PR-STV: voting transfers

gaiscioch

I've a specific question about how PR-STV works. If I vote 1) Murphy 2) Burke 3) Duffy and Murphy is eliminated, does my full vote get transferred to Burke in that count providing Burke is still in the race? If he's not in the race, does my full vote transfer to Duffy in that count?

ArmaniJeanss

Yes to both questions.
If your vote transfers by elimination, then its the full vote that transfers.
And if the next candidate on your list is already eliminated then it transfers to the next available on your list.

gaiscioch

ArmaniJeanss wrote: »

Yes to both questions.
If your vote transfers by elimination, then its the full vote that transfers.
And if the next candidate on your list is already eliminated then it transfers to the next available on your list.

Excellent. Thank you. Now, if my first vote is for a candidate who has, say, 500 votes surplus. Does the system transfer to the next preference of each of the final 500 votes counted, or does something else happen?

cml387

gaiscioch wrote: »

Excellent. Thank you. Now, if my first vote is for a candidate who has, say, 500 votes surplus. Does the system transfer to the next preference of each of the final 500 votes counted, or does something else happen?

It may.If you want the whole thing explained in detail, here's some reading for you.

Poxyshamrock

Quick question. If a candidate exceeds the quota by let's say 600 votes. Those 600 votes are then transferred. My question is how do they determine which 600 ballot papers are transferred?

NuMarvel

Poxyshamrock wrote: »

Quick question. If a candidate exceeds the quota by let's say 600 votes. Those 600 votes are then transferred. My question is how do they determine which 600 ballot papers are transferred?

They distribute the 600 votes in proportion to the next preferences indicated on all of the winning candidate's ballot papers.

So if the proportion of next preferences worked out at 10% for John, 18% for Mary, and so on, then the 600 votes are distributed on that basis.

gaiscioch

NuMarvel wrote: »

They distribute the 600 votes in proportion to the next preferences indicated on all of the winning candidate's ballot papers.

So if the proportion of next preferences worked out at 10% for John, 18% for Mary, and so on, then the 600 votes are distributed on that basis.

Ahh. So they have to recount all the votes again in order to get the overall breakdown of second preferences, or do they note 1st and 2nd preferences simultaneously when doing the first count?

Carawaystick

The only have to count the elected person's votes again, not every vote.

ArmaniJeanss

NuMarvel wrote: »

They distribute the 600 votes in proportion to the next preferences indicated on all of the winning candidate's ballot papers.

So if the proportion of next preferences worked out at 10% for John, 18% for Mary, and so on, then the 600 votes are distributed on that basis.

This appears to be incorrect according to the earlier link in cml387s post (section 3.IV of link).
It appears they just take the last parcel(or parcels, to exceed 600) of counted votes for the elected candidate and regard these as the surplus.
Must surely be a risk of these being a freak parcel with a non-representive transfer.

Godge

ArmaniJeanss wrote: »

This appears to be incorrect according to the earlier link in cml387s post (section 3.IV of link).
It appears they just take the last 600 counted votes for the elected candidate and regard these as the surplus.
Must surely be a risk of these being a freak 600 votes with a non-representive transfer.

Not the last 600, the last bundle he received.

If the candidate is elected on the first count then it is all of his votes.

If he is elected on a subsequent count, it is the last vote he received.

This may well become an issue in the European counts because of the huge geographical disparity.

gaiscioch

Godge wrote: »

Not the last 600, the last bundle he received.

If the candidate is elected on the first count then it is all of his votes.

If he is elected on a subsequent count, it is the last vote he received.

This may well become an issue in the European counts because of the huge geographical disparity.

Interesting; I presume 'last vote' means all the votes from that 'last count'. Are all the EU votes for, say, MNW not brought to the one count centre?

Godge

gaiscioch wrote: »

Interesting; I presume 'last vote' means all the votes from that 'last count'. Are all the EU votes for, say, MNW not brought to the one count centre?

Yes,

The issue is with party loyalty versus geographical loyalty.

Say there are three FG candidates, one from Kerry, one from Cork and one from Wicklow. The candidate from Cork gets elected. The last votes that candidate gets are from a FF candidate in Cork. The distribution of the surplus is of the last bundle that the candidate gets. The votes being distributed are really Cork votes not FG votes. These votes "leak" to other parties much more than the FG votes the candidate got in the first count.

That is the type of distortion that can happen with the current system.

