Page 4 of 7 « First 23 4 56 Last »
Reply
Thread Tools
brontide's Avatar
brontide is offline brontide
2008-09-05 , 23:01
Posts: 868 | Thanked: 474 times | Joined on Oct 2007 @ Capital District, NY, USA
#31
Originally Posted by allnameswereout View Post
PS: not everyone is allowed to vote.
Like myself.

http://lists.maemo.org/pipermail/mae...er/000761.html

I won't rehash this here, just wanted to point out the discussion on the list.
 
Benson's Avatar
Benson is offline Benson
2008-09-05 , 23:47
Posts: 4,930 | Thanked: 2,272 times | Joined on Oct 2007
#32
Originally Posted by brokenladder View Post
If you are having an at large election (many winners as opposed to just 1) Score Voting (aka Range Voting) is actually still excellent, per massive Bayesian regret calculations.
=> http://rangevoting.org/UniqBest.html
That page doesn't say anything for multi-seat elections, as expected from the title, "Why range voting is "uniquely best" among all common proposals for single winner voting systems". Following the link to Warren Smith's site didn't reveal anything particularly relevant to multi-seat races at a glance. (I did notice that Smith was factually wrong re: the electoral college, but then, I've got a thing about that.) From the abstract, it appears to have been aimed entirely at single-seat races.

If you would like results which more closely model the diversity of opinion among your community, however, you could use a proportional method like Proportional Score Voting (aka Reweighted Range Voting), or Asset Voting. Both of these systems are far superior to and simpler than methods like STV, MMP, etc.

http://rangevoting.org/RRVj.html
http://rangevoting.org/Asset.html

Asset Voting is really fascinating in how simple it is. It was originally invented by Lewis Carroll, who was an Oxford mathematician.
Actually, I believe it was invented by Charles Dodgson, who was not only an Oxford mathematician, but also Lewis Carroll.

I'd never heard of Asset before, and it's very interesting and effective. It resembles old-style party conventions (back when multiple candidates surviving to the convention, and deals being cut to select a nominee, was normal); of course, with the reaction to the suggestion of a ''brokered convention'', it's hard to see it having a chance for application to primaries. (Of course, Pres and Veep nominations aren't a simple multi-seat, so maybe it's not the most suitable.)

RRV seems to effectively combine the benefits I was seeing for STV and cumulative. It's a clean system, but I really like Asset, now that I've seen it.

Originally Posted by brokenladder View Post
Well, the sum of all advantages and disadvantages is a scaler utility efficiency value, that you can compare. The best voting method is the one with the lowest Bayesian regret.

It's analogous to arguing about race cars. You may talk about the superior horsepower of car X, and I may talk about the superior aerodynamics of car Y, and some other guy may talk about the lightness of car Z. But talking about these advantages and disadvantages doesn't address the issue. What we should instead do, if we are rational economists, is put the cars through a zillion random trials, with random drivers and road conditions (modeling real life frequencies) and see which car gets the best average time.

That's what Bayesian regret does for voting methods.
Yes, but there's two issues which keep that from being an unequivocal, objective result:

First, a difference is that we can't measure utility, so we're stuck with models; I'm an engineer and live on models, so I'm not saying that's wrong, but I'm loath to raise it as an argument that I've got an objective solution. Especially when there's no possible way to validate the model by measurements. (Like I said, I'm an engineer.)

And there's always the meta-modeling question; to borrow your racecar analogy, different people will establish different figures-of-merit; some will say time for one lap starting at speed, others will want a 1/4 mile speed, and still others 1/4 ET. Is total utility the right measure? Would some sort of average or median utility be a better measure? While I favor Bayesian regret, I can't claim it's objectively better than all other possible figures of merit.

You could argue that complexity of the voting method is also an issue, but it just so happens that Score Voting (aka Range Voting) is the second simplest alternative voting method, right after Approval Voting (which itself is actually just a limited form of Score Voting, with a 0-1 "score").
True, but as I said, I see complexity only as a political obstacle, and perhaps a hurdle to good analysis. We've got decent analysis already, though, for all election methods under consideration.

I would like to point out that the Gibbard-Satterthwaite theorem only applies to rank-order method, and does not apply to Score Voting or Approval Voting.
http://rangevoting.org/GibbSat.html

Not to say that Score Voting is immune from tactical voting of course. (Just reacts mildly to it.)
Agreed; I was linking on Gibbard-Satterthwaite for those who might instinctively expect that tactical voting is an artifact unique to plurality. While it's not proven for cardinal systems, it is my best guess that (sane) cardinal systems are all subject to tactics and strategy, so trying to eliminate it is pointless. (And a system that reacts "mildly" to it is exactly what I want!)

