Instant-runoff voting: Difference between revisions
Content added Content deleted
(→Notes) |
(Add center squeeze reference, clarify LNHarm compliance note.) |
||
| Line 172: | Line 172: | ||
==== Failure to pick a good compromise ==== |
==== Failure to pick a good compromise ==== |
||
IRV can ignore a good compromise in favor of a polarized choice that enjoys smaller actual support. |
IRV exhibits [[center squeeze]. That means that IRV can ignore a good compromise in favor of a polarized choice that enjoys smaller actual support. |
||
This failure mode occurs in a 3-choice election where parties A and B are bitterly opposed, and party C is first choice for a minority but tolerable for a large majority. For a real-life example, consider the 17th-century Europe struggle over "government-enforced [[Catholicism]]" versus "government-enforced [[Protestantism]]", with "freedom of private worship" as the compromise C. |
This failure mode occurs in a 3-choice election where parties A and B are bitterly opposed, and party C is first choice for a minority but tolerable for a large majority. For a real-life example, consider the 17th-century Europe struggle over "government-enforced [[Catholicism]]" versus "government-enforced [[Protestantism]]", with "freedom of private worship" as the compromise C. |
||
| Line 238: | Line 238: | ||
== Notes == |
== Notes == |
||
Though IRV is often praised for passing [[later-no-harm]], which is claimed to encourage voters to rank all of their preferences, it doesn't tend to use as much of the information provided by the voters as other ranked methods, such as [[Condorcet methods]]. |
Though IRV is often praised for passing [[later-no-harm]], which is claimed to encourage voters to rank all of their preferences, it doesn't tend to use as much of the information provided by the voters as other ranked methods, such as [[Condorcet methods]]. This is a less extreme analog to how [[First Past the Post electoral system|first past the post]] technically passes [[later-no-harm]] by ignoring later preferences altogether. So IRV's [[later-no-harm]] compliance has to be evaluated in context of the other criteria it fails due to using less information than other methods. |
||
IRV always elects a Condorcet winner who receives over [[Dominant mutual third|1/3rd]] of 1st choice votes. More generally, a candidate who at any point when they are uneliminated receives over 1/3rd of all active votes and [[Pairwise counting|pairwise beats]] (is preferred by more voters than) all other uneliminated candidates is guaranteed to win. This is because when all but two candidates are eliminated, the one preferred by more voters is guaranteed to win in IRV, and a candidate with over 1/3rd of active votes is guaranteed to be one of the final two remaining candidates, because at most only one other candidate can get more active votes than the over-1/3rd pairwise victor. |
IRV always elects a Condorcet winner who receives over [[Dominant mutual third|1/3rd]] of 1st choice votes. More generally, a candidate who at any point when they are uneliminated receives over 1/3rd of all active votes and [[Pairwise counting|pairwise beats]] (is preferred by more voters than) all other uneliminated candidates is guaranteed to win. This is because when all but two candidates are eliminated, the one preferred by more voters is guaranteed to win in IRV, and a candidate with over 1/3rd of active votes is guaranteed to be one of the final two remaining candidates, because at most only one other candidate can get more active votes than the over-1/3rd pairwise victor. |
||