One of the big advantages of the electronic voting system was that it could get rid of this anomaly. It is quite a difficult logistical issue to go back through every bundle of votes in a manual count and allocate accordingly. The machine can do it automatically.

ArmaniJeanss

So MNW is about to go into the final count, the distribution of Cartys surplus.

And it would appear that only Bryne->Carty votes will make up this surplus?
It does seem an unfair method.

Fred Swanson

This post has been deleted.

Robert McGrath

ArmaniJeanss wrote: »

So MNW is about to go into the final count, the distribution of Cartys surplus.

And it would appear that only Bryne->Carty votes will make up this surplus?
It does seem an unfair method.

It potentially means that a full recount can produce a different result as well (I think) ... Not because of a clerical error by the counters first time round, but because a different random batch of surplus votes can be selected for distribution second time round.

Riskymove

Basically the way to look at it is:

number one to the person you most want elected

if he gets eliminated you want to see 2 elected and so on


If 1 is elected then that's all your vote counted for regardless of how many preferences you gave

Godge

Robert McGrath wrote: »

It potentially means that a full recount can produce a different result as well (I think) ... Not because of a clerical error by the counters first time round, but because a different random batch of surplus votes can be selected for distribution second time round.

Not exactly, if Byrnes votes elect Carthy, it will always be that bundle which determines where Carhy's surplus goes.

In a close election, this may be crucial.

The only way it could change is if Carthy picks up other votes in a recount (or a recheck) that allow him to be elected before Byrne's transfers. In that case, he will have a different surplus to distribute.

gaiscioch

Riskymove wrote: »

Basically the way to look at it is:

number one to the person you most want elected

if he gets eliminated you want to see 2 elected and so on


If 1 is elected then that's all your vote counted for regardless of how many preferences you gave

On the other hand say if somebody knows it's likely Sinn Féin will get a seat but they care about environmental issues even though it's unlikely the GP candidate will be elected. Could they vote GP 1 and SF 2 and that way the SF candidate will see that he received a certain number of GP transfers and he might therefore be more willing to integrate green issues into his politics to keep such transfers coming? In other words, you can give a message of what way you'd like to see their politics develop by giving them No 2 rather than No 1.

echo beach

gaiscioch wrote: »

On the other hand say if somebody knows it's likely Sinn Féin will get a seat but they care about environmental issues even though it's unlikely the GP candidate will be elected. Could they vote GP 1 and SF 2 and that way the SF candidate will see that he received a certain number of GP transfers and he might therefore be more willing to integrate green issues into his politics to keep such transfers coming? In other words, you can give a message of what way you'd like to see their politics develop by giving them No 2 rather than No 1.

There is a certain logic to this and many people vote on the basis of what they expect the outcome to be rather than what they want it to be, so they may not give their no 1 to a candidate they expect will get in anyway and parties try to 'manage' the vote.
However this is dangerous as until the votes are cast nobody can know who will be elected, as was shown in several close outcomes. Many candidates hate to ride high in opinion polls and will present their seat as being in danger to get the maximum vote out. Because opinion polls can bias voter choice many feel they should be limited and perhaps banned for a certain period before polling day.

The motivation behind transfers is so difficult to work out that it is unlikely any party would base their policies on where transfers came from. In the example you give SF have no way of knowing if you voted Green because you care about the environment or because you know the candidate personally.

Sometimes you can over-analyse a simple instruction, 'vote 1,2,3, etc in order of YOUR preference'.

Jayop

I'm going to bump this thread rather than start a new one because it's a question that I'm still not overly happy with the contradicting answers I've been given.

PR-STV
http://spunout.ie/news/article/our-voting-system

So I know it's spunout.ie but I'm sure they have researched that article and it's accurate.

My problem is that when I vote I tend to go right down the ballad the whole way to whomever I least want elected. I was told that's the best way to ensure your vote counts. However going by the information on that they only take the second and subsequent preferences on the votes over the quote and not a percentage of the entire quota.

Quota 1000
Candidate A gets 1100

By what they are saying, the first 1000 votes are locked in and the 100 over are the only ones which they count for the transfers.

From Extra 100 second preference
Candidate B 90
Candidate C 10


Surely that's nonsense. Surely they should count the entire 1100 second preferences and calculate the percentage of votes to be transferred to B & C from there. If Candidate A's entire second preference was a 50/50 split then because of an irregular bundle B & C are getting a 90/10 split.


