Board of Governors Vote opens 15/08/2011 23:59

[tempest3k]

Dear Member,

The member vote for the 4 vacancies on the board of Governors will open at 23:59 tonight (15/08/11): https://www.pirateparty.org.uk/vote/56/

For this election we will be using the STV voting system. You may select candidates in order of preference rank; 1=1st choice, 2=2nd choice, etc, for as many as you wish to express a preference for. The order you rank candidates must be in sequence.

The vote will be carried out and counted using the OpenSTV software and the Electoral Reform Society 1997 guidelines.

There are 6 candidates for 4 positions. You may also choose to vote to reopen nominations. The vote will run for two weeks, until 00:01 on 29/08/2011.

Election monitors appointed by the board for this vote are Shaun Dyer - s.dyer@pirateparty.org.uk (Governor) and Daniel Hertz - d.hertz@pirateparty.org.uk (London RAO).

If you have any technical problems please contact e.geraghty@pirateparty.org.uk. If you have any other questions, please advise the nominating officer (p.cooke@pirateparty.org.uk) or either of the election monitors, who will attempt to resolve any issues and answer any questions.

Regards,

Phil Cooke
Nominating Officer
Pirate Party UK

[scuzzmonkey]

cheers Phil

[cabalamat]

The vote will be carried out and counted using the OpenSTV software and the Electoral Reform Society 1997 guidelines.

There are 6 candidates for 4 positions. You may also choose to vote to reopen nominations.

I believe allowing re-open nominations (RON) to be a mistake and, more importantly, a violation of our constitution.

We should not allow RON because our constitution says to use the ERS 1997 rules and the ERS 1997 rules don't include RON. In addition, previous PPUK multi-member STV elections have not had RON (this was discussed by the NEC before the 2010 election for the BOG, when I pointed out that RON is problematic for multi-member elections, and it was agreed not to have RON).

The ERS 1997 rules don't specifically mention RON, but it is implied that RON isn't an option, e.g. in section 2.2 where voting papers don't include it. RON is problematic for multi-member elections because if it is interpreted as meaning that seats below RON don't get filled, it can distort the proportionality goal of STV. For example, in a 12 member STV election, any faction supported by more than 1/13th of the voters can win a seat, but if RON is allowed, then a large faction can deprive a smaller faction of its proportional share of seats.

The other interpretation of RON is that if it wins more first preferences than any other candidate, nominations are re-opened, and the election held again. Under this interpretation, RON doesn't break proportionality. It is not clear what interpretation of RON is intended for this election.

Update: I have emailed Phil, Shaun, Daniel and Ed, asking them to look into this issue.

[tempest3k]

I have referred this to the board for clarification, as it has been raised as a constitution issue.

[tuoni]

Update: I have emailed Phil, Shaun, Daniel and Ed, asking them to look into this issue.

I have received and read this e-mail, however my only responsibility in the votes is to provide and maintain the platform and any problems which arise through that - the implementation of votes themselves is entirely down to the Nominations Officer and the election monitors.

[M2Ys4U]

Hi all,

Because of the issue that Philip Hunt raised in this thread, the Board is issuing a clarification on what will happen and how RON will be handled in this election.

The election will proceed as planned, with the option to re-open nominations.

Because the ERS97 procedure does not have a specific mechanism for dealing with RON, it shall be dealt with as if it were any other candidate, that is to say that if RON is deemed to be elected to the Board, then only one seat shall remain vacant, to be filled at a later date.

Following the conclusion of the election, the Board shall look at how, if at all, having an option for re-opening nominations shall work in multi-seat elections in the future.

We understand that this is not an ideal situation, but believe that the vote as it stands is constitutionally valid.

Regards,
Jack Allnutt
on behalf of the Board of Governors.

[cabalamat]

Hi all,

Because of the issue that Philip Hunt raised in this thread, the Board is issuing a clarification on what will happen and how RON will be handled in this election.

The election will proceed as planned, with the option to re-open nominations.

Because the ERS97 procedure does not have a specific mechanism for dealing with RON, it shall be dealt with as if it were any other candidate, that is to say that if RON is deemed to be elected to the Board, then only one seat shall remain vacant, to be filled at a later date.

That is, as far as I know, a new interpretation of RON for multi-seat elections. No matter.

Following the conclusion of the election, the Board shall look at how, if at all, having an option for re-opening nominations shall work in multi-seat elections in the future.

