Wednesday 16 February 2011

Candidate lists

At the last election there was a constituency (hypothetical) that had the following candidates running, with the following number of votes...

Jeff - 18,000 votes
Billy - 17,000 votes
Ashley -4,000 votes
Jane - 800 votes
Kerry - 200 votes
Tom - 100 votes

Under FPTP Jeff wins. Fair? It must be, if you support FPTP.

Next election the same candidates run again. Every voter votes the same way...except Ashley has pulled out. Instead Ashley's voters decide how they'll vote in Ashley's absence. 2000 decide they won't bother voting at all, 1,600 decide they'll vote for Billy, and 400 decide they'll vote for Jeff.

New vote shares for the next election are...

Jeff - 18,400 votes
Billy - 18,600 votes
Jane - 800 votes
Kerry - 200 votes
Tom - 100 votes

Another FPTP election, but this time Billy wins. Fair? It must be if you support FPTP.

Yet this situation is exactly what AV deals with, except it does it within the same election. Before claiming that it's "unfair" that someone's 2nd or further preference counts as much as your 1st, think about the above scenario. Would you complain about your election being won by someone because another candidate wasn't present? Would you think that Billy's votes aren't as worthy as Jeffs?

Of course not, you'd not bat an eyelid at the situation, and you'd congratulate the winner. Yet this is what AV does to ensure that the most popular of the most popular win each seat.

Edit: Someone on twitter tried to claim that the above results are like "earning £10" whilst the AV method is like stealing £10. It should be clear to those that don't need to abandon logic to parrot a party line...what I've demonstrated above is the exact process AV takes, though over a period of years not hours. If you believe the above FPTP result is fair you lose any right to claim that AV is unfair. It is, possibly unfortunately for you, unavoidable. :) /Edit

For more on why AV is a fairer choice than FPTP, read on.

10 comments:

  1. Of course it is fair that the person with the most votes wins in both cases, that's how most elections work. And only a moron would expect the same result in an election if there were different candidates standing.

    ReplyDelete
  2. I think you've missed the illustrative point, much as you find it so hard to apply basic mathematics to work out why there's no such thing as people getting more votes than someone else.

    ReplyDelete
  3. The initial thinking behind it is.. this also happens under AV too, if billy pulls out ashley could win.

    I think the true problem no2av have with it is that there IS a candidate running in the election, they just don't win.

    If under Fptp someone doesn't run then their normal voters would vote for someone else...once.

    The simple fact is that yes, when someone pulls out of a 5 person election, there will be people voting for their 2nd choice, but most their first. In the final round, it would be the equivalent of 3 parties pulling out...leading to up to 4th preferences being mixed in to the result, something that would never happen under Fptp.

    The reason why this wouldn't happen under.Fptp is that there are more than two view points and someone would fill in if there was a big enough base of support for that political leaning,, expecting people that have no interest or support to be the king maker between two candidates he doesn't really give a crap about, is not going to give u an informed or fair winner

    ReplyDelete
  4. I'm always amused by how you simplify things. Instead of jumping straight to "3 people pulling out", why not follow the process as it happens and pull one person out at a time. 3 more elections, same situation of zero change between anything but a single candidate pulling out...every time you would state that FPTP election is fair. It's why you cannot claim with any authenticity that AV is unfair at the same time.

    ReplyDelete
  5. As I understand the system (from rithompson's article on the subject) it's the candidate with the least votes in each round who gets eliminated. In your example, this would mean Tom would be eliminated, not Ashley, so it's not really accurate to say that 'this situation is exactly what AV deals with', because the AV system would have no influence over whether Ashley is eliminated or pulls out of his own accord.

    I'm afraid I disagree with your premise that '[p]eople are voting in each round for their favourite'. Since the thing could be won after one round, we must (and might as well!) give our first preference vote to our favourite candidate. I may, for example, agree with every policy of the Green Party, the vast majority of Labour's policies, and most policies of the Lib Dems, and would thus order my preferences accordingly. If the Greens were eliminated after the first round of the count, I think it would be fair if the status of my second preference vote for Labour were recognised as such.

    I also think your (related) rhetorical questions – 'Would you complain about your election being won by someone because another candidate wasn't present?' and 'Would you think that Billy's votes aren't as worthy as Jeffs?' – are red herrings. The fact of a candidate's standing or not is a significant factor in the voting logic.

    ReplyDelete
  6. "it's not really accurate to say that 'this situation is exactly what AV deals with', because the AV system would have no influence over whether Ashley is eliminated or pulls out of his own accord."

    A fairly pedantic point, imagine instead that there had been three more elections in between where Jane, Kerry and Tom all pulled out each year and their votes didn't get redistributed, resulting in a two person election the final time with the top two candidates on the same vote numbers.

    I shortened the sequence, but that doesn't change the point that AV's process is simply removing candidates from the ballot and making sure votes are only cast against candidates still standing...just like FPTP.


    "Since the thing could be won after one round, we must (and might as well!) give our first preference vote to our favourite candidate. "

    Right, so people are as I said voting for their favourite...

    " If the Greens were eliminated after the first round of the count, I think it would be fair if the status of my second preference vote for Labour were recognised as such."

