, compared to the now-elected X, on the relevant voting documents.
Therefore, contrary to the assertion of the writer referred to, above, Y was not a candidate that X’s supporters liked and inadvertently helped defeat. Under STV / PV, that simply cannot happen.
-
Legal Beagle: Voting in an STV election, in reply to Nicci Coffey,
Nicci—
While your understanding of how surpluses are distributed, and Graeme’s confirmation of it, are correct, your example of 2,000 votes cast and 1,334 being needed to win, could not actually happen. If two candidates are being elected, the initial quota would be 666.6666… (2000 / 3).
To give you a brief, but slightly more detailed explanation of how surpluses are transferred, I’ll use the 2013 Southern Ward election (see the Calculator Commentary and Iteration Report at http://wellington.govt.nz/your-council/elections/past-election-results/results-and-voting-documents-returned/southern-ward , then scroll down to the bottom of page 9, of 12).
There were six candidates vying for two seats, and 7590 votes were cast. That means the initial quota was 2530.000000001 votes (7590 / 3, plus a billionth of a vote). Only two candidates could attain that many votes, there being at most only 2529.999999998 votes left for anyone else to attain.
Paul Eagle received 3626 first-preference votes, giving him a surplus of 1095.999999999 votes, which had to be distributed before the lowest candidate was excluded from the count (there being only 170 votes separating the lowest two candidates).
He kept 0.697738555 of each of his 3626 votes (being the quota of 2530.000000001 / 3626), giving him a new total of 2530.000000430 votes – fractionally above the quota. The remaining 0.302261445 of each of his 3626 votes were distributed to the other candidates in accordance with the voters’ second-preference choices.
At the second iteration, David Lee received 386.290126710 votes (1580.290126710 less his total at the first iteration of 1194 votes). That means he was the second-preference choice of 1278 of the 3626 people who voted for Eagle (1278 × 0.302261445 = 386.290126710).
Ginette McDonald was the second-preference choice of 797 people, Bryan Pepperell of 486 people, Will Moore of 260 people, Brent Pierson of 313 people, and Don McDonald of 75 people. That means 417 people voted only for Eagle (i.e., they gave him their unique first preference), the remaining value of their votes (that Eagle did not need, to stay above the quota) becoming non-transferable – a total of 126.043022565 votes.
At the commencement of the third iteration, the lowest candidate, McDonald, was excluded from the count and his keep value was permanently re-set at 0.0; a new, lower, quota was calculated, to take account of those 126.04… votes that had dropped out of the election; Eagle’s keep value was re-calculated (because the new, lower, quota gave him a new surplus – of about 42 votes, not enough to save McDonald, who was now about 252 votes behind Pierson); and all 7590 votes were recounted, using Eagle’s new keep value. And so on.
While the process is extremely tedious (which is why a computer is needed to produce the result), mathematically, it’s quite straightforward.
-
Hmm, I was completely unaware of this event (and of last year's). No worries. There's no way this kid's paying anything like $25 for a burger. $14 tops, once in a blue moon, is my lot.
-
Legal Beagle: It’s Carter/Docherty Day;…, in reply to Graeme Edgeler,
You put the lowest number next to your *most*-favoured candidate!
Hmm, to me, when rank-ordering candidates (in the Wellington mayoralty contest), 8 is lower than 1, but I take your point.
I assumed it was because they were scared of either: (1) appearing defeatist; or (2) scaring off voters who might vote for them, but are against the person they name.
Yep, them, too; further examples of FPP thinking. There is clearly much more voter-education to be done.
-
welcome this sort of candour from politicians. I’ll be voting for everyone, although in what order, I’m still not sure, but this sort of information from candidates is very helpful in understanding where everyone fits.
I, too, will be rank-ordering all the candidates – and putting the lowest number against my least-favoured candidate – for both the mayor and councillor elections.
Back at your STV posting, you were asked “is it illegal for candidates to send out how-to-vote cards in New Zealand, or is it just not common practice here yet?”
From my reading of the relevant Act and Regulations, this aspect of campaigning is not covered. Therefore, if it is not expressly prohibited, then presumably it can be done.
Given that we have postal voting, candidates could include on (the reverse side of) the “Vote-for-Me!!” leaflets their helpers put in our letter-boxes, their recommendation as to who we should give our second, third (and perhaps, in some cases, depending on circumstances) subsequent preferences, to. As you say (in most cases, I suspect) they know more about the candidates opposing them than us punters out here in Voterland do.
We must, of course, keep in mind that, in New Zealand local elections, 85% (give or take) of candidates are Independents / No party. It’s not like in New South Wales, for example, where most candidates are “grouped” (whether the groups be local or national), and it’s easy to rank the groups / ungrouped candidates on the how-to-vote cards. (I know one candidate who is having a party at his place on election night, 10 September. Condition of entry? You have to have handed out his how-to-vote cards that day.)
Failing that, there is nothing to stop candidates from putting up on their own campaign websites / Facebook pages (which they should all have) their rank-orderings of at least two or three other candidates (in their ward) whom they think they could work with, should they be elected, so as to inform our own rankings.
I have a feeling, however, that the resistance you encountered at that meeting, was at least partly to do with the fact that the candidates instinctively felt that by answering your (great) question, they would be helping their opponent(s), perhaps at their own cost. In other words, they are still in an FPP mindset. They do not yet instinctively know that a later preference can never harm an earlier preference (for themselves).
This, of course, goes back to the discussion at your STV posting, but I think I have made my point.