We understand that this is not an ideal situation, but believe that the vote as it stands is constitutionally valid.

I agree that it is not an ideal situation. However, I do not think the interests of the party will be served by delaying this election, so I am not going to contest this clarification.

When this election is over, I believe the highest priority for the board should be to review our procedures for elections. Whether I am elected to the board or not, I will be proposing constitutional amendments to that effect. As well as dealing with the length of the nomination period, and whether RON should be included in multi-seat elections (and if it is, what version of RON), we should also look at which varient of STV is used to count the election, because I have just noticed that the ERS97 rules are flawed.

The problem with ERS97 is that because of how it counts the quota (see section 5.1.6) it can sometimes elect more people than there are places. For example, if there are 4 places and 50 votes, the ERS97 quota is (50/(4+1))=10, and you could have 5 candidates each getting 10 1st preferences, so all 5 would be elected.

For this reason, many variants of STV use the Droop quota, calculated as:

Code: Select all

quota = int(numVotes/(numPlaces+1)) + 1


This is the smallest integer that guarantees that the number of candidates reaching the quota cannot be bigger than the number of places.

I think the Board would benefit if someone on it has a level of understanding of STV, and attention to detail, that they can pick up on these problems. I therefore ask Pirates to elect me to the board.

[azrael]

That is, as far as I know, a new interpretation of RON for multi-seat elections.

Is it? Can you cite your source for this?

The problem with ERS97 is that because of how it counts the quota (see section 5.1.6) it can sometimes elect more people than there are places. For example, if there are 4 places and 50 votes, the ERS97 quota is (50/(4+1))=10, and you could have 5 candidates each getting 10 1st preferences, so all 5 would be elected.

For this reason, many variants of STV use the Droop quota, calculated as:

Code: Select all

quota = int(numVotes/(numPlaces+1)) + 1


This is the smallest integer that guarantees that the number of candidates reaching the quota cannot be bigger than the number of places.

You are right that in a very rare circumstance you could have 5 candidates getting an equal number of votes, and each precisely matching the threshold/quota. The response is to break the tie by chance and eliminate a candidate. This isn't much different to flipping a coin, something which does happen in real elections.

Taking your example where using the droop (whole) method would provide a threshold of 11 - what happens when you have 5 equally matched candidates? Instead of all of them meeting the threshold, none of them do. How do you then decide which candidate to eliminate in order to redistribute votes? Standard method is to break the tie by chance and eliminate a candidate. Votes then redistribute during a second round of counting at which point you have the remaining four win.

There is one big difference between these two methods, and that isn't the result, but that the ESR97 method is quicker and easier.

I think the Board would benefit if someone on it has a level of understanding of STV, and attention to detail, that they can pick up on these problems. I therefore ask Pirates to elect me to the board.

I agree. I hope that if you are elected you will research the subject further.

[cabalamat]

That is, as far as I know, a new interpretation of RON for multi-seat elections.

Is it? Can you cite your source for this?

It's a new one known to me, so I don't have a source other than my personal knowledge. It wouldn't surprise me if someone else had suggested or used it at some point. Other interpretations of RON I've come across are:

1. once RON is elected, candidates elected before RON remain elected, and there is a new election for all currently-unfulfilled places.

2. if RON is elected, all of RON's surplus vote goes to RON(2), RON(3) etc so you can have multiple RONs elected. A new election is held for all places where RON is elected.

3. if RON wins the greatest number of first preferences, the election is suspended, nominations are re-opened, and a new election is held. Otherwise, all RON's votes go to their 2nd preferences, and the vote is held as normal.

RON seems to be mostly used in UK universities.

The problem with ERS97 is that because of how it counts the quota (see section 5.1.6) it can sometimes elect more people than there are places. For example, if there are 4 places and 50 votes, the ERS97 quota is (50/(4+1))=10, and you could have 5 candidates each getting 10 1st preferences, so all 5 would be elected.

For this reason, many variants of STV use the Droop quota, calculated as:

Code: Select all

quota = int(numVotes/(numPlaces+1)) + 1


This is the smallest integer that guarantees that the number of candidates reaching the quota cannot be bigger than the number of places.

You are right that in a very rare circumstance you could have 5 candidates getting an equal number of votes, and each precisely matching the threshold/quota. The response is to break the tie by chance and eliminate a candidate. This isn't much different to flipping a coin, something which does happen in real elections.

