Defining the Maemo Community Council voting procedure
 

Due to the uproar caused by the voting system used in the last election, a new system needs to be selected for next year. To this end, I'm committing a new task for the November Sprint to define the new system.

As I don't much care about voting theory myself, I'm asking those people who do so obviously care to move beyond the loud-complaining stage to the constructive-change stag and outline a better voting system on the wiki.

I'll stipulate 3 things first, though:

• Yes, we know plurality methods suck, get over it and define something that doesn't suck. We're after a concrete voting system, so no proselytizing about your favorite system, or ranting about the evils of plurality methods.
• Keep the theory reasonable, and try to keep the content appropriate for people with no background in voting theory. Yes, we're all very impressed by howe much you know, but none of that needs to be paraded around to define a new system.
• This has nothing to do with voter nor candidate eligibility, this is only about the voting system to be used by eligible voters to elect eligible candidates. We'll deal with that more contentious issue somewhere else.

With that in mind, let's see what sort of awesome system you can come up with!

The deadline is November 30th, 2008.

Thanks!

Re: Defining the Maemo Community Council voting procedure
 

My opinions as community member: http://en.wikipedia.org/wiki/Preferential_voting

"Preferential voting (or preference voting) is a type of ballot structure used in several electoral systems in which voters rank a list or group of candidates in order of preference."

All the rest is just theory and implementation details you can skip. :)

In practice, instead of voting one best candidate (like last time) or choosing 5 candidates at the same level (like the GNOME Foundation does), each voter ranks the candidates by preference and maths do the rest.

In practice this option provides better chances to those that are consistently liked by most voters, and less chances to those more controversial being in the top for some and in the very bottom for others.

Example:

a - Zoe is a candidate, a nice person. She is not in people's mind for the top runner and for most is not even in the top 5 but everybody has her at least in the 6th position.

b - Joe is a candidate, a strong character. A minority thinks of having him as primary runner, another minority don't but still would have him as 5th candidate to get in the Council. The majority actually dislike him.

In a preferential system Zoe has more chances of being elected than Joe. In the current system Zoe has no chances and Joe will probably get in.

On a more theoric approach, the preferential system advocates that more "community inteligence" is put into use to determine who are the elected candidates. Another example:

a - Jim is an amazing candidate. Everybody loves her. A huge majority votes for her and she gets 300 votes more than needed to obtain a seat.

b - Tim and Ming are very good forerunners. A lot of people vote for any or both of them. At the end Tim gets the sit, Ming is left out - just for one vote.

How fair is that? The "inteligence" unused in the 300 extra votes from Jim could have been put into better used in a preferential system, deciding with much more ground whether Tim, Ming or both were the most supported for a seat.

Software available to do this? No idea. http://um.com.au/cgi-bin/cassandra/ is linked in the wiki page, and ther are websites offering the service for free (we used one in GNOME to determine the winner of the GUADEC logo contest a few years ago).

Re: Defining the Maemo Community Council voting procedure
 

Sounds very much like the Eurovision Song Contest. Douze points to qgil for this.

Re: Defining the Maemo Community Council voting procedure
 

That's it! The candidates all need to record a song!

But I also agree with the preferential system. I'd like to see something like this implemented in "real world" politics, too, especially in areas where there's a lot of choice. Americans have created a workaround for their political system that involves reducing the choice to two, but that's not a good solution.

Re: Defining the Maemo Community Council voting procedure
 

What, so much time to moan about how much the voting system we picked sucked (I'm recalling pages and pages of text here), but not just a few minutes to step up to fix it now that you're presented with the opportunity?

Pity.

Re: Defining the Maemo Community Council voting procedure
 

Why belabor it, GA?

Quim wins. Instead of debating further, we can all just Thank his post and move on. :D

Oh, and if I'm no longer with Nokia I'm running next year. Beware, beyotches!

Re: Defining the Maemo Community Council voting procedure
 

I'm feeling spiteful. ;)

Well, you could've run this year if you had chosen. ;) Nokia employees, being a part of the Maemo Community, are permitted to run. You just have to disclose your relationship with Nokia.

