AV - how to vote?

[mattleonard]

One of the big pro-AV arguments which seems to be coming up is that under FPTP it is likely that more people will not want the winner as their MP as do want them. Another problem with this argument seems to me to be that AV doesn't actually provide any guarantee that this will be the case. It is quite plausible (although it will probably become less common) that at the end of every round there is still no winner with a majority.

 

What seems an even bigger problem is that if the current problem is a majority of people who don't want a particular candidate or candidates to be returned then AV doesn't actually take this problem seriously enough. A voter can express dissatisfaction with all candidates by not placing a vote - so should this dissatisfaction not be taken into account? Surely the logical conclusion to these pro-AV arguments ends up being that a constituency which cannot provide a majority of all eligible voters for a candidate shouldn't be able to return an MP.

[HeadingNorth]

One of the big pro-AV arguments which seems to be coming up is that under FPTP it is likely that more people will not want the winner as their MP as do want them. Another problem with this argument seems to me to be that AV doesn't actually provide any guarantee that this will be the case. It is quite plausible (although it will probably become less common) that at the end of every round there is still no winner with a majority.

 

Eventually you're reduced to only two remaining candidates, all others having been eliminated. It is then a certainty that someone will secure more than 50% of the preferences.

 

It's possible, if people don't list large numbers of preferences, that "more than 50% of preferences" still means that less than half of the voters actually marked your name at all. I don't know how often that will happen. It cannot happen more often than it does already under FPTP.

[truman]

Consider the following selection of votes, between the Brown, Blue, and Pink parties:

 

6,000 people vote "Brown 1, Blue 2"

5,000 people vote "Pink 1, Brown 2"

4,000 people vote "Blue 1, Pink 2"

2,000 people vote "Blue 1, Brown 2"

 

The Pink party, with only 5,000 votes, is eliminated, and their second choice votes mean that the Brown party takes the seat.

 

Now consider if the 2,000 people who voted "Blue 1, Brown 2" reverse their order of preference, so you have:

 

6,000 people vote "Brown 1, Blue 2"

5,000 people vote "Pink 1, Brown 2"

4,000 people vote "Blue 1, Pink 2"

2,000 people vote "Brown 1, Blue 2"

 

It's now the Blue party that is eliminated, with only 4,000 votes. Their second-choice preference are for the Pink party, which takes the seat. The Brown party has lost because they secured 2,000 extra votes. Worse still, if the votes are analysed, it turns out that two thousand people who voted for everyone except the Pink party, are responsible for them winning the seat!

 

 

 

If AV should be adopted, how long do you think it would last beyond a headline saying "Labour lose Sheffield Heeley due to too many people supporting them?"

 

OK I understand what you're saying but after blue are eliminated in the second instance are their voters' 2nd choice, ie pink, still transferred even though at this stage (before the transfer) brown has > 50% ?If you see what I mean..?

[mattleonard]

Eventually you're reduced to only two remaining candidates, all others having been eliminated. It is then a certainty that someone will secure more than 50% of the preferences.

 

It's possible, if people don't list large numbers of preferences, that "more than 50% of preferences" still means that less than half of the voters actually marked your name at all. I don't know how often that will happen. It cannot happen more often than it does already under FPTP.

 

I agree, but in FPTP it is implicit that such a scenario is OK. In AV it is implicit that such a scenario is not OK. So surely that raises the stakes if such a result occurs.

 

I take your point that there will always be a majority of any remaining votes in an AV result - but by this point surely it is quite likely that this "majority" is exactly equivalent to the scenario of simply having more votes than everybody else which you would get in FPTP?

[HeadingNorth]

OK I understand what you're saying but after blue are eliminated in the second instance are their voters' 2nd choice, ie pink, still transferred even though at this stage (before the transfer) brown has > 50% ?If you see what I mean..?

 

I think what you mean is that if you eliminated the 4,000 people who voted Blue, then Brown has 8,000 out of the 13,000 other votes. But that's not how AV works; in fact, that is essentially just a first-past-the-post result. Brown has 8,000 votes, which is more than either Blue or Pink, therefore Brown wins.

 

Under AV, the votes of the person who gets eliminated are redistributed, before you examine the result again.