Taking your example where using the droop (whole) method would provide a threshold of 11 - what happens when you have 5 equally matched candidates? Instead of all of them meeting the threshold, none of them do. How do you then decide which candidate to eliminate in order to redistribute votes? Standard method is to break the tie by chance and eliminate a candidate. Votes then redistribute during a second round of counting at which point you have the remaining four win.

You're right, in this instance the result is the same. But even so, I don't like the ERS97 method, since the whole point of a quota is that any candidate who fulfils it can be safely elected: this isn't the case with ERS97, so their definition of a quota breaks what a quota is for.

There is one big difference between these two methods, and that isn't the result, but that the ESR97 method is quicker and easier.

Having a different quota does however make a difference in other circumstances: if a candidate gets lots of votes, then the lower the quota, the more of those votes are redistributed. So a lower quota helps those voters whose first preference is for a popular candidate, since they get more oomph out of their vote.

As to one method being quicker and easier, since they're being counted by computer, who cares? And both methods are O(n) anyway.

I think the Board would benefit if someone on it has a level of understanding of STV, and attention to detail, that they can pick up on these problems. I therefore ask Pirates to elect me to the board.

I agree. I hope that if you are elected you will research the subject further.

I intend to.

[borgs8472]

Can I go on the record to say I oppose STV in many situations because it's too complicated?

[azrael]

Phil/Cabalamat, I've seen examples of the types of RON use you describe. There does seem a general feeling that 'RON reaching quota stopping the vote and preventing all other places being filled' is unfair and against the spirit of STV. The fairer solution is the rollover RON. How that is achieved however is complicated. Either:
(1) you make X RON candidates where X is the number of seats up for grabs - and then hope voters who want to vote only for RONs do so in the 'correct order, or
(2) you use just one RON candidate in voting, and then transfer surplus votes to extra added RONs at count stage.
The second sounds like a better solution to avoid confusion to voters and would produce an election that didn't look any different to this current one. The software however can't handle it and would require manual intervention - probably a manual count - which isn't ideal.

In practice if it turns out that RON gets so many surplus votes in this election that a RON(2) (or more) would also have placed for a seat, I am willing to admit that a single non-rollover RON was not sufficient. If RON does get votes (even if not enough to reach threshold) I will feel that the inclusion gave members an option they felt met their needs.

I don't know if I'm able to get too worked up over a difference where you get the same result faster, but have possibly redefined terminology. If one system is redefining that reaching threshold isn't quite winning, the other system is redefining what a 'loser' is when eliminating a candidate randomly even though they have no fewer votes than the other candidates. Perhaps the disappointment for a candidate is less if they are eliminated before reaching threshold rather than after - but that's psychology

I think the vote system is complicated enough that we don't want it any more complicated (I guess to some people ... borgs ... it is probably so complicated to be indistinguishable from magic ). Yes it is being counted by a computer, but I like the idea that anyone can print off the ballot, sit down and cut it up, and run the vote manually to confirm it was accurate rather than rely on a black box giving the right answer. As such any additional complexity, or more voting rounds than necessary in my opinion obfuscates the process. Though I must throw in that any of the systems can require a tie being broken and OpenSTV gives the option of that person being selected by hand or randomly. I believe if there is ever a tie including RON, that RON should be manually chosen for elimination. If there is ever a tie between real candidates, I believe some other mechanism (trial by combat) would be better.

[tuoni]

I believe some other mechanism (trial by combat) would be better.

I like this system and hope to see it enshrined in our constitution.

[tempest3k]

I believe some other mechanism (trial by combat) would be better.

I like this system and hope to see it enshrined in our constitution.

^this

[cabalamat]

Phil/Cabalamat, I've seen examples of the types of RON use you describe. There does seem a general feeling that 'RON reaching quota stopping the vote and preventing all other places being filled' is unfair and against the spirit of STV.

Indeed. Obviously this isn't a problem in single-winner elections, because they aren't proportional anyway. There is only a problem if some people are elected (but not all the places), then the rest of the places filled in a separate election. So if it's a 12 seat election, and 11 members elected and one RON, followed by a 1 seat election, then that will be less proportional. Of course we could, as a Party, decide that proportionality isn't all that important, in which case it's not an issue.

