How complicated is the Alternative Vote?

Thanks! Chris (and anyone) - feel free to use as you please!

Ian - thanks for the comments, though could you explain this part: "Steps 2-5 in your FPTP chart also apply to AV"?

FPTP is also a very poor example of PR!

Ian - the point is that under FPTP you need to guess how everyone else will vote in order to cast your vote in the most effective way. Under AV you do not (although some people might still feel they need to do that, for no good reason).

My presentation assumes that voters will use their mental energy to decide on their order of preference. Under FPTP they don't have much time for that, as they have to spend most of their energy watching opinion polls, staring at bar charts, and reading conflicting claims from the different parties about who "can" or "cannot" win in that constituency.

The referendum isn't about PR.

Iain (spelt correctly now - sorry!),

It would be helpful to spell out what you mean. I grant that tactical voting can under some circumstances be appropriate under AV, but it's much less of an issue than under FPTP, and I'm not convinced the opinion polls would ever be sufficiently reliable to give you the information you need in order to cast an effective tactical vote. But you clearly think differently and it would be helpful to know why.

Also I'd be interested to know who is spinning this as a referendum on PR.

Sam - I would call that expressive voting rather than tactical voting, since the person is trying simply to make a point by voting for a small harmless party first, rather than trying to influence the outcome.

(And in the last General Election in my constituency Greens came first and Lib Dems came fourth!)

Clay - thanks, helpful link. In the "When IRV fails" section, I would have thought M2>m>M1 is more likely than M2>M1>m, assuming a bitter rivalry between the two main candidates. And if that's the case, the argument breaks down. But it would be interesting to see empirical evidence for one or the other.

But I think we can all agree that IRV/AV does a better job than FPTP at encouraging voters to put their genuine favourite candidate first.

"M2=Republican, M1=Democrat, m=Green"

In that case, I can see how M2>M1>m would be realistic, but not m>M2>M1. If it was m>M1>M2, then m would no longer be a spoiler, and the argument no longer works.

And I don't understand the real scenario at the bottom of that page. The highlighted voters did exactly what you suggest: they top-ranked the frontrunner - the one who ended up with the most first-preference votes - which was W! (Or are you saying that people need to guess everyone's second preferences as well as their first preferences in order to determine which two candidates are most likely to win?)

You obviously have much better opinion polls in the US than we do in the UK... Is it really possible in an IRV/AV election to determine which two candidates are most likely to win? Maybe it is in the US, with its essentially two-party political system. But elsewhere?

AV makes it more difficult to determine which two candidates are most likely to win, compared with FPTP, so in that way it gives less incentive to vote tactically and hence more encouragement to put the genuine favourite as first preference. You don't know of any evidence that supports that, but are you aware of any evidence against that?

And you seem to think the aim of voting is to back the winning candidate in the final round. I disagree. If I don't particularly like either M1 or M2 then my strategy is to vote for m, and that is always the best strategy, in order to maximise the chances of m winning. True, it could be that my vote for m leads directly to victory for whichever of M1 and M2 I dislike the most, but I don't really care. My favourite candidate, m, made it to the final round, and that is a huge success for a minority party candidate. Maybe next time he/she will win! Okay, I backed the losing candidate in the final round, but so what?

“AV makes it more difficult to determine which two candidates are most likely to win, compared with FPTP”

Not really.

Well, does it or doesn't it?

If you don’t have a strong indicator from pre-election polls,

which you don't: the Republican candidate in that Burlington election was 253 votes short of victory - was that predicted by the pre-election polls?

you can just go off historical data.

which are unable to take account of the current situation: remember that the Republican candidate was 253 votes short of victory - was that predicted by historical data?

In fact, maybe those 253 voters did vote tactically M>W>K rather than W>M>K: how do you think they feel now? Wouldn't they have been better off voting sincerely for W>M>K rather than taking the misleading advice of some voting guru?

Sorry, but I'm not convinced.

Anyway, do you seriously think that FPTP is better on this account? (Who would have won in Burlington under that system?) If not, then I don't see any point you carrying on. We're deciding between FPTP and AV here, so your favourite score voting method is irrelevant.

The point is that a weak party is statistically more likely to defeat ONE of the major parties than to defeat BOTH of them. This should be obvious.

That's not obvious at all, if by "one" you mean "one and not both". If by "one" you mean "one or both" then that is obvious (but irrelevant).

Now say that swapping your sincere vote for a first-ranking of M1 has:
A 0.0001 chance of changing the winner from M2 to M1.
A 0.000001 chance of causing M1 to win instead of m.