Hopefully my point/question makes sense and someone can put my mind at ease.

Jayop

I have another question and this one is even more complicated in my mind.

So Candidate D is being eliminated and his transfers are being split. Do they count all the transfers he's already received and go one further down the ballad or do those voted essentially get dumped and they only count the original first preferences that he got??

jlang

Jayop wrote: »

However going by the information on that they only take the second and subsequent preferences on the votes over the quote and not a percentage of the entire quota.

Quota 1000
Candidate A gets 1100

By what they are saying, the first 1000 votes are locked in and the 100 over are the only ones which they count for the transfers.

From Extra 100 second preference
Candidate B 90
Candidate C 10


Surely that's nonsense. Surely they should count the entire 1100 second preferences and calculate the percentage of votes to be transferred to B & C from there. If Candidate A's entire second preference was a 50/50 split then because of an irregular bundle B & C are getting a 90/10 split.

Onb the first count, they check each one but the next lower preferences are not checked. The assumption is that the original mixing was randomisation enough. If elected on the first count, each vote had as much of a chance of being on top as any other.

If elected on a later count, only the bundles on top (that just got transferred in) get looked at again. The lower preferences for the original vote do not get looked at. That vote has already elected someone and will stay in their pile.

Jayop wrote: »

So Candidate D is being eliminated and his transfers are being split. Do they count all the transfers he's already received and go one further down the ballad or do those voted essentially get dumped and they only count the original first preferences that he got??

All the votes in his pile get their next preference checked and distributed appropriately.

A compulter could do it better by looking at all the preferences on all the votes and transferring fractional votes, but that is not the way the rules of the count are written.

In fact, in the one election where computer counting was used, instead of implementing a better/fairer system they had to correctly implement the official rules (mixing, transferring the last pile in, etc) to make it equivalent to a hand count.

johnnyskeleton

Jayop wrote: »

Quota 1000
Candidate A gets 1100

By what they are saying, the first 1000 votes are locked in and the 100 over are the only ones which they count for the transfers.

From Extra 100 second preference
Candidate B 90
Candidate C 10


Surely that's nonsense. Surely they should count the entire 1100 second preferences and calculate the percentage of votes to be transferred to B & C from there. If Candidate A's entire second preference was a 50/50 split then because of an irregular bundle B & C are getting a 90/10 split.


Hopefully my point/question makes sense and someone can put my mind at ease.

http://www.environ.ie/en/Publications/LocalGovernment/Voting/FileDownLoad,1895,en.pdf

That's the guidelines on how practically it should be done. Is it as precise as one may theoretically hope? I'm afraid not.

It is however a practical way of calculating the vote in a timely manner.

In your scenario, Candidate C is entitled to request a recount if he/she feels that the surplus hasn't been distributed correctly.

In respect of your specific query, if the 1100 are all 1st prefs, then they will count all 1100 and if they are 50:50 they will give the first 50 in the respective piles to each candidate.

But if candidate A got 900 in the first vote and then candidate D is eliminated, transferring all his 200 votes to candidate A, then those 200 votes will be looked at. If, of them, 90% vote for candidate B and 10% vote for candidate C, they will take the top 90 votes from candidate B's pile and then the top 10 from candidate C's pile.

mewso

Not the clearest of articles on spunout. Citizens Information might be a bit more helpful. Or for even more detail this document on the Dept. Of Environment site is worth a look.

For a surplus all* of that candidates next preferences are counted and percentages calculated to apportion the surplus amount so it is incorrect to think that a random 100 votes are pealed off the top. All 1100 votes in your example would have the next preference counted to determine the percentages.

Re. the elimination all of that candidates votes are distributed. It's important to bear in mind that voting papers are moved from count to count. It's a while since I worked on a count but when the surplus in the first example is calculated papers will be physically moved to the candidate. So if it was calculated that they had 10% next preferences from elected candidate then 10% of those papers will be moved to that candidate (off the top) so that if and when that candidate is eliminated the next preference will be counted. When counting an eliminated candidates' papers nobody above that candidate on a paper will still be in play if all has gone correctly. i.e. you look at one of their votes and they are number 5 on it. 1 to 4 on that paper will be either elected or eliminated candidates.

Open to correction of course but from memory this is how it goes.

* correction - on the first count

Jayop

Thanks for that. Personally I think that assumption you mentioned is an absolute joke. You can't just assume that a 9% sample of votes are going to be the same as all the other votes.

