Monday, 14 September 2009

Their motives should be clear

Labour stand on the edge of a political wilderness. They will certainly lose the next election, that much is clear. They have already been in power for twelve years, and the natural swing of voter tendencies is moving against them.

If Cameron’s government delivers the changes in politics that the public are demanding, if they manage to remain united over Europe, if they remain competent, if Labour divide and squabble over party direction in the wake of a collapsed New Labour project, then they face the prospect of remaining in opposition for anything up to 20 years. Sound unrealistic? It has happened before:

Between 1931 and 1945, Labour were out of power, following the collapse of Ramsey McDonald’s second government and expulsion from the Labour party. They only regained power after a titanic radicalisation of public opinion during the war. During these 13 years out of office, they ran through Arthur Henderson and George Lansbury very quickly as leaders, before settling on Clement Attlee in 1935.

Between 1951 and 1964 Labour were once again in opposition, following Churchill’s re-election. The Conservatives had four Prime Ministers in this period (Churchill, Eden, Macmillan and Douglas-Home), changing leadership in order to remain in office. Labour came into power again with Harold Wilson, who took over from the previous leader, George Brown, after an impressive three weeks and six days at the helm.

Final example, and possibly the most obvious. From 1979 until 1997, Labour remain confined to the opposition benches, despite running through Foot, Kinnock, Smith and Beckett (who lasted a little over two months) as leaders. It was not until Blair came and shook up the roots of the party that they once again became electable.

With this slightly inglorious history nagging in the minds of the current Labour ministers, a drastic option has come to mind – electoral reform.

Now would be the perfect time to change the current system, but it also comes at a period that must warn us against such a change.

On the surface, the season is ripe for reform. With the tempestuous expenses scandal still fresh in memories, and the ever-increasing focus on changing the broken political system, the public would welcome a change in the way that the votes work. With the argument presented in the Independent editorial last Friday, saying that many are fed up with the tedium of meaningless votes cast in safe seats, the case may look convincing.

Be warned.

A new voting system would carry endless problem for governance. The European Elections aided the election of representatives from the BNP in the areas of Yorkshire and the Humber, and the Northwest.

The proposed voting system to be introduced instead, the AV system, would allow voters to list candidates in order of preference. If no candidate wins an overall majority of first-preferences, the second preferences are counted, and so on. This system can very easily lead to people being elected by mistake. As many put a candidate in second place who they do not want to see in power, they are effectively voting for them without realising it, as AV systems always go through to the second round of voting. The result – an unwanted candidate can gain office without genuine support. For Labour, this would mean a substantially higher election rate with many putting them in second place behind either a Conservative or a Lib Dem first place vote, second place votes that would then count as first place votes in the second round of voting.

The government are not as concerned with democracy as they might like to appear. This new measure will give them a far greater chance of regaining office in four years’ time, and subsequently after that.

Do not be deceived – it is very much a political move.


  1. The Alternative Vote system has many problems. It fails many of the basic criteria of voting system: It can inconsistent -- if a constituency after voting was split into two, and both halves would elect X, there's no guarantee that the whole constituency would also elect X. It is not monotone: A candidate x can be harmed by being raised on some ballots without changing the orders of the other candidates. Etc, etc.

    But with all this, it's still so much better than First Past the Post it's not even funny.

    FPtP encourages 'tactical voting' to a huge extent. It forces a mainly two-party system (Duverger's law), and then gives tiny, extremist parties a disproportionate influence (as they can shift the resulting 50-50 balance). It can (and often does) result in a candidate who would lose in a head to head runoff against every other candidate, winning (e.g. population of 34% extremists and 66% liberals, but two liberal candidates split that vote 33% each). And it fails so many of the voting criteria it's a joke.

    Your criticism of AV -- that "As many put a candidate in second place who they do not want to see in power, they are effectively voting for them without realising it" -- makes no sense whatsoever. You rank the candidates in order. So, yes, people who put a candidate in second place would prefer to see the person they put in first place in power, but would rather see the person they put in second place in power than the people they put even lower. That's what ranking means. Also, your assertion that AV would help Labour is dubious. More likely, it'll help the Lib Dems slightly, by helping break the two-party-stranglehold that tactical voting in FPtP has given, and making it easier for a compromise candidate to get elected.

    Yes, AV has big problems. Eliminating people from each round by who comes the lowest in first choice rankings, rather than taking into account the complete rankings, is not ideal. I, and many other people, would prefer a condorcet method, such as Schulze beatpaths; or even just something like range voting to sidestep the whole Arrow's theorem problem. But this is gilding the lily compared to the clusterfsck that is FPtP. From where we are now, AV is a godsend.

  2. I would not say IRV ("Instant Runoff Voting", which you are calling "alternative vote" here) is a godsend compared to plurality voting. We have objective Bayesian regret figures comparing it to plurality voting, e.g. these:

    With lots of sincere voters, IRV is quite a bit better than plurality. But as we've seen with Australia's "How to Vote" cards, the vast majority of people vote strategically, making IRV approach equivalent behavior to plurality (the strategy with IRV is *still* to top-rank your favorite front-runner, regardless of who your sincere favorite is -- a fact which is virtually unknown among IRV proponents).

    Then you have to consider that IRV causes about 7 times as many spoiled ballots.

    And on top of that, it is more prone to fraud, because it eliminates the ability to do precinct-subtotals, thus one central fraud effort is much easier.

    Score voting, and its simpler variant of approval voting, are both far superior (according to objective Bayesian regret calculations, not just according to arbitrarily chosen criteria). Further, they are simpler than IRV or any Condorcet method. And it appears they have about as good or possibly better results than any Condorcet method.

    But in closing, yes you are quite right that FPtP is simply horrendous.

  3. But surely, Peter, AV is the democratic way to go?

    Take any two-way marginal constituency: let's call it Campbellville. Campbellville is a fight between two main parties - say, perhaps, the Lib Dems and the Tories - with either, though not realistically anyone else, having a good chance of winning the seat.

    In Campbellville, any party that isn't Lib Dem or Tory is, for all purposes, irrelevant. A vote for anyone else is a wasted vote, because no one else can win. In reality, the voters of Campbellville have to decide either to go with the Tories or the Liberals, and that's all there is to it.

    The problem with FPTP is the wasted votes. Anyone who puts down anything other than the two leading parties might as well not have voted.

    What AV does is recognise who the contest is really between, and make sure that's the contest voters are getting a say in. In France - or with our Parliament's elections of the Speaker - FPTP is used but with multiple rounds, with the trailing candidates being knocked off the list each time. The French then vote for the candidates left on the list, wittling it down until there's just two left or someone's got a majority (which would make anyone else winning impossible).

    AV is the same system but bunged into a single round: voters put down their preference and their votes are allocated to the highest-ranked candidate still left in the race, giving exactly the same effect as with those multiple rounds. But the key is that voters are making a real choice, and no vote is wasted. Just because the Tories aren't someone's first choice needn't mean they don't prefer them to the Lib Dems (or vice versa), and their voice should be counted.

    p.s. Unfortunately, Simon's wrong about AV not helping Labour. It does so significantly, and to an even greater extent than the Lib Dems. The reason is that Labour and Lib Dem voters are politically closer, and would usually - though even this is more split than you might expect - rank each other above the Tories. The overlap means there's a core vote split between two parties (with half of it essentially wasted), giving the Tories a lead which extends beyond their political support.