If M1 and M2 are very close in their levels of support, then the probability of m landing precisely in that tiny gap between M1 and M2 is extremely small. In that case, m is much more likely to come first or third than to come second, in which case a supporter of m should put m first on the ballot.

Continued here...

http://www.anthonysmith.me.uk/2011/01/26/no-more-tactical-voting-under-av/

Dean,

Thanks, I'd be interested to see that evidence.

But anyway, I was making a different point. Obviously it is more complicated to write "1 2 3..." than to write "X". But the process of choosing where to place the "X" is much more complicated than the process of choosing where to place the "1 2 3...".

Anthony

Peter - you'll need to explain that a bit more. Could you give a hypothetical example of how to vote tactically under AV? A real example would be even better.

Bill - that's great. But please explain why you say there will "definitely" still be tactical voting under AV.

Tim,

Thanks for taking me up on that. Let's put some names and numbers to flesh that out:

A = (T)ory
B = (L)ib (D)em
C = (L)abour

Tory voters detest Labour
LD voters typically prefer Labour to Tory

40 vote T then LD
40 vote L
15 vote LD then L
5 vote LD then T

In that case, the LDs go out in the penultimate round, and the final result is

55 L
45 T

You are suggesting that the T voters can vote tactically to ensure a LD victory. At least 11 T voters need to do this, as follows:

29 vote T then LD
40 vote L
15 vote LD then L
16 vote LD then T

Then in the penultimate round, T is knocked out and the final result is

60 LD
40 L

In that case, a tactical vote would be successful.

However...

I think (as I have said in a more recent post) that this presupposes a totally unrealistic level of knowledge.

Suppose, first, that the opinion polls have it wrong, and that the true (non-tactical) votes would have been:

46 vote T then LD
32 vote L (then LD)
17 vote LD then L
5 vote LD then T

In that case T would win the seat. But if your 11 voters switched to LD first preference, then that would push L out in the penultimate round and cause LD to win against T in the final round!

So, in order for a tactical vote to be successful, you have to persuade 25% of the T voters that T cannot possibly get 46% of the first preferences (given that the opinion polls say 40%) and that they should therefore vote for LD as their first preference instead. So it all depends on how accurate the opinion polls are (not very).

I don't doubt that people will still attempt to vote tactically under AV. But I just don't think it makes any sense to do so, given the information available to us. In contrast, it very often makes perfect sense under FPTP.

Tim - in that case, with L on 45%, and T and LD on around 27.5%, I would say it's definitely worth T pushing for a victory. To say "T can never win" is simply not true. It wouldn't take much to change it to 40% L, 35% T and 25% LD, then if the LD candidate is politically quite similar to the T candidate (which you seem to be assuming), then T would probably win on LD's second preferences.

So I'm still not convinced that tactical voting would ever be sensible under AV. Certainly it would be much less prevalent than under FPTP.

Go back home??!

Please do elaborate, and explain why these "minority votes" should be worth less than other people's votes.

Andrew - please explain how the diagram is misleading.

Dave - under AV it's fine to put down only one preference, so changing to AV will make no difference to you. But some people do vote tactically, and a change to AV will make it simpler for them. And just because you don't want to express your opinions in any more detail than your first preference, that's no reason to prevent other people from doing so. It's called democracy, I think

And if there were more than one excellent candidate to choose from, I don't understand why you wouldn't want to be able to vote for both of them, one as first and one as second preference.

JR - I don't understand any of that.

Andrew - under AV the least favourite candidate (i.e., the one who the majority thinks is worse than every other candidate) cannot win. It's impossible. AV guarantees that 100%. (In fact, nor can the one who gains the fewest first-preference votes. That person is the first to be eliminated.) Under FPTP, the least favourite most certainly can win.

You'll have to elaborate on the thought processes between the lines for AV.

Thanks for dropping by, Mr/Ms What-a.

No, that's not all I have to offer.

The counting is quite simple too. There's a flowchart for the counting here, for example:

http://ulstersdoomed.blogspot.com/2009/06/effect-of-electoral-reform.html

But the majority of people will be voting rather than counting the votes, so I think the voting process is more relevant to most people than the counting process.

(An example of why I focus on the voting rather than the counting would be the D'Hondt method used for our MEP elections. I haven't the foggiest idea how it works, but I don't care, because it's obvious that I should just cast my vote for the party I want to win.)

Thanks again for your comment. I genuinely enjoyed reading it.