View source for Talk:Center squeeze

This is the discussion page (the "Talk:" page) for the page named "[[{{BASEPAGENAME}}]]".  Please use this page to discuss the topic described in the corresponding page in the main namespace (i.e. the "[[{{BASEPAGENAME}}]]" page here on electowiki), or visit [[Help:Talk]] to learn more about talk pages.

== Bucklin, Score, and center squeeze ==
<blockquote>Systems which can do either well or poorly in a center squeeze situation include most graded Bucklin systems and score voting. </blockquote>

Under what circumstances do they not do well? — [[User:Psephomancy|Psephomancy]]&nbsp;([[User talk:Psephomancy|talk]]) 03:44, 6 February 2019 (UTC)

== Questions about center squeeze ==
I think center squeeze is an interesting criterion & it's valuable to compare systems which exhibit it vs don't exhibit it. Some questions:

* Is there an existing criterion that is a superset of exhibiting/preventing center squeeze? What I mean is, is there a criterion that if met by a voting system, center squeeze can never happen?
* If not, is "exhibits center squeeze" worthy of a criterion?
* Lastly, whether or not it's called out as a criterion, what voting systems do & don't exhibit it?

--[[User:Marcosb|Marcosb]] ([[User talk:Marcosb|talk]]) 19:52, 20 July 2021 (UTC)

== "Liberal" and "Conservative" ==

In section [[Center_squeeze#Effect_of_strategy]], I would suggest not using real party/ideology names like "Liberal" or "Conservative", as this will often be misunderstood as advocating in favor of a "Centrist" or moderate winner.  (FairVote conflates these concepts on https://fairvote.org/why-the-condorcet-criterion-is-less-important-than-it-seems/ when they say "Condorcet winners are centrist by nature, regardless of the preferences of the electorate" and "choosing the centrist candidate every time is just falling into the fallacy of the middle ground".)

Center-squeeze is bad because it elects poor representatives, not because it elects non-centrists on some absolute spectrum.  If the voters are all pro-women's suffrage extremists and anti-prohibition extremists, then a good voting system that doesn't suffer from "center-squeeze" would elect a pro-suffrage and anti-prohibition candidate, not someone who is centrist on both issues.  It's an unfortunate name, we should try not to reinforce this misconception. — [[User:Psephomancy|Psephomancy]]&nbsp;([[User talk:Psephomancy|talk]]) 02:43, 14 May 2023 (UTC)

:Feel free to edit this section to your liking (though expect others, like me, to swoop in and change what you wrote, possibly reverting your change).  The section was originally added by [[User:BetterVotingAdvocacy]] in 2020 (see https://electowiki.org/w/index.php?title=Center_squeeze&diff=prev&oldid=8773 ) and could use some copyediting (such as the RanSOM noTE capitalIZATION).  Personally, I suspect that most voters (even outside of the United States) think of the political spectrum as a single-dimensional space with "[[left-wing politics]]" and "[[right-wing politics]]", so it's difficult NOT to talk about electoral issues without describing the two traditional extremes in political discourse, but I'd be happy to see a different version of this section. -- [[User:RobLa|RobLa]] ([[User talk:RobLa|talk]]) 20:15, 14 May 2023 (UTC)

== I need a different example. ==

I have an improved version of IRV that is not vulnerable in the example given.
I would elect B (the Condorcet winner). Can I claim that it is not vulnerable or much less vulnerable?
(Is it still considered IRV? See the link on my user page). [[User:RalphInOttawa|RalphInOttawa]] ([[User talk:RalphInOttawa|talk]]) 16:58, 13 December 2023 (UTC)

: I don't think there's a formalized "center squeeze criterion". But as an indication, you could try to see if your IRV version elects Center in every configuration of the [[Left, Center, Right]] scenario; that is, all possible elections with x < 50%, y < 50%, x + y > 50%, where x, y, z are defined as on that page. [[User:Kristomun|Kristomun]] ([[User talk:Kristomun|talk]]) 14:40, 18 December 2023 (UTC)

: Presumably, Center should also be elected if any of the candidates is cloned in a way that does not introduce a Condorcet cycle among the cloned candidates. This could exclude methods that check a fixed number of candidates to otherwise obtain Condorcet compliance, but it would also be more difficult to test. [[User:Kristomun|Kristomun]] ([[User talk:Kristomun|talk]]) 14:41, 18 December 2023 (UTC)

:: Good thinking. I will try more variations in the vote counts and comment on the results. Also, will have a look at your cloning scenario. Thanks for the feedback. [[User:RalphInOttawa|RalphInOttawa]] ([[User talk:RalphInOttawa|talk]]) 19:10, 19 December 2023 (UTC)

::: I'm interested in a formal definition of a "center squeeze" criterion as well, [[User:RalphInOttawa|RalphInOttawa]]. Some reasonable definitions:
:::: 1. Candidates can fall outside of their region on a Yee diagram. (I think this is the best definition.)
:::: 2. Candidates may have an empty set on a Yee diagram.
:::: 3. Satisfying the traditional definition of candidate-later-no-harm (i.e. later preferences have no impact on the election, so there's no consensus-seeking).
::: --[[User:Closed Limelike Curves|Closed Limelike Curves]] ([[User talk:Closed Limelike Curves|talk]]) 21:20, 18 February 2024 (UTC)

:::: You could formalize the "elects C in every [[Left, Center, Right]] election" condition (e.g. for all positive x, y, z_1, z_2, v so that v = x+y+z_1+z_2, x+y > v/2, x < v/2, y < v/2, the method must elect C for the election "x: L>C>R, y: R>C>L, z_1: C>L>R, z_2: C>R>L"). But I don't think that would be very transparent a criterion.
:::: Ultimately, the problem is that center squeeze is loosely defined as "strong wing candidates mustn't exclude the candidate more voters are closer to". Its importance didn't come from theory, but from examples of methods failing it and producing intuitively wrong results. It's thus much harder to find an elegant theoretical phrasing of just what it means.
:::: LNHarm doesn't need to imply center squeeze, though: see for instance [[MMPO]]. [[User:Kristomun|Kristomun]] ([[User talk:Kristomun|talk]]) 18:24, 21 February 2024 (UTC)