Talk:Smith criterion: Difference between revisions
Content added Content deleted
(Created page with "Do all Smith-efficient methods produce Smith set rankings of the candidates for their orders of finish? ~~~~") |
No edit summary |
||
Line 1: | Line 1: | ||
Do all Smith-efficient methods produce Smith set rankings of the candidates for their orders of finish? [[User:BetterVotingAdvocacy|BetterVotingAdvocacy]] ([[User talk:BetterVotingAdvocacy|talk]]) 16:35, 22 February 2020 (UTC) |
Do all Smith-efficient methods produce Smith set rankings of the candidates for their orders of finish? [[User:BetterVotingAdvocacy|BetterVotingAdvocacy]] ([[User talk:BetterVotingAdvocacy|talk]]) 16:35, 22 February 2020 (UTC) |
||
: That's really two questions. The first is "is it possible to create a Smith-efficient method that doesn't rank them in this way?", and the second is "do all publicly known/nonpathological methods rank them in this way?". The answer to the first is yes (e.g. Smith//IRV and then rank the rest of the candidates randomly below the candidates in the Smith set). The former is, I think, "most but not all". The intuitive reason is that if you make a method that embodies Smith-compatible logic, then the path of least resistance for most such simple types of logic is to extend to a full Smith ranking. Schulze is like this, for instance. But composed methods (Smith,Minmax) and methods that glue together sufficiently different logic (BTR-STV, Benham) are not necessarily so. [[User:Kristomun|Kristomun]] ([[User talk:Kristomun|talk]]) 19:39, 22 February 2020 (UTC) |
Revision as of 19:39, 22 February 2020
Do all Smith-efficient methods produce Smith set rankings of the candidates for their orders of finish? BetterVotingAdvocacy (talk) 16:35, 22 February 2020 (UTC)
- That's really two questions. The first is "is it possible to create a Smith-efficient method that doesn't rank them in this way?", and the second is "do all publicly known/nonpathological methods rank them in this way?". The answer to the first is yes (e.g. Smith//IRV and then rank the rest of the candidates randomly below the candidates in the Smith set). The former is, I think, "most but not all". The intuitive reason is that if you make a method that embodies Smith-compatible logic, then the path of least resistance for most such simple types of logic is to extend to a full Smith ranking. Schulze is like this, for instance. But composed methods (Smith,Minmax) and methods that glue together sufficiently different logic (BTR-STV, Benham) are not necessarily so. Kristomun (talk) 19:39, 22 February 2020 (UTC)