An excellent point. It's nice that Score Voting has this "pleasant surprise" theory, that if all voters give a maximum score to every candidate they like better than their expected outcome, the average voter will be happier than the expected outcome on which he based that vote.
http://rangevoting.org/PleasantSurprise.html
That's very interesting; while my support for range voting was already fairly solid, that's excellent ammo.
__________________
World's first inductively-charged N900!
 
brokenladder is offline brokenladder
2008-09-06 , 03:32
Posts: 11 | Thanked: 4 times | Joined on Sep 2008
#33
Originally Posted by Benson View Post
That page doesn't say anything for multi-seat elections, as expected from the title, "Why range voting is "uniquely best" among all common proposals for single winner voting systems".
The Bayesian regret figures show that Score Voting elects candidates who are very satisfying to the average voter. This is still more or less true when it's used to elect multiple winners, with the caveat that you don't get proportionality. Although the Fedora steering committee uses this system precisely because they don't want proportionality. They presume that having a bunch of candidates who are relatively average is more expedient and leads to less infighting than having a group that, while it may average out to the same average-ness, is composed of semi-ideologues. That's basically true, since countries which use P.R. have historically demonstrated more quarreling between factions. There are clearly pros and cons to each system. My point is that a group of good representatives should reasonably be a good representative as a whole - though I acknowledge that this is a simplification.

Following the link to Warren Smith's site didn't reveal anything particularly relevant to multi-seat races at a glance.
http://rangevoting.org/PropRep.html

(I did notice that Smith was factually wrong re: the electoral college, but then, I've got a thing about that.)
Please describe what you believe he's wrong about. He's not wrong very often, and when he is he likes to correct the site immediately.

Yes, but there's two issues which keep that from being an unequivocal, objective result:

First, a difference is that we can't measure utility, so we're stuck with models; I'm an engineer and live on models, so I'm not saying that's wrong, but I'm loath to raise it as an argument that I've got an objective solution. Especially when there's no possible way to validate the model by measurements. (Like I said, I'm an engineer.)
Yes, models are imperfect. But Smith used 720 different combinations of the 5 fundamental parameters in his simulations, and then got the average utility efficiency for hundreds of thousands of elections in each of those combinations. Score Voting won in all of them. And by a pretty good measure. It his hard to conceive of any obvious flaws in his simulations, whose correction could possibly make up the difference. Therefore I see those figures as being extremely reliable.

And there's always the meta-modeling question; to borrow your racecar analogy, different people will establish different figures-of-merit; some will say time for one lap starting at speed, others will want a 1/4 mile speed, and still others 1/4 ET.
Sure. Some people will think your simulation should model a society of 10% strategic voters and the rest expressive. Other people will believe that at least 60% of voters are strategic. So you just do simulations covering all of those values. If Score Voting wins under some models, but fails in other models, then you have the very tricky task of figuring out which of those models were closest to reality.

Luckily for Smith, Score Voting won in every model. So the people who want a 1/4 mile, and 1/2 mile, and whatever else, all find that the same race car wins, no matter which set of circumstances they try.

Is total utility the right measure?
Yes. You can more or less prove that.
http://rangevoting.org/UtilFoundns.html

The most compelling "proof" of it (aside from the fact that most every other social utility function can be disproved via reductio ad absurdum) is Harsanyi's observation that a rational voter should want the system that maximizes his expected utility (do not confuse that with expected value, since utility and money are not linearly related).

Would some sort of average or median utility be a better measure?
Not to a rational voter. If a benevolent god wants to give society a voting method which all rational voters will want, he'll give them the one with the lowest Bayesian regret.

While I favor Bayesian regret, I can't claim it's objectively better than all other possible figures of merit.
You can if you allow for some very sensible assumptions. There are some alternative social utility functions that violate sensible notions, such as "maximin". That is, the best state is the one in which the minimum utility for any voter is the highest. That is, if outcome X causes 99% of voters to be extremely happy, and 1% of voters to be unhappy, and outcome Y causes every voter to be totally neutral (no increase in happiness, nor a decrease), then outcome Y is "better" for society.