    Right..again, you've ordered your preference so that in each round you are putting your mark against your favourite still in the election... so far you're not really disproving my point about people voting for their favourite in each round.

    "The fact of a candidate's standing or not is a significant factor in the voting logic."

    It is entirely a significant factor. In AV a candidate being removed from the ballot because they don't have enough preferences is the same in every practical way as running the election again with that candidate not standing under FPTP.

    ReplyDelete
  7. Lee, as far as I understand your first point, I think we're in agreement. My point to which this one refers was really on a matter of accuracy: had you compared AV's treatment of Tom and his potential absence, as opposed to that of Ashley and his, then any perceived inaccuracy of the comparison would have been more trivial.

    On the second (regarding weighting), our disagreement is with who we consider to be our favourite. For me, my favourite is the person I want to win the election. In your argument, the favourite is the person you want to win the round. My second preference, had my first been eliminated in the previous round, may well then be promoted to being my favourite for that round, but they nevertheless remain my second preference for the election. You could say that if they went on to win the election with, say, 40% 1stPref and 10% 2ndPref, their mandate to govern would be less than that for my first preference (had the latter won with 50% first preferences) because within those 10% 2ndPref votes, not all of their policies (or at least not as many as those of the 1stPrefs) are sanctioned.

    I do concede, though, that it would be far from easy to weight the preferences fairly, and it would perhaps be at 'disproportionate cost', as they say, to do so. However, weighting the preferences would, in my view, provide the ultimate in fair results within the AV framework.

    On the third, I'm afraid I've got a bit tangled up, and am not entirely sure whether we agree or disagree. Should we just do the former to do the latter?

    ReplyDelete
  8. Miblo, dont mention weighting for god sake, he goes mental

    ReplyDelete
  9. "For me, my favourite is the person I want to win the election. In your argument, the favourite is the person you want to win the round."

    I don't see these as different in that if your favourite is eliminated, they can no longer be your favourite for the purposes of the ongoing election. I think perhaps it's more accurate for me to say "Favourite out of the remaining candidates", I don't think we're really disagreeing here :)

    "I do concede, though, that it would be far from easy to weight the preferences fairly"

    The main problem with weighting preferences fairly is that weighting only relates to your individual view. My example above is how I'm attempting to illustrate this. Under FPTP we wouldn't discuss weighting, we would just accept that Jeff's policies were preferred more than any other candidate standing, and in the second example that Billy's were preferred.

    The essence of this problem is that Nothing has changed aside from a candidate leaving, so it highlights why AV is so important...that in reality in a head to head situation Jeff may not be more popular with the electorate than someone else he has beaten under FPTP.

    This is because in each FPTP election (and thus, consequently, every round of AV) we don't try and second guess how much people care about a candidate compared to another person, just whether they care enough *more* about one candidate than another to put your preference against them.

    A quick mathematical representation of how FPTP could, theoretically, land a result in exactly the same manner of problem you describe...

    50 voters vote for the Labour candidate, with an average "approval" for policy, etc. of 35%.
    30 voters vote for Tories with an average "approval" of 59%.
    20 voters vote for Lib Dems with an average "approval" of 90%.

    If you then try to use these people's approval, the result is turned on it's head.

    Lib Dems get 18 full votes, Tories 17.7 votes, and Labour 17.5.

    For me this is why weighting isn't a relevant argument in the case of FPTP vs AV. we don't weight FPTP results, FPTP is a "who has the most plurality out of these candidates" situation and every vote can only be counted as a "1". Equally each round in AV is a "who has the most plurality out of these candidates" and equally each vote can only be counted as a "1", with each candidate only being measured against their remaining competitors...past opinion is no longer relevant.

    ReplyDelete
  10. As for disproportionate cost of trying to deal with the weightings...I doubt it's the case, not that I care for arguments that money should come in the way of representative democracy.

    You could do something like ranking all your candidates, and applying a different "score" to each level at the count stage, though this doesn't get accurate information on how much you "approve" of each candidate compared to the others, so is perhaps redundant..it'd also do worse than AV in that it would actually give voters that preference more power than those who don't regardless of the level of support that those single preference'rs give...not an equal or fair system.

    Perhaps you could ask people to rank each candidate out of 10 instead, where 0 is definitely don't approve, and 10 is totally approve. An average score could be composited then, however the disadvantage would be that those who are very middle of the road would definitely have a strong advantage going through, while divisive candidates might lose out.

    You could also try to take in to account everyone's preferences in a condorcet method, but this actually potentially leads to a logical problem whereby you enter an infinite loop of "I don't like this winner, I would have voted these instead, and as such changed the winner"!

    The problem is without proportionality there is always a problem with a voting system, by not doing the most simple task of just representing people related to the number of 1st preferences they have in a multi-member format, you introduce elements that mean an imperfect result can occur.

    However systems like AV, and perhaps some other variations of preferential voting, reduce the number of imperfect results that can occur.

    ReplyDelete

Got something to say about my post? I'd love to hear it!

Try to keep it civil, I don't delete comments unless obliged to or feel the thread is getting too out of hand, so don't make me do it.