Correct me if I'm wrong then;

It's most likely that the last votes to be counted are all going to come from the same polling booth. If there's a local candidate there surely the chances are they will poll higher than average.

Really if your polling box is the last to be counted there's much more chance of your preferred candidates being elected.

4ensic15

Jayop wrote: »

Thanks for that. Personally I think that assumption you mentioned is an absolute joke. You can't just assume that a 9% sample of votes are going to be the same as all the other votes.

Correct me if I'm wrong then;

It's most likely that the last votes to be counted are all going to come from the same polling booth. If there's a local candidate there surely the chances are they will poll higher than average.

Really if your polling box is the last to be counted there's much more chance of your preferred candidates being elected.

It is the last seats that matter. It often happens that the second last candidate elected exceeds the quota and their surplus will then decide the destination of the last seat. The surplus will be the top of the last transfers that the second last candidate got. They will have come from the same source which elected the second last candidate and will benefit someone similar in part or geography.

Jayop

4ensic15 wrote: »

It is the last seats that matter. It often happens that the second last candidate elected exceeds the quota and their surplus will then decide the destination of the last seat. The surplus will be the top of the last transfers that the second last candidate got. They will have come from the same source which elected the second last candidate and will benefit someone similar in part or geography.

It's an interesting one.

Personally I feel the list system could be a better option, but with so many indo's getting elected in Ireland it'll not work.

4ensic15

Jayop wrote: »

It's an interesting one.

Personally I feel the list system could be a better option, but with so many indo's getting elected in Ireland it'll not work.

The senate elections are fully proportional and so were the computerised elections.
There were 2 problems with the computerised voting

1. Security faults.
2. there was no possible manual validation.

What would be needed would be a system where voters would choose their preferences on a screen and the vote would be printed out. Dropping the vote into the box would complete the process. The computer could do the count with its record of the votes printed out and the hard copies would be available as backup if there was a challenge to the result.

oscarBravo

4ensic15 wrote: »

What would be needed would be a system where voters would choose their preferences on a screen and the vote would be printed out. Dropping the vote into the box would complete the process. The computer could do the count with its record of the votes printed out and the hard copies would be available as backup if there was a challenge to the result.

That's something approaching a least-worst design for computerised voting, but it's not without potential problems. What happens if when a printed ballot doesn't match the preferences selected?

Brussels Sprout

NuMarvel wrote: »

They distribute the 600 votes in proportion to the next preferences indicated on all of the winning candidate's ballot papers.

There was a lady on the Pat Kenny show yesterday discussing the mechanics of the system and she claimed that they just took 600 "Off the top of the pile" rather than working them out proportionally from the entire batch. I hope you're correct as this would be a a fairer representation.

oscarBravo

Brussels Sprout wrote: »

There was a lady on the Pat Kenny show yesterday discussing the mechanics of the system and she claimed that they just took 600 "Off the top of the pile" rather than working them out proportionally from the entire batch.

As I understand it, the "pile" is created by shuffling the ballots from all the boxes, so they are somewhat randomised before they are picked off the top.

Sam Russell

oscarBravo wrote: »

As I understand it, the "pile" is created by shuffling the ballots from all the boxes, so they are somewhat randomised before they are picked off the top.

It is this randomised format that allows the whole system to work.

The council votes for Listowel were recounted in the last week or so as the final seat was won by two votes. The Supreme Court decided that votes had to start with a '1' as some had continued their preferences from the Euro vote at '3' or '4'. This eliminated a large (compared to 2) votes but the recount did not change the result.

If a consistent scheme is followed, then it does not matter. A complete recount, following reshuffling the votes could give a different result, but it normally is just a recheck - that is a check that all bundles are correct and in the right pile.

Sam Russell

oscarBravo wrote: »

That's something approaching a least-worst design for computerised voting, but it's not without potential problems. What happens if when a printed ballot doesn't match the preferences selected?

The whole eVoting fiasco was a result of trying to answer the wrong question.

What was needed was eCounting, not eVoting. We needed a system of counting the votes, instead we bought a whole load of obsolete voting machines that had a limited life of 5 years and were going to have reliability problems beyond that, and did not own the software written by a Dutch company that did not understand the voting system. No wonder it was thrown out.

