|Luis Villa||Feb 16, 2008 7:53 am|
|Dave Neary||Feb 16, 2008 8:19 am|
|Jonathon Jongsma||Feb 16, 2008 8:35 am|
|Quim Gil||Feb 16, 2008 11:20 am|
|Shaun McCance||Feb 16, 2008 3:47 pm|
|Luis Villa||Feb 17, 2008 5:32 am|
|Quim Gil||Feb 17, 2008 11:36 am|
|James Henstridge||Feb 23, 2008 5:32 pm|
|Elijah Newren||Feb 23, 2008 8:10 pm|
|Telsa Gwynne||Feb 24, 2008 12:41 am|
|Shaun McCance||Feb 24, 2008 9:44 am|
|James Henstridge||Feb 24, 2008 7:14 pm|
|James Henstridge||Feb 25, 2008 5:08 am|
|Behdad Esfahbod||Feb 25, 2008 10:47 am|
|Vincent Untz||Feb 27, 2008 3:44 am|
|Bruno Boaventura||Feb 27, 2008 2:23 pm|
|Subject:||Re: time to (re)consider preferential voting?|
|From:||James Henstridge (jam...@jamesh.id.au)|
|Date:||Feb 23, 2008 5:32:39 pm|
On 17/02/2008, Shaun McCance <sha...@gnome.org> wrote:
On Sat, 2008-02-16 at 10:53 -0500, Luis Villa wrote:
[Speaking purely as a Foundation member and not as a member of the Board; I've not discussed this with the Board at all.]
Some years ago the Foundation considered the use of preferential voting to select the board. At the time I opposed it, for reasons I don't fully recall but which in retrospect probably boiled down to 'I'm unfamiliar with it.' I believe that at the time we'd also have had to write the software, which would not have been fun. But I've come around to believing that this is a better way to run elections.
It appears that by the time of our next election, we'll have a third-party, free software solution available for the problem, used recently and successfully by FreeCulture.org. http://blog.selectricity.org/?p=4
I'm still trying to puzzle through the bylaws (which are a bit of a mess wrt voting) as to what it would take to actually enact this change (bottom line is probably that the board can just say 'it should be this way'), but in the meantime I thought it might be good to have a bit of discussion here around whether or not this is a good idea.
Maybe I'm the only one, but I don't really see the point. For the record, I strongly advocate preferential voting in situations where you are electing exactly one person. In these cases, non-preferential voting systems tend to lock out candidates.
I think the same arguments about not locking out candidates stand when you generalise single seat instant run-off to multiple seat single transferable vote: if the candidate you prefer is unpopular and gets eliminated at the start, this does not penalise you for choosing them.
In instant run-off, you are effectively eliminating candidates until one has more than 50% of the first preference vote. For seven seat STV, you do the same but pick candidates with more than 12.5% of the first preference vote (since 8 candidates could not satisfy this criteria).
Where STV deviates from instant run-off is in what happens when a candidate is elected. With STV, when a candidate is selected, all the votes for that candidate get redistributed at a reduced weighting based on how far over the quota they were. So if 25% of voters gave their first preference to Luis, then he'd be elected, and all the votes would be redistributed a weighting of 50%.
This gives you another benefit over our current system: you aren't penalised for picking a popular candidate.
For the board elections, we are electing seven people, and we each get to cast up to seven votes. I don't think we've ever seen the list of candidates unfairly cut due to non-preferential voting. And I'm sure I've never made a strategic vote for one person instead of another I like more, simply to block another person.
You may never have made a strategic vote with our current system, but it is definitely possible.
As an example, in the last election there were only 15 votes separating the board member who got the least votes, and the next candidate. In contrast Luis received 80 votes more than what he needed to be elected.
If 16 people who wanted Luis on the board, instead voted for George Kraft they'd have changed the outcome of the election while still getting Luis elected. In effect, votes for marginal candidates have a higher value than those for safe candidates. Of course, if everyone adopted this practice then the safe candidates would not be very safe -- one of the problems with systems that encourage strategic voting.
Any preferential voting systems is going to make the voting process more difficult. If I had had to order my votes in previous elections, I'm sure it would have been mostly arbitrary. If it's not solving any real problems, why bother?
Is it really that much more difficult to order a list of ten candidates as opposed to selecting 7 out of the 10?
Even if you aren't sure of a total ordering, you can probably pick a few candidates that you definitely want elected (put them at the top) and some candidates you definitely don't want elected (put them at the bottom). You might decide to order the remainder randomly if you don't care about them.