[HeadingNorth]

I agree, but in FPTP it is implicit that such a scenario is OK. In AV it is implicit that such a scenario is not OK. So surely that raises the stakes if such a result occurs.

 

I take your point that there will always be a majority of any remaining votes in an AV result - but by this point surely it is quite likely that this "majority" is exactly equivalent to the scenario of simply having more votes than everybody else which you would get in FPTP?

 

I don't know how likely it is. I accept your point; I don't know if the advocates of AV will do so. I'm trying to explain the pros and cons of both systems without telling people which one I think is best.

 

FPTP essentially only has one flaw; which is this one. You can be elected despite a majority of voters not wanting you; indeed, a majority could want anyone at all except you, but if their votes are split you can still win.

 

The argument boils down to whether that one flaw is a big enough flaw to warrant throwing it out and replacing it with a system which has a whole boatload of flaws, but all of them much smaller ones.

[mattleonard]

The argument boils down to whether that one flaw is a big enough flaw to warrant throwing it out and replacing it with a system which has a whole boatload of flaws, but all of them much smaller ones.

 

This seems like a very fair and accurate representation. In my case, as I've thought out loud on this thread the conclusion I've come to has been that I don't think the big flaw is as big as the cumulative flaws of AV. I suspect that I'm probably in the minority though.

[ArchimedesUK]

Consider the following selection of votes, between the Brown, Blue, and Pink parties:

 

6,000 people vote "Brown 1, Blue 2"

5,000 people vote "Pink 1, Brown 2"

4,000 people vote "Blue 1, Pink 2"

2,000 people vote "Blue 1, Brown 2"

 

The Pink party, with only 5,000 votes, is eliminated, and their second choice votes mean that the Brown party takes the seat.

 

Now consider if the 2,000 people who voted "Blue 1, Brown 2" reverse their order of preference, so you have:

 

6,000 people vote "Brown 1, Blue 2"

5,000 people vote "Pink 1, Brown 2"

4,000 people vote "Blue 1, Pink 2"

2,000 people vote "Brown 1, Blue 2"

 

It's now the Blue party that is eliminated, with only 4,000 votes. Their second-choice preference are for the Pink party, which takes the seat. The Brown party has lost because they secured 2,000 extra votes. Worse still, if the votes are analysed, it turns out that two thousand people who voted for everyone except the Pink party, are responsible for them winning the seat!

 

 

 

If AV should be adopted, how long do you think it would last beyond a headline saying "Labour lose Sheffield Heeley due to too many people supporting them?"

 

I don't see a problem - in both of those scenarios the candidate with >50% won. The thing is that in reality we aren't dealing with a scenario with such few candidates and predictable blocks of preferences. As well as people switching from brown to blue you could have people going the other way to counteract it, those who don't number every candidate, those who do and so on!

 

The complexities of optional preferences makes it virtually impossible to successfully engineer tactical voting to your advantage so everyone might as well vote honestly - which is why AV is better than FPTP, under which tactical voting is very easy to engineer.

[mattleonard]

The complexities of optional preferences makes it virtually impossible to successfully engineer tactical voting to your advantage so everyone might as well vote honestly - which is why AV is better than FPTP, under which tactical voting is very easy to engineer.

 

If you only see tactical voting as the problem, yes. But some might see a scenario where a candidate wins because people don't vote for them and another loses because people did vote for them favourably as a bit twisted.

 

If tactical voting is the problem and it doesn't matter too much if there are occasional anomalous results then a fairer system would be to give everyone one vote which is put into a hat and the winner is picked out.

 

I don't really see how a voting system can really be sold on the basis of its complexity.

[ArchimedesUK]

Regarding the 50% of votes threshold. It's 50% of votes in the final round and 50% of votes under AV is the same as 50% of votes under FPTP. In both cases you don't count those members of the electorate who didn't fill in a ballot or voted for a candidate who wasn't running, etc.

 

When you number your preferences under AV or any preferential system you are essentially saying that "Of the remaining candidates, I would vote for this person if my previous choice wasn't running or was eliminated". You keep numbering until you have no opinion on the remaining candidates. This is the same as stating that in a FPTP election if none of your numbered candidates were running then you wouldn't have voted as you have no opinion between the candidates on offer.