[ home ] [ pony / townhall / rp / canterlot / rules ] [ arch ]

/townhall/ - Townhall

A place for civilized animals
Name
Email
Subject
Comment
File
Flags  
Embed
Password (For file deletion.)

[Return][Go to bottom]

 No.6118

Let me ask you a seemingly simple question: In an election with more than two candidates, if one candidate always wins against any other candidate in a head to head battle, should they be considered the winner?

This is known as a Condorcet winner, and it has caused a major paradoxical issue in today's voting systems. Largely because a Condorcet winner does not always exist. Anyone who has ever played a three-way game of Rock Paper Scissors should recognize the possibility of a no-win condition. But many people instinctively believe that any voting system should always choose the Condorcet winner if one does exist.


I was shown a wonderful article recently showing the flaws in different voting systems, and how ranked voting can have wildly different results depending on what system you use to count the tallies.
http://www.ams.org/publicoutreach/feature-column/fcarc-voting-decision
Additionally, all of these ranked voting systems introduce a measure of strategic voting, which always pushes voting toward a two party system where voters vote against the candidate they hate, instead of for the candidate they want.

Cardinal voting is often proposed as a solution to this problem, because it preserves independence of irrelevant factors, which many ranked systems do not. there is no penalty for voting up your favorite candidate, so this feels like a more fair system for finding a popular candidate.

Interestingly, none of the four most popular ranked voting methods today choose the Condorcet winner, or the winner who would win against any other candidate in a head to head battle, when using the proposed sample ballots in the linked article. Nor does cardinal voting under most conditions. and even more interesting, not all ballot conditions produce a Condorcet winner

This proposes an interesting philosophical dilemma. If crafting a voting system that always chooses the Condorcet winner if one exists is impossible, how important is it to always choose the Condorcet winner? Should this be our primary concern, or should we focus more on the most popular winner overall? This is a largely undecided philosophical dilemma, so all opinions are welcome in addition to facts and figures.

 No.6119

File: 1596083607232.jpg (202.85 KB, 1280x539, 1280:539, nonstick-pan.jpg) ImgOps Exif Google

>>6118
> If crafting a voting system that always chooses the Condorcet winner if one exists is impossible
Don't such systems already exist though?  E.g.:
https://en.wikipedia.org/wiki/Schulze_method
https://en.wikipedia.org/wiki/Ranked_pairs
Seems like the only issue is that they can be rather complicated to implement.  Copeland's method is simpler but often produces a tie:
https://en.wikipedia.org/wiki/Copeland%27s_method
However, the tie could be resolved by other means.  So I'd say that if we do ranked-choice elections, we should use a Condorcet method, because it has nice, easily understood properties if there is a Condorcet winner.  If we don't use a Condorcet method due to its complexity, we might as well just use the much simpler Approval Voting method.  Non-Condorcet ranked-choice systems like Instant Runoff Voting have some rather strange pathologies, e.g., https://rangevoting.org/IrvPathologySurvey.html

>which always pushes voting toward a two party system
Probably not nearly as much as the current first-past-the-post system, in which third parties get votes only from people who really hate both parties approximately equally.

>most popular winner overall
How would you define "most popular" in a ranked-choice system?

 No.6120

>>6119
Ooh, I like that ranked pairs method. Very easy to understand, though it would be cumbersome to implement.

>How would you define "most popular" in a ranked-choice system?
Well the simplest method would be a mean value rank, which I'm not sure the name of. effectively the inverse of the Borda Method
>assign a point value equivalent to the rank voted: 1 point for 1st, 2 points for 2nd, etc
>tally up the points for each candidate
>divide each candidate's score by the number of voters who ranked that candidate
This system is very simple and helps show the intuitive most popular choice, however, it fails the Condorcet criterion.

Ultimately I think any of the systems we've just named would be preferable to the US's current system, as they allow a 3rd party to have a fair chance.

 No.6121

I don't think there should be any weight given to a "Concordet Winner".  It's just more of what we've got now, something everyone's unhappy with.  Like it's true that Range Voting wouldn't inherently support a Concordet Winner, but it's weird to present that as any kind of counter argument.  That's the type of winner we wanted to avoid in the first place.

 No.6122

File: 1596111541966.jpg (50.21 KB, 720x720, 1:1, 1476853410092.jpg) ImgOps Exif Google

>>6121
>It's just more of what we've got now, something everyone's unhappy with.
Huh?  The first-past-the-post system (which we use now for most elections) can elect a Condorcet loser, someone who would lose in a head-to-head election against every other candidate.

 No.6124

>>6121
Actually the current system almost never elects the Condorcet winner, because it mathematically forces two party voting.

 No.6126

>>6122
>>6124

If there's a two party voting system, then every winner is the Condorcet winner.  That's what a Condorcet method checks for in the first place as far as I can tell.  In every potential two party system you could pull from ballot results, who would win?  It is technically better than what we have now (what isn't?), but by no means something we must respect.  Is there even necessarily always a Condorcet winner to promote?

And most importantly, compared to Range Voting as described by the OP video, it largely fails to account for impact.  A minority who would be immensely negatively impacted by a candidate should be able to voice the extent of that impact in some way, beyond simply "we'd prefer someone else".

 No.6131

File: 1596143559970.jpg (102.65 KB, 850x1200, 17:24, 1596060181820.jpg) ImgOps Exif Google

Another good thing about Condorcet methods is that they can dispense with the party primaries.  Primaries sometimes have a tendency of selecting extreme candidates (e.g., Trump).  If voters in the general election had the option of voting for a more moderate and well-behaved Republican over Trump, I think the odds of Trump winning in November would fall a lot.

>>6126
>Is there even necessarily always a Condorcet winner to promote?
No.  That is how the various Condorcet methods differ, in how they pick a winner when no candidate beats all the other candidates head-to-head.

>be able to voice the extent of that impact in some way
That sounds really hard to do in a way that isn't extremely susceptible to tactical voting.  

 No.6133

>>6131
>That sounds really hard to do in a way that isn't extremely susceptible to tactical voting.  

Hmm...perhaps.

 No.6146

>>6126
if only two candidates ever ran, you'd be correct, but in any system more than 2 running candidates, First-past-the-post voting nearly always forces a non Condorcet winner who more than half the population hates, and in some cases it grants victory to the Condorcet loser.

And yes, the minority should be able to have a voice, which is why ranked voting or even range voting would be better than plurality votes.

 No.6336

File: 1597524998218.png (46.96 KB, 323x254, 323:254, 1409709433720.png) ImgOps Google

Come to think of it, the presidential primaries probably would be a good place to adopt a Condorcet method.  It likely would have avoided selecting Trump as the Republican nominee in 2016.  (As I understand it, most Republican primary voters were against Trump, but they split their votes among many other candidates, so Trump got a plurality, even though he wouldn't have won in a head-to-head contest with any moderate candidate.)  And I think the parties, as private organizations, can just make this change themselves, without the need for new laws.  


[]
[Return] [Go to top]
[ home ] [ pony / townhall / rp / canterlot / rules ] [ arch ]