Thesh Voting

For the discussion of math. Duh.

Moderators: gmalivuk, Moderators General, Prelates

User avatar
Xanthir
My HERO!!!
Posts: 5426
Joined: Tue Feb 20, 2007 12:49 am UTC
Location: The Googleplex
Contact:

Thesh Voting

Postby Xanthir » Wed Apr 20, 2016 4:59 am UTC

So we all know that first-past-the-post (the standard voting system in most countries, such as America) is terrible. It forces you into a 2-party system (attempting to run a third party just steals votes from whichever major party you're closest to, increasing the likelihood the opposing major party wins), and is generally almost the worst you could do without being purposely perverse.

Another popular voting system is approval voting, where you vote for as many people as you want (indicating that you approve of them, and wouldn't mind them winning). This eliminates the "vote-stealing" problem, but it introduces a new one - it massively favors "centrist" candidates who are blandly acceptable to the largest number of people. We *want* popular outsiders to be able to shake things up, tho, but they're unelectable under approval voting. This then encourages strategic voting, where you *don't* vote for people you find acceptable, to make it less likely they'll beat your favorite candidate; this devolves into a messy version of first-past-the-post.

Over in FaiD, Thesh proposed a voting system that combines the best of both.

1. You vote twice on the ballot, casting a single "fave" vote for your favorite candidate, and then as many "approval" votes as you want for everyone you find acceptable. (Your fave vote is automatically considered an approval vote, too.)
2. If any candidate gets over 50% of the faves, they win.
3. Otherwise, if any candidate gets over 50% of the approvals, whichever one has the most approval wins.
4. Otherwise, do an instant run-off of the two candidates with the most approval. The winner is either the one with the most faves, or the most approvals, whichever category has a larger *difference* between the two candidates. (In other words, if the two highest-approval candidates are like 45% approval and 10% approval, the 45% approval wins; if they're both 45% approval, whoever gets the most faves wins.)

On first glance, this looks really good. It feels like it minimizes the downsides of both voting types - it allows popular outsiders to win even against the omni-bland candidate that gets 100% approval, but defaults to the least regret (highest approval) if no one can achieve that. And in the bad case that nobody can get a majority of people to like them, you decide between the two that are least offensive, letting strong feelings (faves) win out over bland acceptability unless the approval difference is *large*.

Can anyone spot any obvious problems with this? Can anyone provide a reasonable argument that this'll provoke strategic voting, or has some weird corner cases?

I can see one strange inflection point: Candidate A gets 25% faves and 25% approval (only A's fave supporters approve of them at all), Candidates B-E all get ~19% faves and less than 25% approval, Candidate F gets 0% faves and 49% approval. Nobody has majority faves or majority approval, so instant run-off occurs between A and F, with A winning - the fave difference is 25% while the approval difference is only 24%, so A wins on faves. The runoff was correct - B-E were clearly less favorable candidates than A, and I think it's easy to argue they were worse than F. Deciding between A and F is a toss-up between two bad choices - A is only liked by a quarter of the population, while F is a milquetoast that *nobody* actually likes, but almost half the population can at least *tolerate*. This is a knife's edge - it A had *slightly* less faves, or F had a *tiny bit* more faves or approval, F would have won. It's just a slightly funky situation, but I don't think it's killer.
(defun fibs (n &optional (a 1) (b 1)) (take n (unfold '+ a b)))

User avatar
cyanyoshi
Posts: 418
Joined: Thu Sep 23, 2010 3:30 am UTC

Re: Thesh Voting

Postby cyanyoshi » Wed Apr 20, 2016 7:45 am UTC

What advantages/disadvantages would this system have over regular instant runoff voting? This system seems very similar to a coarser version of IRV where each voter is forced to declare a single 1st preference, an n-way tie for 2nd preference, and nothing else.

Let's see if I understand this correctly. Say there are 4 candidates A through D with the following votes:

A: 48% fave, 48% approve
B: 10% fave, 49% approve
C: 40% fave, 40% approve
D: 2% fave, 49% approve.