The answer to your question - 'What happens if the printed ballot does not match the the electronic vote?' - well, they go with the printed ballot.

oscarBravo

Sam Russell wrote: »

The answer to your question - 'What happens if the printed ballot does not match the the electronic vote?' - well, they go with the printed ballot.

Well, yeah. But what happens in the polling booth? I select candidates in a particular order (which is the order I want), the machine says it has recorded my preference, but it prints out something different for me to drop in the ballot box. What's the procedure?

Yes, this is an edge case - but if a system doesn't have a way to handle edge cases like this, then that system doesn't work.

Another example: what happens if a routine check of paper ballots finds a large discrepancy from the electronically-recorded tally that can't be accounted for? Does that cast doubt over every electronic count? Will every paper ballot have to be hand-counted to get a definitive result?

I'm not asking you specifically to answer these questions in relation to this hypothetical voting system; I'm pointing out that if the design of a system doesn't include answers to these and dozens of other such scenarios, then that design isn't fit for purpose.

Jayop

oscarBravo wrote: »

Well, yeah. But what happens in the polling booth? I select candidates in a particular order (which is the order I want), the machine says it has recorded my preference, but it prints out something different for me to drop in the ballot box. What's the procedure?

Yes, this is an edge case - but if a system doesn't have a way to handle edge cases like this, then that system doesn't work.

Another example: what happens if a routine check of paper ballots finds a large discrepancy from the electronically-recorded tally that can't be accounted for? Does that cast doubt over every electronic count? Will every paper ballot have to be hand-counted to get a definitive result?

I'm not asking you specifically to answer these questions in relation to this hypothetical voting system; I'm pointing out that if the design of a system doesn't include answers to these and dozens of other such scenarios, then that design isn't fit for purpose.

Surely a machine could be capable of taking the printed ballad back and destroying it and giving you a new one with the correct selection.

Not that any of it matter though, we're never going to go down that road again.

Sam Russell

Alternatively we could go the route of machine reading the current hand marked ballot papers. Each paper is scanned, the official verifies it is the same as the hand written one and it is entered. Computer then does the counting.

Possible, but the distrust will still be there so it will not happen.

oscarBravo

Jayop wrote: »

Surely a machine could be capable of taking the printed ballad back and destroying it and giving you a new one with the correct selection.

...and if it gives you another wrong printout?

I'm deliberately labouring the point, because this is a pet peeve of mine. There's a tendency to insist that eVoting must be a simple problem to solve, when it's actually a problem that approaches impossibility.

Sam Russell wrote: »

Alternatively we could go the route of machine reading the current hand marked ballot papers. Each paper is scanned, the official verifies it is the same as the hand written one and it is entered. Computer then does the counting.

Works for me. 90% of ballots could probably be scanned without controversy; the remainder would be interpreted by consensus among the count staff and the tallymen.

Jayop

Sam Russell wrote: »

Alternatively we could go the route of machine reading the current hand marked ballot papers. Each paper is scanned, the official verifies it is the same as the hand written one and it is entered. Computer then does the counting.

Possible, but the distrust will still be there so it will not happen.

That's really the only proper way of doing it. Completely verifiable.

If the count is in doubt with the computer then just recount the paper ballads. Also, if there's people there from all the candidates monitoring them being scanned in then it should be fairly foolproof.

Sam Russell

Jayop wrote: »

That's really the only proper way of doing it. Completely verifiable.

If the count is in doubt with the computer then just recount the paper ballads. Also, if there's people there from all the candidates monitoring them being scanned in then it should be fairly foolproof.

I suppose that is why there is a song and dance about it.

Jayop

Sam Russell wrote: »

I suppose that is why there is a song and dance about it.

( I don't think that's against the charter)

cml387

It should be remembered that in all the cases that I can think of, recounts have only ever differed by a handful of votes(the problem is of course if candidates are only separated by a handful of votes).
So the system does work and computerisation would only speed up the process,it wouldn't be any fairer.

4ensic15

cml387 wrote: »

It should be remembered that in all the cases that I can think of, recounts have only ever differed by a handful of votes(the problem is of course if candidates are only separated by a handful of votes).
So the system does work and computerisation would only speed up the process,it wouldn't be any fairer.

The computerised system was fully proportional the whole way through as ooposed to the manual system which has random elements. the computerised system also eliminates spoiled votes and so is more certain.

Scofflaw

4ensic15 wrote: »

The computerised system was fully proportional the whole way through as ooposed to the manual system which has random elements. the computerised system also eliminates spoiled votes and so is more certain.

