Still seeking to understand the system

[Christel]

What system do you use in your home country and do you think it is better?

I have to admit I don't know. I'm away from it many years, and wasn't too interested in electoral systems when I was still there. Here's a wikipedia link:

Mixed member proportional representation - Wikipedia, the free encyclopedia

I suppose I could ask as many questions about this.

But I liked the simplicity of the Gemeinderatswahlen in Baden Wuerttemberg, which I was told about by a visitor from Germany recently. She had 40 votes (there were a few hundred candidates), which she could distribute by giving 1, 2 or 3 votes to candidates of choice. I learned today from wikipedia that it's called first-past-the-post, i.e. those with highest amount of votes are elected.

[Ballyfermot Man]

That's an odd one alright. I was thinking of one way in which it might happen and I think I had it but lost my train of thought. Could another way be if you vote Candidate A #1, B #2 and C #3, C being your preferred candidate, and A and B transferring heavily to other candidates who are in competition with C for a seat, and your first and second preferences help push A and B over the quote meaning that instead of them being eliminated and all their votes being redistributed, only their surpluses are which would be considerably smaller?

I'm very tired so that probably doesn't work, I dunno.

A classic example of this occured in Carlow Kilkenny in 2007... if more people had voted Fergal Browne No. 1 than JP Phelan, he would subsequently have kept ahead of the 2nd Labour candidate, soak up his transfers, and have a significantly larger dollop of preferences to pass to Phelan.

But JP got a biteen greedy and poached....

[caulfield-the-yank]

I think stringjack is correct - because the order in which candidates are eliminated is important. If putting the first two preferences to other candidates would have meant the 'right' candidates were eliminated, things could theoretically have gone better for the preferred candidate. On the other hand, it takes hindsight to see what would have worked...

With a GE coming up, this topic becomes timely again.

Yes, Stringjack is correct.

The surplus of a candidate who wins on the first count (e.g., Bertie in Dublin Central in 2007) is distributed in proportion to the second preferences of all his voters.

But candidates who go over quota on subsequent counts only have "their" surplus chosen from among the last "packet" of votes that put that candidate over the quota.

Take the example of Dublin Central in 2007.

If a left-leaning voter's first choice was Tony Gregory, but he wished just as much to see Mary Lou McDonald (SF) elected after Gregory, or, if that was not possible, Joe Costello (Lab) beat Cyprian Brady (FF) and Paschal Donohue (FG), the only way that voter could have any chance of having all those preferences count would be to give a higher preference to Frances Fitzpatrick (FF) than to any of the three candidates he actually preferred.

If he voted for Gregory #1, his vote would stay in that pile, no help to anyone else (unless, like Bertie, Gregory had also gone over the quota on the first count).

If he voted McDonald #1, he would risk the chance (however remote) of McDonald beating Gregory by 1 vote for the fifth seat.

Ditto, if he voted Costello #1. That would risk either Gregory or McDonald losing out to Costello for the fifth seat.

If he "strategically" voted for McKenna (for instance), and then for his actual three choices in order, his vote would again go to Gregory, and then stay there, no help to anyone else, because the distribution of McKenna's vote did not put Gregory (or anyone else) over the quota.

As Gregory was not likely to be elected on the first count, it was only by being among the "packet" of votes that put Gregory over the quota that this voter's ballot could have had any chance of being transferred from Gregory to McDonald. (And since McDonald would be eliminated after that count, the vote would then go to Costello.)

But the only way that ballot could be in the packet that would put Gregory over quota would be if it gave Fitzpatrick a higher preference than any of the three candidates this hypothetical voter actually favored.

Admittedly, it is only after the votes are counted that this voter would know for sure that it was a relatively high preference vote for Fitzpatrick that was the "magic bullet" needed to effectuate all of his preferences. I realize that. But it still seems bizarre that it would be voters who preferred Fitzpatrick to Gregory (or at least voted that way) who would have the sole say over how "Gregory's surplus" would be distributed.

A better way, it seems to me, would be to count all of Gregory's votes in calculating the distribution and destination of of his surplus -- just as, for example, Bertie's surplus had been calculated after his first-count victory.

[SideysGhost]

A better way, it seems to me, would be to count all of Gregory's votes in calculating the distribution and destination of of his surplus -- just as, for example, Bertie's surplus had been calculated after his first-count victory.

There are other oddities in the southern system too like random sampling of votes to determine the destination of transfers from eliminated candidates, IIRC. Madly, the FF/PD Govt actually requested that the company supplying the e-voting machines re-write the code to keep random sampling in rather than take the opportunity to fix it and count all lower preferences properly and completely.

I've been told in the PR-STV election in the North though, all votes are counted and transfers done properly. Begs the question as to why the south has such opaque idiosyncrasies and why they've never been fixed when the job gets done properly a few miles up the road...

[caulfield-the-yank]

There are other oddities in the southern system too like random sampling of votes to determine the destination of transfers from eliminated candidates, IIRC. Madly, the FF/PD Govt actually requested that the company supplying the e-voting machines re-write the code to keep random sampling in rather than take the opportunity to fix it and count all lower preferences properly and completely.

Why in God's name would "random sampling" be needed to determine the destination of the ballots of an eliminated candidate?

All of those ballots (if transferrable at all) get transferred.

I can understand with surpluses. The case for the present system might be taht it is unduly burdensome to reexamine 11,000 ballots in order to determine how to allocate the 300 or so that might be "surplus".

But when a candidate is eliminated, ALL of his or her transfereable ballots are transferred, whether that number is 500 or 5000.

So if the RO determines (by whatever means) that 200 of A's 1000 votes go to B and 800 to C, don't 200 votes have to be physically switched from A's pile to B's pile, and 800 physically switched from A's pile to C's?

So don't all of the eliminated candidate's votes have to physically handled in the transfer process?

Then why not simply send each one to the destination indicated by the voter?

[caulfield-the-yank]

There are other oddities in the southern system too like random sampling of votes to determine the destination of transfers from eliminated candidates, IIRC. Madly, the FF/PD Govt actually requested that the company supplying the e-voting machines re-write the code to keep random sampling in rather than take the opportunity to fix it and count all lower preferences properly and completely.

May I safely assume that "random sampling" of the votes of an eliminated candidate is only done prior to the final count that is mathematically certain to fill the remaining seat (or seats)?

That would be the only circumstance under which, arguably (if you could infer the distribution from the sample), you would not have to physically move each and every one of the eliminated candidate's ballots.