The problem here is that a single voter is effectively a dictator, able to make all of society much worse off so that he can be a little better off. Now we can wax philosophical about whether that's a sensible way to handle things, but there's no need. We know that this "dictator" voter, were he to have the choice of any social utility function prior to knowing the outcome, would want an additive social utility function - if he was rational.

The fact that the greatest utility sum is what all rational voters want, any arguments about additive social utility are "academic" in my mind.

While it's not proven for cardinal systems, it is my best guess that (sane) cardinal systems are all subject to tactics and strategy, so trying to eliminate it is pointless. (And a system that reacts "mildly" to it is exactly what I want!)
It is proven that all deterministic voting methods that do not employ revealed preference are susceptible to gaming.

Here's a voting method that employs revealed preference:
http://rangevoting.org/CTT.html

That's very interesting; while my support for range voting was already fairly solid, that's excellent ammo.
So let's find a city that will adopt it already, and save the world.
http://rangevoting.org/LivesSaved.html
Last edited by brokenladder; 2008-09-06 at 03:34.
 

The Following User Says Thank You to brokenladder For This Useful Post:
Benson's Avatar
Benson is offline Benson
2008-09-06 , 06:31
Posts: 4,930 | Thanked: 2,272 times | Joined on Oct 2007
#34
Please describe what you believe he's wrong about. He's not wrong very often, and when he is he likes to correct the site immediately.
http://math.temple.edu/~wds/homepage/electoral.txt
3. The electoral vote system does NOT somehow restore the balance of power between big and small states, allowing the small
states to have more say. (That goal IS accomplished by
giving each state 2 senators, regardless of its size.)
That is because the number of electoral votes
a state has, which is the same as the number of seats it
has in the House, is (up to roundoff) proportional to its population. Thus this system simply introduces more noise
and more risk of unjudgeable cliffhanger elections, with no compensating benefit.
The number of electors allocated to a state is not the same as the number of representatives in the House; from the U.S. Constitution, Article II, Section 1:
Each State shall appoint, in such Manner as the Legislature thereof may direct, a Number of Electors, equal to the whole Number of Senators and Representatives to which the State may be entitled in the Congress: but no Senator or Representative, or Person holding an Office of Trust or Profit under the United States, shall be appointed an Elector.
I'm actually receptive to the notion of abolishing the electoral college, because the states as entities (i.e. governments) no longer play a role in the selection of the Senators or Electors; giving the additional representation of the state to the people of that state, regardless of their number, seems useless and unfair. (The discretization effects are ugly, too; if it is to be preserved, doubling or tripling the number of electors and using proportional allocation would be much better.) But arguing that it does not give less populous states more influence (Smith), or that this influence never served a purpose (electionmethods.org, referenced earlier), rubs me the wrong way.
Yes, models are imperfect. But Smith used 720 different combinations of the 5 fundamental parameters in his simulations, and then got the average utility efficiency for hundreds of thousands of elections in each of those combinations. Score Voting won in all of them. And by a pretty good measure. It his hard to conceive of any obvious flaws in his simulations, whose correction could possibly make up the difference. Therefore I see those figures as being extremely reliable.
I agree it seems reliable, but the impossibility of experimental validation leaves me short on dogmatism, in a discussion; that's all.
Sure. Some people will think your simulation should model a society of 10% strategic voters and the rest expressive. Other people will believe that at least 60% of voters are strategic. So you just do simulations covering all of those values. If Score Voting wins under some models, but fails in other models, then you have the very tricky task of figuring out which of those models were closest to reality.
That's not the sort of difference I meant; I'm thinking more of those who would argue that:
  • Certain people "count less" than others.
  • Certain political trends are intrinsically bad (/good) and to be avoided (/mandated), even though they cause no harm (/benefit) to anyone.
  • and a number of similar bases for choosing an inherently harmful result.
Essentially, upon coming up with specifics, I'm realizing I was wrong; these do represent deviations from something that may reasonably be termed democracy -- i.e., Bayesian regret can be claimed as a uniquely democratic figure-of-merit.

(Which leaves the question of whether this "democracy" is the best principle, but there's always an axiom somewhere, and always someone who will dispute it. )

Yes. You can more or less prove that.
http://rangevoting.org/UtilFoundns.html

The most compelling "proof" of it (aside from the fact that most every other social utility function can be disproved via reductio ad absurdum) is Harsanyi's observation that a rational voter should want the system that maximizes his expected utility (do not confuse that with expected value, since utility and money are not linearly related).
Close, indeed; but isn't that only valid as long as the rational voter doesn't know where the result falls, or equivalently, where he falls? (i.e., a non-pre-established or a highly dynamic political environment.) Still, I'll go with your label 'academic'; even if only rational and ignorant voters are guaranteed to prefer it, that's a strong argument for it as the best system if anything close to democratic ideals are accepted.