The only people I might, personally, discourage from running are the maemo.org folks (X-Fade, dneary, andre, karsten, etc) since they're already directly involved in the process and I'd rather bring in some additional faces than piling more hats on a single person. ;)

That said, I'm glad to hear it. The turnout for candidates was ever so slightly underwhelming last time around.

Re: Defining the Maemo Community Council voting procedure
 

You might as well just anoint me now so we can move on to more trivial things. :D :p

Re: Defining the Maemo Community Council voting procedure
 

Qgil's suggestion is good, as long as one is also allowed to state indifference between 2 candidates. In other words to specify the same preference between one candidate. E.g. Ming 1, Tim 1 (then Jim gets 3).

Re: Defining the Maemo Community Council voting procedure
 

Well, perhaps the lengthy thread discussing voting systems last time actually persuaded everyone who cared to the same side...

Anyway, I'm not sure yet; I have been sticking my head back into Wikipedia and other resources, plus just plain thinking. I'm a range voting fan for single-winner contests, but it's not clear to me that the likely outcome of range voting extended to a multi-winner contest is optimal. It will tend to pick centrists a group of similar winners, while it could be argued that we should always have, say, at least one my-N810-is-a-laptop-you-insensitive-clod geek and one what's-this-"root"-thing-anyway mainstream type on the Council; I'm not sure I buy into that notion here, but I'm not sure I don't either.

As that distinguishes between very different types of voting systems (proportional vs. non-proportional), it seems that would be an important discussion to have, so let's have that first instead of (or in addition to) implementation details!

Question for discussion:

Should the goal be:

• Choose several similar candidates that will make all factions somewhat happy and none too angry.
  Effect with enough randomly-distributed candidates tends to all winners identical in all significant issues.
• Choose some candidates to please each faction, proportional to the voters in that faction?
  Effect with enough randomly-distributed candidates tends to a spread of dissimilar winners corresponding roughly to clumps of voter position.

Re: Defining the Maemo Community Council voting procedure
 

Then you should explore the proportional score voting system (also called "reweighted range voting"), or "asset voting", described here:
http://scorevoting.net/RRV.html
http://scorevoting.net/Asset.html

Nope. See these pictures. Warren Smith explains

Re: Defining the Maemo Community Council voting procedure
 

Thanks for those links; I'm pretty sure you gave them in the other thread, but I hadn't got back over and looked them up yet. ;)Ah, correct. That was bad (i.e. wrong!) wording. I meant that straight multi-seat range voting (simply taking the top 5 candidates) would result in a cluster of similar (good, not centrist) candidates; i.e. it's vulnerable to cloning by the single-seat winner. (Which might not be a problem if people are more concerned about candidates' personalities than on positions in some issue space, but needs pointed out.) I'll go fix it now.

Re: Defining the Maemo Community Council voting procedure
 

I could point out that someone with Easy Debian and the new GUI edition of DBus Switchboard installed could (thee-or-et-ic-ally) use their N810 as their laptop without knowing how to gain root... but that would be a combination of shameless plug and qolling.

Re: Defining the Maemo Community Council voting procedure
 

Its basically the same concept as my most helpful poll, but without the rankings.
http://www.internettablettalk.com/fo...ad.php?t=24633

I personally think I prefer it that way, if its 5 people, the its multiple choice pick your favorite 5.

It seems possible that with your way, that a person could receive more direct votes and lose to a lesser candidate who received more secondary votes.

Re: Defining the Maemo Community Council voting procedure
 

Okay, just to make GA happy: I wrote the seminal paper on voting yeeears ago. For your jeering pleasure:

http://www.corpse.org/archives/issue...ews/arnold.htm

Re: Defining the Maemo Community Council voting procedure
 

What's wrong with the plain "vote for 5 or fewer candidates" method (5 as that's the number we want, fewer so people don't just fill in the gaps with people they don't know anything about).

Shouldn't this accurately represent the wishes of the community. I don't really see why we should rank the choices, which has uses when you're electing and wish to also know the positioning of the elected.

