Talk:Later-no-help criterion: Difference between revisions
(→Incompatible with majority-Condorcet?: new section) |
|||
Line 30:
== Incompatible with majority-Condorcet? ==
Is later-no-help incompatible with majority-Condorcet, or just simple (plain) Condorcet? Later-no-help plus monotonicity seems to correspond to a voting method that's immune to turkey-raising.
: Feels like I'm dancing around a proof/argument that score and approval are "maximally-strategy-resistant" in some sense, in that they minimize the opportunity for order-reversal. But I'm not sure.—[[User:Closed Limelike Curves|Closed Limelike Curves]] ([[User talk:Closed Limelike Curves|talk]]) 17:52, 26 April 2024 (UTC)
|
Revision as of 17:52, 26 April 2024
Failing Later-no-help in IRV
As discussed in the Favorite Betrayal Criterion, my improvement to IRV makes favorite betrayal unnecessary but apparently it will now violate later-no-help.
12 A
8 B>A
6 C>B
IRV and my “improved” IRV elects B.
When 3 voters change their ballots: IRV still elects B. My “improved” IRV changes the result into a paradoxical tie.
9 A
3 A>C
8 B>C
6 C>B
When three supporters of A add the preference C, the probability of electing A has increased. A goes from a total defeat to a small chance. By definition, I believe this violates later-no-help. Frankly, I don’t see that as bad when the intent is to help the candidate being added to the ballot.
If the voter’s intention is to help C because there really is a preference over B, then this change is sincere. That’s well worth creating a tie in which this voter’s candidates of choice have two thirds of a chance to win. It’s later-some-help.
If these voters were replacing A>B (their sincere choice) with A>C, then it’s a tactical move to make the result into a tie when it isn’t there. I can see a candidate wanting to do that, but not a voter. Before changing, these voters were hoping for A or B, and they got B (having indicated they didn’t want C). After the change, they are getting A or B or C. Why choose a one third chance of losing completely when you already have a guarantee you won’t? RalphInOttawa (talk) 14:10, 20 December 2023 (UTC)
Incompatible with majority-Condorcet?
Is later-no-help incompatible with majority-Condorcet, or just simple (plain) Condorcet? Later-no-help plus monotonicity seems to correspond to a voting method that's immune to turkey-raising. LNHe+monotonicity definitely implies Monroe's "nonelection of turkeys", but it's not clear to me that the converse holds, in part because nonelection of turkeys is imperfectly defined.—Closed Limelike Curves (talk) 17:52, 26 April 2024 (UTC)
- Feels like I'm dancing around a proof/argument that score and approval are "maximally-strategy-resistant" in some sense, in that they minimize the opportunity for order-reversal. But I'm not sure.—Closed Limelike Curves (talk) 17:52, 26 April 2024 (UTC)