Legal Beagle by Graeme Edgeler

Read Post

Legal Beagle: STV Q&A

37 Responses

First ←Older Page 1 2 Newer→ Last

  • Graeme Edgeler,

    So if STV were to be implement in New Zealand at government level would it be likely to be single member electorates?

    The upcoming referendum on the voting system gives both preferential voting (single member electorates) and STV (multi-member electorates) as options.

    If we opted for STV, it would have multi-member electorates. The 1986 Royal Commission, which recommended MMP, looked into other systems for New Zealand. It said that is New Zealand was to introduce STV, that most electorates should have 5 MPs, but that up to 20% of electorates could differ from this, having between 3 and 7 MPs.

    Wellington, New Zealand • Since Nov 2006 • 3215 posts Report Reply

  • Paul Campbell,

    (note: I think that applying this condition is probably only fair if you take one more step and redistribute the votes of the 2nd last candidate as well)

    Dunedin • Since Nov 2006 • 2623 posts Report Reply

  • Graeme Edgeler,

    Paul - my referring to the Wellington City elections in 2007 was for exactly that purpose. Kerry Prendergast was declared elected without having an absolute majority of votes cast. Whether she'd still have fallen short of an absolute majority on the final iteration is an open question. Maybe. Maybe not.

    If anyone was gonna do it, they'd have done it then, although I suspect the result may happen again this year. Feel free to challenge the result if it happens this time =)

    Wellington, New Zealand • Since Nov 2006 • 3215 posts Report Reply

  • bmk,

    If we opted for STV, it would have multi-member electorates.

    In that case I wouldn't be too opposed to STV (would still prefer MMP) and it would definitely be preferable to preferential voting.

    I now feel comfortable making STV my candidate for the second question in the referendum - if it were only the preferential version then I would choose SMP.

    Though that now leads me to the can of worms in that answering the second question of whether to choose my most preferred option or the one I think most likely to lose in a run-off against MMP:)

    Since Jun 2010 • 327 posts Report Reply

  • Graeme Edgeler,

    that now leads me to the can of worms in that answering the second question of whether to choose my most preferred option or the one I think most likely to lose in a run-off against MMP:)

    Choose your most preferred option. It's not worth the risk.

    Plus, STV is probably the one most likely to lose in a run-off with MMP. Most people opposed to MMP will not want STV, they want the good ol' days of first-past-the-post, or it's near equivalent supplementary member. The option of replacing MMP with something even more complicated will not be particularly popular with such people.

    Wellington, New Zealand • Since Nov 2006 • 3215 posts Report Reply

  • Tim McKenzie,

    By ranking a candidate lowly, you're not helping them beat people you like more than them

    In the context of questions about the (non-)danger of filling in all your rankings, this is a reasonable thing to say; giving someone a low ranking (as opposed to no ranking) never helps them beat people with higher rankings. But giving someone a low ranking (as opposed to a high ranking) can, perversely, help them beat people with higher rankings.

    Consider three candidates: A, B, and C, in a mayoral election.
    39 people vote A first, B second, and C third;
    35 people vote B first, C second, and A third;
    26 people vote C first, A second, and B third.
    C has the least votes, and all their second preferences say A, so A wins with 65 votes to B's 35.

    But now suppose 10 of B's supporters had sufficiently accurate polls to know that this was going to happen. They then change their votes to put A at the top, and the votes look like this:
    49 people vote A first, B second, and C third;
    25 people vote B first, C second, and A third;
    26 people vote C first, A second, and B third.
    Now B has the least votes, and all their second preferences go to C, so C wins with 51 votes to A's 49.

    The 10 highly-informed B-supporters got their second preference by raising their third preference to the top of their vote. They didn't get a better outcome for themselves by not ranking their least-preferred candidate (which is the alternative Graeme was contrasting with a fully specified vote); they got a better outcome for themselves by ranking their least-preferred candidate at the top of their ballot.

    The above example was adapted from a Wikipedia article, which gives a more plausible manifestation of this perversity of STV (single-winner STV, really, which is also known as PV, and which the Wikipedia article calls "instant runoff", which is sometimes abbreviated to IRV). Specifically, the Wikipedia example doesn't require strategic voters informed by highly accurate polls; instead, the supporters of B genuinely change their minds and support A instead, causing A to lose the election they would have won without that support.

