Vote transfers and late counts

takun

I broadly understand how the system of transfers works.

In the case of an eliminated candidate it is simple, the entire pile of voting papers is taken and redistributed.

I am less clear about the mechanics of redistributing excess votes over the quota when a candidate is elected, but my understanding is that a bundle of voting papers equal to the number of excess votes is chosen at random for redistribution, and these papers are the ones that go on to the next and subsequent counts.

I can see the logistical difficulty of doing it any other way, but it means that the vast majority of individual papers don't in fact move on to the later counts. It could be as few as a couple of dozen papers that actually move on - certainly too few to be statistically representative.

So, to make your vote hang around as long as possible, is there a logic in voting first for candidates you think will be eliminated early (thus ensuring your actual vote hangs around) then for someone you'd like to see elected but suspect will be in a battle for the last seat.

Maybe my thinking on this is screwy, but given that the real battles this time will be for last seats, where lower preferences go is of huge importance. Of course it involves being right about who gets eliminated, and if everyone did it it would completely eat itself - but is it a reasonable method of voting?

An additional question - if I am right about how excess votes are redistributed, is the same randomly chosen bundle of papers used in a recount, or is a new one chosen?

Nody

Why not vote directly for the lower seats instead of trying to get a handed over vote? That way you can ensure it gets included and has a bigger impact (as there are over all fewer votes in the batch).

jimmycrackcorm

We should introduce electronic voting so that every preference counts instead of something random. Wait.....

making waves

jimmycrackcorm wrote: »

We should introduce electronic voting so that every preference counts instead of something random. Wait.....

I posted this on another forum - but it addresses you point -

Finally on the issues of electronic voting that was used in 2002 - I had some limited involvement in the investigation by party representatives at the time. There were four major issues -
1. The system was being run by a private company who refused to release the source code for independent analysis to ensure it operated as was intended. The government backed the right of a private company to maintain a trade secret over the democratic right of the people to ensure public accountability of the democratic process.
2. There was one memory bank in each electronic booth - no back-up system if the thing crashed and no receipt for the voter to indicate that their vote had been recorded properly.
3. The hardware being used in the electronic machine was ancient - and I mean something equivalent to the old Commodore 64 (slight exaggeration to emphasise the point) - with a very basic operating system that was subject to crashing.
4. The machines were capable of being hacked into - they had no security system in place and any competent hacker with the right equipment could have manipulated the vote.

This is what the government spent nearly €70 million on - and along with that they were second hand machines.

Electronic voting is an excellent idea. It would make the system more proportional in terms of surplus distribution and in terms of speed and I would be fully supportive of it if it was done on the following basis -
1. The machines used open source and verifiable code for developing the software (it's not a complex programme).
2. Each machine has an independent back-up system in case the machine crashes.
3. A paper trail is available in the form of (a) a receipt to the voter and (b) a paper record in the event of a dispute.
4. The machines use up to date hardware that are properly tested to minimise the potential for hacking - with the above paper trail as back-up.

Respublica

my understanding is that a bundle of voting papers equal to the number of excess votes is chosen at random for redistribution

Yes, this is a flaw in our system but it would be quite easy to improve. Instead of chosing ballot papers at random you can transfer all of the ballots, but as only fraction of a vote. This is known as the "inclusive Gregory method".

I gather that a version of this is used in Northern Ireland with very little difficulty. All the returning officer needs is the aid of a calculator, so there's no reason it couldn't be done here.

(An example: a candidate receives 60 votes, making a surplus of ten. The candidate is declared elected and then all of the votes are transferred but each is only counted as a sixth of a vote.)

So, to make your vote hang around as long as possible, is there a logic in voting first for candidates you think will be eliminated early

Yeah, that's correct. It's often best to give your first preference to a no-hoper candidate. This is a tactic called "raising a turkey".

The fact that this tactic is possible shows up another flaw in our version of STV: it doesn't treat all votes equally, so supporters of weak candidates may have somewhat more influence over the outcome of an election.

However there's another improvement to STV that eliminates this issue. It's called the "Meek method" and is used for STV local elections in New Zealand. It's quite complicated but again, I think if the Kiwis' can manage it then so can we.

Electronic voting is an excellent idea. It would make the system more proportional in terms of surplus distribution and in terms of speed

I think it's important to distinguish between electronic voting and electronic counting. There's no reason why we can't use paper ballots but then work out the result using computers. This would avoid the most controversial aspects, such as lack of a paper trail.

As noted above the experts have put a lot of work into finding ways to improve STV. But the very best methods, like "CPO-STV", can't realistically be implemented using a hand-count.

takun

Thank you Respublica, that's a very clear answer.

I like the "inclusive Gregory method" - it seems much fairer. Given that the excess over quota can be very small, especially in the later counts, it seems too likely with the current random method that redistributed votes may not be a true representation of the total.

I am assuming, now that I think about the logistics of this, that on a recount a new random bundle is chosen, right? Makes recounts more interesting.

For what it's worth, it's made me think this is probably how I will vote.

1. 'Turkey' 2. My favourite 3. My 2nd favourite 4. The last seat battler END.

I will 'raise a turkey' this time, partly to make my vote last but also because I think this particular turkey, though not likely to get to the final mixup, is fully deserving of support. Then move on to my other top candidates - who will I believe already have been elected (but just in case). If they have been elected my preferences for them will be ignored and the vote end up with the person I believe will in a battle for the last seat.

But not decided on anything yet.

Respublica

I will 'raise a turkey' this time, partly to make my vote last but also because I think this particular turkey, though not likely to get to the final mixup, is fully deserving of support. Then move on to my other top candidates - who will I believe already have been elected (but just in case). If they have been elected my preferences for them will be ignored and the vote end up with the person I believe will in a battle for the last seat.

I'm not an expert but that seems like a good strategy to me.

Turkey-raising has the advantage that it is simple and safe (provided the turkey is definitely a no hoper). In theory there are other ways to vote tactically but they are difficult to pull off, and they can backfire on you and elect the wrong candidate.

Another thing to bear in mind in this country is state funding. As I understand it, a party needs to get at least 2% of the first preference vote on a national basis in order to qualify for funding. So if a party you like is at risk of falling just below the 2% mark nationally it might be worth giving them a first preference for that reason.

http://www.sipo.gov.ie/en/Guidelines/StateFinancing/GuidelinesforPoliticalPartiesonExchequerfundingundertheElectoralActsMarch2009edition/Name,2146,en.htm