Then again, I don't know anything about voting theory so please do shoot me down :)

Re: Defining the Maemo Community Council voting procedure
 

It depends what you want to achieve. It certainly keeps the options for the voter simple. However, keep in mind the number of candidates is a variable.

If it is not possible to define 2 (or more) candidates as same rank it might lead to some voting for all the candidates (after the first preferred in a random order) because they'd rather not see some candidates win. That leads to sloppiness. Being able to specifiy explicitly the indifference or negative sentiment regarding candidates is a good thing IMO.

Re: Defining the Maemo Community Council voting procedure
 

What if I do take the time to research all the candidates? In the last mayoral election here in San Francisco, we had like 10 choices.

If people don't know enough about a candidate, they can just give him a zero - which is what you're proposing we force them to do.

We can even allow an "abstain" option, and look at average scores instead of totals.

I'm not proposing ranking, I'm proposing rating. The reason why is, to pick better winners, as measured via Bayesian regret:
http://scorevoting.net/BayRegDum.html

If you are a sane rational person, then you want the greatest expected happiness with election results. Therefore you want the best voting method possible, so you aren't forced to be any less happy than you could be.

Too late. :D

Re: Defining the Maemo Community Council voting procedure
 

Well, since nobody's discussing, I've unilaterally made up my mind, and now anyone who thinks different shall be obliged to overcome my biases to persuade me. :p

I'm gonna say we want proportional representation, and, much as I really, really, really like asset voting, I think people aren't gonna like it, so I propose reweighted range voting. I'm gonna go write it up on the wiki over the weekend, but here's brokenladder's link again for those who missed it the first time around (you guys get a SNEAK PREVIEW!!):
http://scorevoting.net/RRV.html.

Re: Defining the Maemo Community Council voting procedure
 

Are you going to write it up, or do I need to dive in, figure out what it is and write it up? (I'm fine either way, it is my responsibility to see it finished this month, so whatever needs to be done needs to be done. Just realize that my understanding of voting theory is next to none and anything I write up is likely to be flawed. ;))

Re: Defining the Maemo Community Council voting procedure
 

I do so admire your passive-aggressive approach there, General. ;)

Re: Defining the Maemo Community Council voting procedure
 

I noticed that after I read it when I typed it up this morning, and it definitely wasn't intended. I was simply too lazy/tired to go back and rewrite it. :\

Re: Defining the Maemo Community Council voting procedure
 

...cos he's working so hard on our behalf ;-)

Re: Defining the Maemo Community Council voting procedure
 

Yeah, something like that. . . . ;)

Re: Defining the Maemo Community Council voting procedure
 

that is awesome! either way, i think you guys will be at the forefront of election science, out of all the groups in the world. lewis carroll actually invented asset voting, though i don't know that it was ever used. and rrv (which i like to call reweighted score voting or proportional score voting, since the s.v. community recently switched from "range voting" to "score voting") has certainly never been used in any election of any significance (though i once used it in an inconsequential forum election, similar to this one, as a test).

i suspect there is still some other p.r. method out there that is even better, but i have no idea what it might be.

hope you guys go through with this, and share some of your experiences (and maybe even anonymous ballot data??) with the scorevoting.net folks, so we can further our election method research.

Re: Defining the Maemo Community Council voting procedure
 

Would you be able to recommend some voting software that could either be modified to do what we need or act as a blueprint for new software?

Re: Defining the Maemo Community Council voting procedure
 

Apologies for my lateness and forgetfulness. I've put a proposal up on the wiki, and welcome examination by others. If there are any errors or ambiguities, feel free to correct it or contact me.

Re: Defining the Maemo Community Council voting procedure
 

A huge +1 on preferential voting as Quim prescribes. I do want to point out that there are several ways to count preferential votes, though, and they are not all equal. My favourite is single transferable vote, which I grew up with in Ireland and understand very well. If you ask me, Condorcet is a terrible counting method, because it's so hard to understand, and occasionally produces completely correct but very surprising results.

