Matthew's Journal

Your first point makes the important assumption that, if there is a hung parliament, a voter's preferred coalition would be between their first-preference and second-preference candidates. Why would that be true?

Also: it is legitimate and reasonable for a voter to prefer majority government to coalition. AV doesn't let voters express that, either (though I accept it doesn't militate against majority government in the same way proportional systems do). I'm not sure it's typical (within the UK's political landscape) for coalitions to work as well as the current one.

Isn't it better for [marginalising the BNP] then, if, despite a small increase in 1st preferences, they find everyone else has ranked them last?

I don't think so. AV muddies the waters somewhat: there are a great many factoids one could extract from the typical AV poll, with greater or lesser degrees of statistical legitimacy. For example: most people probably won't rank all the candidates, only their preferred ones, so the BNP wouldn't be ranked last by a large proportion of the electorate. And if people did rank all candidates in order to put the BNP last, the BNP would simply say that a majority of the electorate had listed them as one of their preferred candidates. Or whatever.

Given some of the garbled explanations of AV I've encountered from purported experts in the past few days (even on Radio 4), it's clear that a lot of people — perhaps even most people — don't properly understand AV's mechanism, let alone its practical consequences. It would be pretty easy to pull the wool over people's eyes in making statements about a system they didn't understand.

Worse, it would be pretty easy to convince someone that a result they didn't like had happened for unfair reasons. If the reasoning they used was that AV doesn't satisfy the Condorcet criterion people would have a hard time refuting it.

FPTP is also unfair, but is simple, well understood and well established. AV lacks those countervailing benefits.

And if we're designing systems to keep the BNP out, we might as well have a system of "The BNP automatically get 0 votes".

There are two key pragmatic criteria for an electoral system:

• It seems fair
  
• It gets a good government

Importantly, "representing the wishes of the people" is not a criterion, except to the extent that it achieves those two objectives. The ideal voting system would be acceptable to everyone, yet would guarantee that the BNP couldn't get elected no matter how much support they had.

if widespread tactical voting [under AV] ends up with it going disasterously wrong, that in itself will surely put an end to it in future elections

Not necessarily. The compulsive gambler, when faced with a disastrous loss, does not go "I'm no good at this; I should stop". Rather, they think "I think I see where I went wrong; I can make good my losses next time". Especially unwise gamblers double the stakes each time they lose, fallaciously adopting the Martingale system. I'm not quite sure how similar thinking would play out under an AV electoral system, but the results are possibly best observed from the country next door.

On top of that, it seems to me that the strategy for tactical voting would be very much dependent on a particular constituency, so it wouldn't work for the Daily Mail to simply tell people who to vote for (well, I suppose they could publish a list of every constituency...)

They could recommend a nation-wide order in which to rank the parties? "Confused by AV? Want to support traditional family values, curb immigration and stop those filthy poofters taking over our bed-and-breakfasts? Here's how:"