Here's a voting method that employs revealed preference:
http://rangevoting.org/CTT.html
I like that one, too, though it does scale... interestingly. I'm with fpp; I do learn fascinating stuff every time I get in an election discussion.
__________________
World's first inductively-charged N900!
 
benny1967's Avatar
benny1967 is offline benny1967
2008-09-06 , 06:51
Posts: 3,790 | Thanked: 5,718 times | Joined on Mar 2006 @ Vienna, Austria
#35
It's interesting how many of you are concerned with the voting system.... My only concern is: Who should I vote for? Some candidates I don't know. There's one on the list I definitely do not want to see in the council. The others are like "Yes, I know his name, but who is he?"

What do they stand for? Which social skills do they have? Do they have any strong opinion about which direction Maemo should take in the future? (I know they're not about to decide, but it will have some influence in whatever they do.)

So... I have 2 possibilities: Not to vote at all (just because by accident I could choose a person I wouldnt trust in real life) or vote for the cutest, best looking (as usual). ATM, I think it's better not to vote.

No change of the voting system could help me with that.
 

The Following User Says Thank You to benny1967 For This Useful Post:
GeneralAntilles's Avatar
GeneralAntilles is offline GeneralAntilles
2008-09-06 , 07:23
Posts: 5,478 | Thanked: 5,222 times | Joined on Jan 2006 @ St. Petersburg, FL
#36
Originally Posted by benny1967 View Post
What do they stand for? Which social skills do they have? Do they have any strong opinion about which direction Maemo should take in the future? (I know they're not about to decide, but it will have some influence in whatever they do.)
Pay attention? Most of the discussion is taking place on maemo-community.
 
lardman is offline lardman
2008-09-06 , 08:42
Posts: 2,102 | Thanked: 1,309 times | Joined on Sep 2006
#37
What do they stand for? Which social skills do they have? Do they have any strong opinion about which direction Maemo should take in the future? (I know they're not about to decide, but it will have some influence in whatever they do.)
This is a slight worry about deciding on the electorate. I had assumed that people would know of our "social skills" from our previous conversations on IRC, the mailing lists and here on ITT. In terms of our personal interests, there is some info in the blurb we all wrote for our nominations, and likewise what we all normally talk about on IRC/ITT/ml.

It might be that these should be made more explicit for the next run through.

Even if people were to delivery manifestos, there's no easy way to evaluate the nominee without doing some digging through the ml/ITT/IRC logs to see what they have been talking about and what they have achieved.

The other suggestion that nominees should state what they can bring to the community council sound like a nasty interview question. Will it get any useful responses? Perhaps asking for a CV of previous experience, etc., might be worth while?
 
lardman is offline lardman
2008-09-06 , 09:08
Posts: 2,102 | Thanked: 1,309 times | Joined on Sep 2006
#38
This is a slight worry about deciding on the electorate.
I should qualify this just in case people think I've used the wrong word. I do actually mean deciding who should vote. This is to be a community council, therefore the people who vote should be part of the community - but what makes us part of the community? Not just having a Nokia Internet Tablet, it also requires contribution to the community. If people are contributing, it's expected that they will be aware of others who are contributing, and therefore should have an idea of what the candidates stand for and are interested in.

Obviously there may be some troubles where some community members only contribute to ITT and others only on the ml and never the twain shall meet, etc. Something to be worked out, as ever
 

The Following 5 Users Say Thank You to lardman For This Useful Post:
benny1967's Avatar
benny1967 is offline benny1967
2008-09-06 , 11:56
Posts: 3,790 | Thanked: 5,718 times | Joined on Mar 2006 @ Vienna, Austria
#39
Originally Posted by lardman View Post
I do actually mean deciding who should vote. This is to be a community council, therefore the people who vote should be part of the community - but what makes us part of the community? Not just having a Nokia Internet Tablet, it also requires contribution to the community. If people are contributing, it's expected that they will be aware of others who are contributing, and therefore should have an idea of what the candidates stand for and are interested in.
Originally Posted by GeneralAntilles View Post
Pay attention?
These might be two of the very, very few documented cases in the history of democracy where candidates tell an eligible voter he's not worth voting. (Don't worry, as I said before, I won't.)