I do not like range voting (where you give each candidate a score between 0 and 100) - it is complicated for the voter, and complicated to count, giving you the worst of both worlds.

Re: Defining the Maemo Community Council voting procedure
 

What is the status of this?

Re: Defining the Maemo Community Council voting procedure
 

Here's software for counting STV: http://stv.sourceforge.net/

This should be pretty straightforward to use with Maemo, we simply need to dump the voting DB in a format that can be used as input for this - which is straightforward.

Dave.

Re: Defining the Maemo Community Council voting procedure
 

Thanks for the link.

The software supports these platforms: MS Windows, Mac OS X, and Linux. A Linux/ARM(el) or Maemo port would be nice, but I'm sure community council & others have a machine which runs one of those 3 platforms on x86-32 or AMD64/x86-64. So in that regard this software passes the requirement.

Re: Defining the Maemo Community Council voting procedure
 

With various deadlines approaching, it's time to put this to a public vote (i.e. a referendum).

Despite some voices calling for "range voting" (see RRV on the wiki page), some calm heads are calling for a voting mechanism which meets three critieria:

1. Make it easy for people to vote
2. Make the results of the election easily verifiable.
3. Ensure the result well reflects the will of the electorate.

RRV may well be optimal for the third, but the (relatively) complex maths makes it fail on the first two.

"Preferential/preference" voting as described by Quim seems to be - basically - a single transferable vote system (see STV on the wiki page):
Out of any option, this meets all 3 requirements best of all (IMHO), even if it's sub-optimal in one or more categories.

So, for the referendum, I'm imagining there being the following options (language and wording TBD):

1. No change. The current process is fine.
2. A single transferrable vote. Bullets 4 and 5 ("Each community member gets one vote" and "The 5 nominees with the most votes are elected.") will be changed to XXX (TBD, something like "Each community member ranks ranks one or more candidates in order of preference" and "Council members will be selected according to this single-transferrable vote system)
3. None of these options is acceptable.

The Council would decide what to do in the event of the third option getting a majority of votes.

Discussion on this topic should occur on the maemo-community mailing list. Discussion
in this thread and on ITT may be read, but comments on maemo-community will be read. An appropriate thread would be this one - "Election process referendum".

Re: Defining the Maemo Community Council voting procedure
 

I think we should first vote on the voting system used in the referendum for selecting a voting system.

Re: Defining the Maemo Community Council voting procedure
 

I assume you're being facetious/humourous, but - for the record - the referendum is a simple one-member, one-vote, majority wins affair.

Re: Defining the Maemo Community Council voting procedure
 

Given the timescales, it looks like having a week of debate on the voting process and referendum wording is realistic.

Therefore, if you want to have a say on this (even if it's "yes, this is exactly what we want"), please do so as soon as possible: by 1st February 2009, it may be too late.

Re: Defining the Maemo Community Council voting procedure
 

Anyone still care about this? Now would be a great time to read through this thread, and speak up if you've got thoughts on it...

Re: Defining the Maemo Community Council voting procedure
 

Recent discussion on the maemo-community list has explained that, along with the proposed change in voting, the council is also planning on returning to the original rules for eligibility: you have to have 100 karma points to stand, and 25 karma points to vote. Last time, the 25-point limit was dropped because it excluded too many active community members. It is being re-instated because it is believed that karma is now a more accurate reflection of contribution to the community and it is a fair criterion.

I had not realised this was happening until now. Maybe everyone else did realise it but if you did not, and have any concern about it, you had better raise it immediately on the maemo-community list.

Graham

Re: Defining the Maemo Community Council voting procedure
 

It seems that very few people care enough to weigh into the thread (which, in some ways, is a Good Thing).

Other exciting things: people weren't necessarily aware that the next election will re-instate the 25 karma points or higher required for voting rule.

Re: Defining the Maemo Community Council voting procedure
 

After the last election, I made sure that my maemo.org karma was "fixed"*; I want to be able to fully participate in the community. At the time of the first election, I had something like 9 karma... :(


* as in "repaired" (fixed the car) not "tampered with" (fixed the race)