    Every electoral system allows some perversities to happen (there's a proven mathematical theorem that says so). But this particular perversity of STV(/PV/IRV) is avoidable (by, for example, a Condorcet method). Essentially, the problem here is that STV can be very sensitive to the order in which candidates are dropped (edit: and it ignores preferences among lower-ranked candidates until higher-ranked ones are elected or eliminated).

    Lower Hutt • Since Apr 2007 • 126 posts Report Reply

  • Graeme Edgeler,

    Tim - yes. Tactical voting is possible under all democratic voting systems. Situations like this are unlikely in real life. And there's not a lot you can do about them.

    Wellington, New Zealand • Since Nov 2006 • 3215 posts Report Reply

  • Tim McKenzie,

    For things like the DHB where you are voting for several people all at once I'd really like some way to vote in tiers so I could nominate a bunch of people I actively wanted, followed by a bunch of people I didn't hate and then a group who should never even get near public office.

    This sort of thing would require changes to the way STV works. A Condorcet method, on the other hand, could (in principle) cope (in its natural state) with any consistent set of preferences among the candidates, so you could say "all these people are better than all those people, but I have no preferences within either group". In practice, I suspect any implementation of a Condorcet method would still require voters to give a ranking, leaving us with the problem:

    Trying to figure out if candidate A is more or less sort-of-ok than candidate B does my head in.

    which I had last night. My DHB has 15 candidates this year, meaning there are about 1.3 billion possible ways in which I could rank them. Choosing to rank them all by comparing pairs of them (and drawing inferences from previous comparisons), there's no strategy that could guarantee sorting them all in less than 41 pairwise comparisons.

    Lower Hutt • Since Apr 2007 • 126 posts Report Reply

  • Tim McKenzie,

    If a Nationwide election were conducted using STV, and everyone who supported National, voted 1 for John Key, and 2 for Bill English and 3 for Gerry Brownlee, all the way down through the National Party List, and Everyone who supported Labour had voted 1 for Helen Clark, and 2 for Michael Cullen, etc. the results would be the same as a list-based Proportional Representation election

    And voters could achieve the same proportions without even coordinating to agree on the order of the candidates within each party.

    If just over a fifth of all voters rank all the left-handed candidates (in any order) above all the right-handed, ambidextrous, and armless candidates, then I think STV guarantees that roughly a fifth of the winners will be left-handed (assuming there are enough left-handed candidates and at least four seats to be filled in the election). So, to answer the original question,

    And I don't see how it can fairly be called 'proportional'- wouldn't that depend on, y'know, there being political parties involved?

    you can think of STV as a way of letting voters roll their own parties, if they don't like any of the party groupings on offer.

    Lower Hutt • Since Apr 2007 • 126 posts Report Reply

  • Tim McKenzie,

    Tim - yes. Tactical voting is possible under all democratic voting systems.

    Yes, but tactical voting wasn't my only point (although I could have done more to make that clear). I was also trying to point out that STV and PV fail the monotonicity criterion: there are situations where you can improve your favourite candidate's chances by giving them a lower ranking. And this particular flaw is avoidable: a Condorcet method that lacks this flaw could be used.

    (Condorcet methods don't automatically achieve proportionality, but this may be a good thing or a bad thing, depending on your point of view. Also, a Condorcet method could perhaps be adapted to be proportional; I don't know. In any case, I don't see any reason to prefer PV to a Condorcet method; Condorcet elections are also much simpler to count than PV.)

    Lower Hutt • Since Apr 2007 • 126 posts Report Reply

  • Christopher Wingate, in reply to Graeme Edgeler,

    The problem with democracy is crown immunity. People get appointed to governing power are given immunity (permission) that failure is ok.

    The immunity is allowing cover-ups, negligence and hidden agenda. And when people complain to the fiduciary in power he gets ignored.

    Perth • Since Mar 2011 • 10 posts Report Reply

  • Steve Barnes, in reply to Graeme Edgeler,

    Tactical voting is possible under all democratic voting systems. Situations like this are unlikely in real life.

    But isn't this exactly what happened in Epsom?
    The people of Epsom tend to be rather conservative and, to me anyway, would not have voted, en mass, for Rodney Hide. But they are smart enough though to understand that, if Rodney won, National would have more support in the house than if they had voted National in the first instance.
    Yes?

    Peria • Since Dec 2006 • 5521 posts Report Reply

First ←Older Page 1 2 Newer→ Last

Post your response…

Please sign in using your Public Address credentials…

Login

You may also create an account or retrieve your password.