Talk:VoteFair Ranking

From electowiki

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

Merge with Kemeny-Young Maximum Likelihood Method

Is this the same method as Kemeny-Young Maximum Likelihood Method. -- RobLa (talk) 08:14, 16 August 2020 (UTC)

No! They are NOT the same.
What IS needed is to edit the page Kemeny-Young Maximum Likelihood Method to point out that it has two other names, namely Condorcet-Kemeny and VoteFair popularity ranking.
In other words, the word "popularity" is a very important qualifier. It distinguishes between VoteFair ranking and VoteFair popularity ranking.
For reference, the Wikipedia page named Kemeny-Young method has the correct info. However that title was chosen by Markus Shulze who dislikes the Kemeny method because it so closely competes with "his" method, so he made a point not to include the word Condorcet, and instead included the less-important name "Young". VoteFair (talk) 20:16, 17 August 2020 (UTC)
What I'm proposing is that instead of two articles (Kemeny-Young Maximum Likelihood Method and VoteFair Ranking) we have one article with the important information from both of the current articles. Based on my limited understanding (mostly from our limited discussions elsewhere), the merged article should probably be called "Condorcet-Kemeny", but I'm inclined to get other opinions on the topic. The merged article should explain the variations (which can include "VoteFair popularity ranking"). I'm not in a huge rush to make the change, but I am eager to make electowiki into a more useful reference site. -- RobLa (talk) 20:45, 17 August 2020 (UTC)
FWIW, this is what I've gathered from investigation, EM posts etc.
- VoteFair ranking methods are a set of methods, each for a different purpose (e.g. single-winner, PR, aiding negotiations).
- The VoteFair popularity ranking is mathematically defined in such a way as to always agree with Kemeny-Young when the latter is unambiguous (no ties). The only difference is, IIRC, the VoteFair popularity ranking maximizes the sum of pairwise magnitudes agreeing with the final ranking, and Kemeny-Young minimizes the sum of pairwise magnitudes disagreeing with the final ranking.
- The VoteFair reference implementation does not implement the popularity ranking: it takes shortcuts that makes the result diverge from optimum in certain cases with very large Smith sets. In exchange, the implementation is always polytime, whereas calculating the Kemeny winner is NP-hard.
As for the name of the method itself, Kemeny-Young seems okay to me. It attributes credit to both Kemeny and Young, and distinguishes the method from the "other" Young method (where the winner is the candidate who becomes the CW after deleting the fewest ballots). It doesn't include the name "Condorcet", true, but neither does, say, River or Ranked Pairs. Kristomun (talk) 07:54, 18 August 2020 (UTC)
Thank you for this straightforward answer, Kristomun. There was a conversation that I had elsewhere not too long ago where we discussed the possibility of moving the convention for Condorcet-winner-compliant methods away from naming them after the contemporary "inventor" to toward moving them to "Condorcet-_____" (where "_____" is the contemporary inventor). That would make it clearer for readers new to this subject that the differences between the variety of Condorcet-winner-compliant methods are largely negligible in real-world elections. However, in my opinion, it seems more important to stay aligned with the English Wikipedia naming, so it seems we can rename Kemeny-Young Maximum Likelihood Method to Kemeny–Young method. -- RobLa (talk) 19:38, 20 August 2020 (UTC)

Thank you RobLa for the name change. What remains is to remove the banner that suggests the idea of combining the Kemeny-Young method article with the VoteFair Ranking article. VoteFair ranking does use the Kemeny-Young method as one PART of VoteFair ranking, but another pairwise method could be used instead and it would still closely match the VoteFair ranking system. Thanks! VoteFair (talk) 15:28, 23 August 2020 (UTC)

I've removed the banner. This article still seems to read like an advertisement for VoteFair voting rather than a helpful explanation of how VoteFair compares to other methods. -- RobLa (talk) 02:00, 24 August 2020 (UTC)

Further elaboration on proposed merge

I agree that the Kemeny-Young Maximum Likelihood Method should be renamed to Kemeny–Young method. This would match Wikipedia, and match what the method is more often called. Of course there would be redirects from Kemeny-Young Maximum Likelihood Method and Condorcet-Kemeny method, and they would point to the Kemeny-Young article.

There is nothing to merge. The separate articles must remain separate because they cover separate topics.

Kristomun is correct. Specifically the "overlap" is that I independently created the VoteFair popularity ranking method and later learned that it is mathematically equivalent to the Kemeny-Young method. That naming overlap can be added to the Kemeny-Young article, ideally the same way that it's done in Wikipedia, namely in a "history" section, which also clarifies the subtle difference that what I came up with counts support and looks for the biggest score while Kemeny counts opposition and looks for the smallest score.

What might be confusing is that VoteFair Ranking refers to a system that uses different calculation methods for different purposes. Specifically VoteFair representation ranking is analogous to a two-seat version of STV, and VoteFair party ranking is for ranking the popularity of political parties, etc. Those are in separate articles because they can be used separately, with a different underlying popularity-ranking method -- such as using ranked pairs instead of Kemeny-Young for the underlying counting/ranking.

Expressed another way, adopting VoteFair Ranking would solve a nation's many issues regarding gerrymandering, vote splitting, strategic nomination, blocking reform-minded candidates during primary elections (which is why Biden won the recent primary instead of Warren, Sanders, or Harris who may have been more popular if pairwise vote counting had been used), electoral college, accomodating third-party candidates, etc. This contrasts with what is currently popular which is to promote a single method -- such as STAR, IRV, Approval -- without looking ahead to further additional needed changes.

If this isn't clear, please ask for clarification. Thank you both for your help! VoteFair (talk) 19:30, 18 August 2020 (UTC)