[QUOTE=“Juhniallio, post: 949811, member: 53”]They don’t just take the top 10% of the pile. They take 10% at random of the votes. I checked with a returning officer type on saturday.
Your scenario might lead to the entire 10% coming from one area of the constituency, the last one counted (eg FF loving blackrock.)Their second preferences might be very different to those of, say, Blanchardstown folk. To ensure the surplus is distributed as fairly as possible, the 10% is taken at random.[/QUOTE]
Actually I’m going back to something closer to my first answer.
This is how I think it works:
Boylan got over the quota by say 3,000 based on getting 5,000 transferred to her from Smith. That 3,000 can be distributed. If 1,000 can’t go any further because Boylan is the lowest preference at #2 on those ballots then there is 2,000 to distribute. The full 4,000 votes are counted however to work out the % distribution to each candidate. So if the 4,000 transferable votes from that bundle of 5,000 shows 3,000 to Murphy and 1,000 to Childers then that ratio of 75:25 is applied to the transferrable votes. So you get 1,500 to Murphy (75% of 2,000) and 500 to Childers (25% 2,000).
The actual papers they use to transfer across are selected, and that’s probably the random bit but I’m sure I’ve heard criticism of returning officers before for using different methodologies there. I think it is sometimes the case that they will just choose the first 1,500 ballot papers they can find with Smith #1 Boylan #2 and Murphy #3 and the first 500 ballots papers they can find with Smith #1 Boylan #2 and Childers #3 but I guess whether that is considered random or the top of the pile is arguable - and it might be that they do a properly random approach, in at least some count centres.
And those votes ballots are then used for the next count etc. But the ratio in which they are distributed is based on the percentages of the full amount of votes received, not a random selection, and not the top of the pile either.
That could be wrong but I’m very rarely wrong.