The fairer solution is the rollover RON. How that is achieved however is complicated. Either:
(1) you make X RON candidates where X is the number of seats up for grabs - and then hope voters who want to vote only for RONs do so in the 'correct order, or
(2) you use just one RON candidate in voting, and then transfer surplus votes to extra added RONs at count stage.
The second sounds like a better solution to avoid confusion to voters and would produce an election that didn't look any different to this current one. The software however can't handle it and would require manual intervention - probably a manual count - which isn't ideal.

I realise the software can't handle multiple RONs so obviously doing it that way would be logistically difficult.

Of course, Openstv 1.6 is open source, so it would be possible to extend it to include RON. (I wonder, is there anyone with an IT background in the Party? )

In practice if it turns out that RON gets so many surplus votes in this election that a RON(2) (or more) would also have placed for a seat, I am willing to admit that a single non-rollover RON was not sufficient. If RON does get votes (even if not enough to reach threshold) I will feel that the inclusion gave members an option they felt met their needs.

What the question boils down to, IMO, is what do voters intend when they vote for RON? Maybe we could ask members what it means to them (not in terms of the algorithm that will handle it, but in terms of the message they want to give out).

I don't know if I'm able to get too worked up over a difference where you get the same result faster, but have possibly redefined terminology. If one system is redefining that reaching threshold isn't quite winning, the other system is redefining what a 'loser' is when eliminating a candidate randomly even though they have no fewer votes than the other candidates. Perhaps the disappointment for a candidate is less if they are eliminated before reaching threshold rather than after - but that's psychology

Indeed if that was the only difference between different quotas, then it wouldn't be a big issue. But different quotas can lead to different results.

I think the vote system is complicated enough that we don't want it any more complicated

By using STV, we have accepted a certain amount of complexity. The way I like to think of it is that the idea behind STV is quite simple, but the implementation can be complicated.

If there is ever a tie between real candidates, I believe some other mechanism (trial by combat) would be better.

Flamethrowers at ten paces!

[azrael]

Indeed. Obviously this isn't a problem in single-winner elections, because they aren't proportional anyway. There is only a problem if some people are elected (but not all the places), then the rest of the places filled in a separate election. So if it's a 12 seat election, and 11 members elected and one RON, followed by a 1 seat election, then that will be less proportional. Of course we could, as a Party, decide that proportionality isn't all that important, in which case it's not an issue.

There have been mention (discussion might be too strong a word) of not replacing Governors in dribs and drabs, allowing the Board to get smaller and smaller until a certain point, at which point replacing all the vacant seats in one go or even dissolving the entire Board and replacing the full 12 seats. Of course the downside is that a diminished Board find it harder to get things done (I am *so* looking forward to us being back to full strength!!!!).

[M2Ys4U]

Well hopefully we won't have so many people leaving the Board so quickly in future...

[scuzzmonkey]

Well hopefully we won't have so many people leaving the Board so quickly in future...

^this.

[azrael]

I'm stocking up on super-glue.

[cabalamat]

Indeed. Obviously this isn't a problem in single-winner elections, because they aren't proportional anyway. There is only a problem if some people are elected (but not all the places), then the rest of the places filled in a separate election. So if it's a 12 seat election, and 11 members elected and one RON, followed by a 1 seat election, then that will be less proportional. Of course we could, as a Party, decide that proportionality isn't all that important, in which case it's not an issue.

There have been mention (discussion might be too strong a word) of not replacing Governors in dribs and drabs, allowing the Board to get smaller and smaller until a certain point, at which point replacing all the vacant seats in one go or even dissolving the entire Board and replacing the full 12 seats.

That would be one way to do it. The question is at what point does the full election get triggered? If after a few seats are vacant, then full re-elections might be quite frequent. OTOH, if after there are lots of vacancies, then the Board will potentially be undermanned for a long time.

[scuzzmonkey]

...and hence the current situation.

wrt how the process currently works, if any position on the board is empty then any member (as per sections 4, 10 and 11 of the Const.) is able to nominate themselves regardless of if the Board has announced that a nominations period has been opened. Exactly how we would continue from here hasn't been defined anywhere afaik, but I would assume that the declaration would intent would trigger a formal nominations period, that would work like a normal Board-called election.

Personally, I'm much more in favour of a top-up system (so to speak) for the Board, purposefully ran to keep us as close to full strength as possible rather than waiting until we hit some arbitrary level.