Since nobody has over 50% approval, it becomes a contest between candidates B and D, with B ultimately winning. I guess the danger here is that since candidate A seems like a better choice than candidate B, having too many middle-of-the-road candidates could harm overall satisfaction.

lorb
Posts: 405
Joined: Wed Nov 10, 2010 10:34 am UTC
Location: Austria

Re: Thesh Voting

Postby lorb » Wed Apr 20, 2016 2:13 pm UTC

Xanthir wrote:So we all know that first-past-the-post (the standard voting system in most countries, such as America) [...]


That is a very US-centric view. Look at the map at Wikpedia. Most countries have some kind of proportional representation.

edit: the proposed system also provokes strategic voting in the form of giving approval to fringe candidates in order to drive down the approval % of a bland candidate that the voter is afraid could beat his preferred choice in approval but not in faves. This is a consequence of "Thesh Voting" violating the Later-no-help criterion. So that's one point where it is actually worse than regular instant runoff voting.
Please be gracious in judging my english. (I am not a native speaker/writer.)
http://decodedarfur.org/

Derek
Posts: 2181
Joined: Wed Aug 18, 2010 4:15 am UTC

Re: Thesh Voting

Postby Derek » Wed Apr 20, 2016 10:30 pm UTC

Xanthir wrote:Another popular voting system is approval voting, where you vote for as many people as you want (indicating that you approve of them, and wouldn't mind them winning). This eliminates the "vote-stealing" problem, but it introduces a new one - it massively favors "centrist" candidates who are blandly acceptable to the largest number of people. We *want* popular outsiders to be able to shake things up, tho, but they're unelectable under approval voting.

I don't see how this is a problem. A good election system should typically elect centrists (in a single winner election, in multiple winner elections proportionality is desired), those candidates best represent the feeling of their constituency, and this creates a more stable system than alternating between extreme candidates.

This then encourages strategic voting, where you *don't* vote for people you find acceptable, to make it less likely they'll beat your favorite candidate;

The acceptable cutoff can be placed anywhere along the voter's ranking of candidates and it's not considered degenerate strategic voting. It's only a problem if a voter does not approve a candidate that they like more than a candidate that they did approve, but there's basically no reason to ever do this in approval voting.

The most strategic ballot in approval voting goes roughly likes this: If your most preferred candidate is the frontrunner, only vote for him. Otherwise, vote for every candidate you like more than the frontrunner.

this devolves into a messy version of first-past-the-post.

I don't see how?

User avatar
Qaanol
The Cheshirest Catamount
Posts: 3069
Joined: Sat May 09, 2009 11:55 pm UTC

Re: Thesh Voting

Postby Qaanol » Thu Apr 21, 2016 12:15 am UTC

If you rename “fave” to “excellent”, and allow people to mark more than one candidate as excellent, then you get essentially 3-level majority judgment.

That is, you as a voter rate each candidate “excellent”, “acceptable”, or “unacceptable”. If more than half the voters rate some candidate as excellent, then the candidate with the most excellent votes wins. Otherwise, the candidate with the highest number of “acceptable or better” votes is elected.
wee free kings

User avatar
Thesh
Made to Fuck Dinosaurs
Posts: 6598
Joined: Tue Jan 12, 2010 1:55 am UTC
Location: Colorado

Re: Thesh Voting

Postby Thesh » Thu Apr 21, 2016 12:59 am UTC

My goal was to avoid two types of candidates: the "meh" candidate and the fringe candidate, while also keeping the simplicity of plurality/approval voting while allowing for decentralized counting. The meh candidate is a candidate that is only weakly liked by a minority of voters. The fringe candidate is a candidate that is strongly liked by only a small minority of voters. Plurality is susceptible to fringe candidates if there is a lot of vote splitting, while approval is susceptible to the "meh" candidate.
Summum ius, summa iniuria.


Return to “Mathematics”

Who is online

Users browsing this forum: No registered users and 11 guests