However, the only thing that would make it acceptable is a nice big sleep() instruction so that there was a decent interval for speculation between counts.

cordially,
Scofflaw

Charles Babbage

The RTÉ report on the Sligo-Leitrim count suggested that the particular sample of votes used to distribute a surplus were ones recently transferred from an eliminated candidate, and that this had a material effect on the candidate the votes were transferred to.

This is rather unsatisfactory, if a sample is to be used then they should be require to shuffle them a bit. Either than or use the NI system of counting all of them and dividing proportionally.

4ensic15

Charles Babbage wrote: »

The RTÉ report on the Sligo-Leitrim count suggested that the particular sample of votes used to distribute a surplus were ones recently transferred from an eliminated candidate, and that this had a material effect on the candidate the votes were transferred to.

This is rather unsatisfactory, if a sample is to be used then they should be require to shuffle them a bit. Either than or use the NI system of counting all of them and dividing proportionally.

That is the standard practice. The electronic system was truly proportional. The counts take long enough as it is. In this case the surplus and the underlying vote would be likely to favour the eventual beneficiary to a similar extent as it was the distribution of a party colleague's surplus which elected the last candidate.

Sleepy

Question: in a full re-count are all of the votes of an elected candidate checked before transferring the surplus or is the existing "random" selection just re-checked?

Sam Russell

A real recount requires votes to be returned to the table and reshuffled. That is rare because it takes ages.

A recheck is where bundles are checked to look for mistakes of various sorts.

The real problem lies in the allocations of surplus votes. If a candidate gets 1000 votes from a excluded candidate, and only needs 700 to get a quota, which 300 votes get picked for redistribution? Is it a proportion of those 1,000 or is it the last 300 votes which may have come from one candidates votes? Could make a big difference as it may result in one candidate being next to be excluded over another.


High Court Thursday may have to decide.

4ensic15

Sam Russell wrote: »

A real recount requires votes to be returned to the table and reshuffled. That is rare because it takes ages.

A recheck is where bundles are checked to look for mistakes of various sorts.

The real problem lies in the allocations of surplus votes. If a candidate gets 1000 votes from a excluded candidate, and only needs 700 to get a quota, which 300 votes get picked for redistribution? Is it a proportion of those 1,000 or is it the last 300 votes which may have come from one candidates votes? Could make a big difference as it may result in one candidate being next to be excluded over another.


High Court Thursday may have to decide.

It is the top of the pile. Votes are kept in the order they are in when opened and place in each candidates pile.

Sam Russell

4ensic15 wrote: »

It is the top of the pile. Votes are kept in the order they are in when opened and place in each candidates pile.

Quite, but 'top of the pile' could make a difference, as the source of those votes could differ, depending on the order they were placed there.

Take my example and explain it. Do they take a proportion of the 1,000 or do they just take the 300 from the top of the pile. Could make a difference of huge proportions (if you pardon the pun).

The supreme court just adjudicated on voting, ruling that ballots that do not start with a '1' are invalid even though the intention of the voter is clear.

They may have to rule on this.

echo beach

Sam Russell wrote: »

Quite, but 'top of the pile' could make a difference, as the source of those votes could differ, depending on the order they were placed there.

When the ballot boxes are opened, one by one, all the votes are checked to make sure the number in each box agrees with the number cast and to remove the spoiled votes. This is when the 'tallies' you hear reported are done.

The votes are then split into those for each candidate and it is at this point that they get mixed together or 'randomised'. These votes are then counted in bundles or 50 or 100 and it is these bundles that are put into 'piles'. The first preferences for a candidate who doesn't reach the quota on the first count stay in his or her 'pile' until such time as that candidate is elected or eliminated.
When a candidate is eliminated their no 2 votes are distributed and get put on top of the 'pile' for their second choice and so on as the eliminations continue. The pile for those remaining is getting bigger until it eventually gets to the quota, usually with some over. It is these extra votes that make up the surplus and if you can imagine all the ballot papers moving around the count centre into the various piles then it is clear that only the ones from the last pile added could be part of the surplus. By the time they form that surplus then they have been mixed or 'randomised' several times.

Brussels Sprout

Apologies if this has been asked already but in the case of a full recount are the votes "randomised" again so that the votes chosen for a surplus could be different from the first count?


Also, is there a difference between a recheck and a recount?