Our new electoral system - tying up loose ends

[Patroklus Murakami]

The election to be held in May 2010 will be the first one to use the Single Transferable Vote system in CDS history.

There is one loose end that needs to be tied up at the next RA meeting. It's an arcane but very important question - what quota to use?

In an STV election, a candidate needs a certain number of votes - the 'quota' - in order to be elected. To give an example, the most widely used quota is the 'Droop quota'. Droop quota is

without the decimal part.

So, if we imagine 60% turnout (based on previous results) that is roughly 80 votes and 13 seats, the Droop quota would be 80/14 + 1 = 6.7 so each candidate will need 6 first preferences to get elected in the first round. Any candidate that gets more than 6 first preferences would be elected in the first round.

See the wikipedia link for a more detailed description of how seats are allocated under STV. In a nutshell, the candidates that got the least votes get eliminated and their votes are transferred to their 2nd, 3rd preferences to see if their votes help to elect someone who did not meet the quota in the first round. Also, if some candidates get more than the required quota, the 'excess' votes are transferred to other candidates using the 2nd, 3rd preferences which may get them above the quota.

I know, it's rather complicated to follow. The main thing to remember is that STV is really easy for voters to understand when casting their ballot - they just put the candidates in order of preference. The difficult part is the bit I've described above and we have software to do that now! (I recall having to do this manually in student union elections - nightmare!)

But the RA needs to take a decision about the quota to use. I've given an example using the Droop quota, others are the Hare quota, the Imperiali quota and the Hagenbach-Bischoff quota. (I'm not making this up!) I propose we use the Droop quota, it's the most widely used. I've drafted a bill to enact this below:

-----------------------------------------------------------------------------------

• STV Quota Bill
  The CDS Constitution specifies that the electoral system will use the Single Transferable Vote method to allocate seats to the Representative Assembly following an election. The CDS will use the 'Droop quota' to determine the minimum number of votes a candidate needs in order to be elected.

  Droop quota is understood to mean: without the decimal part.

[Soro Dagostino]

Never claimed to be as smart as Pat. Just not lacking common sense. What's wrong with just counting the number of votes cast? The process, as I understand it, calls for a census that defines the number of seats. Each voter gets to cast a vote for those the persons they want elected -- up to the number of seats available.

See it all the time in Governance elections . . .

[Claude Desmoulins]

Soro,

The V in these equations is usually understood to equal the number of valid ballots cast. You wouldn't know the actual number until the election is over. The bill would set the formula used to calculate the number.

[Soro Dagostino]

Why weight it with the number of ballots cast? Just count allocated votes. Simple. Easy -- no math other than adding up the votes for each person. But, then, I am a simple minded fellow.

[Patroklus Murakami]

If I have understood Soro correctly, he is proposing that we just count the number of votes cast and then allocate the seats to the candidates who get the most votes. If there are 13 seats and 15 or 16 candidates, then the candidates with the top 13 votes get elected. This approach is certainly simple, but there are some problems with it which STV is designed to solve.

Here are some potential outcomes to illustrate the point where the letters represent candidates and the numbers are the number of votes.

• A-20 B-15 C-14 D-13 E-12 F-11 G-10 H-9 I-8 J-7 K-6 L-5 M-4 N-3 O-2 P-1
  This one is easy, the top 13 candidates A-M get elected. But note that there is fairness question here. Candidate A with 20 votes and Candidate M with only 4 votes both get elected. This is a problem which STV solves which I'll come to later.

  A-20 B-15 C-14 D-13 E-12 F-11 G-10 H-9 I-8 J-4 K-4 L-4 M-4 N-4 O-4 P-4
  This one isn't so easy There are a number of candidates with the same number of votes. Which ones get elected? There is no clear and fair way of determining which 4 of the last 7 candidates should get elected. STV solves this problem by looking at voters second preferences.

  A-20 B-15 C-15 D-10 E-10 F-5 G-5 H-5 I-5 J-0 K-0 L-0 M-0 N-0 O-0 P-0
  This is a variation on the same problem, except this time 7 candidates get no first preferences. Which of them, if any, should get the 4 remaining seats? Again, STV solves this problem by looking at who people put second and then transferring votes to them.

STV is a system which tries to avoid 'wasted votes'. In the scenarios described above, most of the votes for candidates A are wasted votes. If we assume that the quota or 'victory threshold' (as described in my first post) is 6 then the 'excess' 14 votes would be wasted because they don't help to get the candidate elected. Under STV these votes are not wasted. Instead, we look at the second preferences of all 20 voters and transfer the excess to other candidates. This can then help to get other candidates elected who did not get sufficient first preferences in the first round.

I recommend the following short flash animation for illustrating how STV (they call it 'choice voting') works in practice.

STV allocation of seats is complicated to follow the first couple of times you see it but it is a fair system which gives each vote equal power to influence the result and ensures that few votes are wasted. It is *very* easy for voters who just rank the candidates in the order they prefer and their preferences are respected in the results.

[Patroklus Murakami]

In my first post I proposed using the 'Droop quota' to set the victory threshold for gaining a seat. Assuming 80 votes (60% turnout) and 13 seats, this means the quota or victory threshold would be 6 votes to gain a seat. Here are the other thresholds provided by the other methods of calculation for comparison (assuming the same number of votes and seats):

• Droop quota - 6 votes
  Hare quota - 6.15 votes
  Imperiali quota - 5.33 votes
  Hagenbach-Bischoff quota - 5.71 votes

We will need to make a choice at today's RA meeting so that seats can be allocated after the election.

[Delia Lake]

Our new electoral system is not on today's Agenda for today's (9 May) http://forums.slcds.info/viewtopic.php?f=2&t=2853 nor was it on last meeting's agenda.

Jon Seattle has now written the software for the current election and will be testing it this evening with volunteers. "The new election system is getting close to being ready. My plan is to do several rounds of testing, starting tomorrow evening (Sunday). If you are willing, it will take just a few minutes for each round." http://forums.slcds.info/viewtopic.php?f=1&t=2854 Although I don't know for absolute certain, I believe that Jon has used the Droop quota in this new program. We will know this evening when Jon activates the program for our testing.

Now would seem to be a little late for the RA discuss election methodology for the current election as the polls open next Saturday. Claude Desmoulins posted two questions on the Forum regarding the methodology for this election under

Election Mechanics

on 29 March. http://forums.slcds.info/viewtopic.php? ... 476#p14476. To make an informed vote on a quota methodology would require that the RA members take the time not only to read your post Pat but to do some internet searches as well. Since this topic is not on the RA Agenda, people may not have prepared themselves to vote on this issue. Will they be given time during the RA Meeting to inform themselves? And if the RA were to approve a methodology other than the one Jon has already programmed, who will revise the election program and test it publicly this week in order that the election software be ready for the 15 May opening of the polls?

Under the circumstances I recommend the RA vote to approve for this current election the methodology Jon has already included in the program. If the RA would prefer a different quota methodology, make that change apply to future elections but not to this one.

[Patroklus Murakami]

I agree Delia

If the RA needs time to consider the implications of which option to use then we can simply use whatever has been programmed in for now and revisit the matter for future elections.

My aim, in providing info on other options apart from 'droop quota' was to enable RA members to make an informed choice. Given the problems with any other option apart from droop, it's pretty obvious to me which one we should opt for. The ideal outcome would be for the RA to make an informed choice on this for the avoidance of doubt but, if RA members don't feel ready to take such a decision, it can be avoided.

I have asked Arria to put this on the agenda for today's meeting.