Still, to make my point clearer:
I follow ITT, the dev and the community mailing lists, read the planet regularly, ...

But that's not what it's about. I'd feel more comfortable knowing who these people really are rather than what they do. (We're not hiring somebody for a technical helpdesk, nor are we looking for a good developer.)

I wonder how other people make up their minds.
 
allnameswereout's Avatar
allnameswereout is offline allnameswereout
2008-09-06 , 14:33
Posts: 3,397 | Thanked: 1,212 times | Joined on Jul 2008 @ Netherlands
#40
Originally Posted by GeneralAntilles View Post
I sensed sarcasm and hostility in your post, what can I say.
Even if true, IMO a leader shouldn't lower him- or herself to the same level as its friends or foes for one because a leader is an example for its followers. Social skills are very important for a leader, and being sarcastic is not appropriate behaviour of a leader.

This has nothing at all to do with Nokia. It was Jaffa's idea and lardman and myself assisted him a bit with the inception.
A much better response, but it is still related to Nokia because the 5 winners will be the contact persons for Nokia.

Originally Posted by brontide View Post
Like myself.

http://lists.maemo.org/pipermail/mae...er/000761.html

I won't rehash this here, just wanted to point out the discussion on the list.
Me neither (therefore I didn't look much into the process; I was apparently too late already) while hunderds of inactive people with a profile (can be created easily) are eligable to vote.

The problems I see are

* Consensus process didn't work well.
* Eligable base of voters is inaccurate.
* Votes can be sold or bought.
* Election method archaic.
* Unclear about anonimity.
* Goals of the candidates are not outlined well (far too brief for a serious election).
* Relationships between candidates are unclear; who will do what? I haven't found any discussion between the candidates either. I'd like to know who agrees and/or disagrees with whom.
* Elected persons decide on next voting process?
* Minimum number of votes required to pass election is not defined.

Originally Posted by Benson
While it's not proven for cardinal systems, it is my best guess that (sane) cardinal systems are all subject to tactics and strategy, so trying to eliminate it is pointless. (And a system that reacts "mildly" to it is exactly what I want!)
Thanks for your interesting posts.

The underlying methods have different severity regarding strategic voting. Just because a method does have the side effect of strategic voting doesn't say anything about the severity of this side effect.

Quality of representation, or who gets what degree of representation, are not criteria for that definition, or any other I've seen, for democracy. Indeed, your description explicitly rejects certain types of democracy, e.g. direct democracy.
(Also, if you want the winners to effectively represent the whole rather than the majority, I'd expect you to support Borda or similar; it may be said that Borda is to Condorcet as median is to mean.)
Hmm, I never heard of Borda. Care to enlighten me about this person?

You're right regarding the definition but IMO the definition of democracy is flawed. You need to remember it is a a combination of 2 greek words, and that in any ((other) language) it has an artificial meaning already when applied. The current definition of democracy you'd find in a dictionary is based on the current democratic systems whereas in practice the current democracies lead to a chosen dictatorship of max 4 years. Democracy isn't an abstract word anymore. Like liberty.

There is no effective feedback system (ie. "who moderates the moderators?"). In direct democracy this problem is mitigated, but it isn't efficient for fast decision making, and another problem is lack of time and/or involvement. Good book about this is "From the director's democtacy to direct democracy".

Interestingly, in Europe in general one does not vote for their sheriff. Although referendum is implemented in some European countries, it sometimes doesn't work because not enough people vote. The meaning of 'liberal' is also very different in the US compared to Europe.

Consensus, according to my experience, often leads to the person with the biggest mouth becoming the leader while such so-called "leadership" is in many situations not desired.

It's simple enough, to be sure; but while the alleged complexity of other voting systems is an obstacle in changing government elections, I think it's no problem for us.
Agreed, and if more people get in contact with such methods they might see how our current democratic models are out of date. Unfortunately, it seems IRV is a popular alternative, while it is severely flawed. Earlier mentioned site contains some mathematical examples.
__________________
Goosfraba! All text written by allnameswereout is public domain unless stated otherwise. Thank you for sharing your output!
 
Reply
Page 4 of 7 « First 23 4 56 Last »

« Previous Thread | Next Thread »

 
Forum Jump


All times are GMT. The time now is 06:48.

Contact Us - Home - Archive - Top