[EM] Later No Help? Better Late Than Never (was IRV Finisher DSV)

[Forest Simmons]

This IRV finisher is a partial fulfillment of the IRV (fake) promise of
Reliable Later Help. We should call the method BLTN, Better Late Than Never
(aka Bacon Lettuce Tomato Not)

It tries to do what Kevin's Gradual Approval does, but too little too late
... still, better than nothing!

El jue., 16 de jun. de 2022 4:55 p. m., Forest Simmons <
forest.simmons21 at gmail.com> escribió:

> Thanks, Kevin and Ted.
>
> My idea is to give a fair chance to all of the candidates that were
> eliminated "prematurely" ... to partially remedy the failed promise that if
> your first choice is eliminated, then your vote will transfer to your
> second choice.
>
> In the finisher approval step, each ballot B approves every candidate that
> was once its transferred top choice and every candidate that would have
> been its top choice at some stage had it not already been eliminated ... in
> sum every candidate that is not outranked on B by its last transferred vote.
>
> To preserve the flavor of IRV we add one post final step: each ballot's
> vote is transferred to its approved candidate that is approved on the
> greatest number of ballots.
>
> Then elect the candidate with the greatest number of final transferred
> votes.
>
> This method (like IRV itself) is too cumbersome to be a good public
> proposal.  But it can serve as a good educational tool.
>
> In simulations a comparison of the VSE of the outcomes with and without
> the finisher steps will reveal some of the social cost of the premature
> eliminations ... i.e. the cost of the betrayal of the failed promises.
>
> -Forest
>
> El jue., 16 de jun. de 2022 9:33 a. m., Kevin Venzke <stepjak at yahoo.fr>
> escribió:
>
>> With Forest's idea, the sense seems to be that the IRV winner can win as
>> long as that
>> is somewhat reasonable. I.e. that there isn't some eliminated candidate
>> who can defeat
>> all the final-round support for the IRV winner.
>>
>> If the IRV winner will be considered below all cutoffs, it looks like we
>> want the IRV
>> winner to be defeated if at all possible. That doesn't sound good from a
>> monotonicity
>> standpoint.
>>
>> That said, I'm not seeing how the two rules give a different treatment of
>> the "Z>X>..."
>> ballot in that scenario, where X was the IRV winner and Z was the other
>> finalist.
>>
>> Kevin
>>
>>
>> Le mercredi 15 juin 2022, 21:34:38 UTC−5, Andy Dienes <
>> andydienes at gmail.com> a écrit :
>> > It is an interesting idea but I think it would be more fair if it were
>> like:
>> >
>> > IRV winner X, approval winner Y where X is (exclusive) cutoff on all
>> ballots, then
>> > pairwise of X and Y wins
>> >
>> > Otherwise, what if the final round of IRV is X and Z, but my
>> preferences happen to be
>> > Z > X > (everything else), seems like my vote would be mostly ignored.
>> >
>> > On Wed, Jun 15, 2022 at 10:22 PM Forest Simmons <
>> forest.simmons21 at gmail.com> wrote:
>> > > One Approval step determines the winner W after the IRV winner X has
>> been found:
>> > >
>> > > The IRV winner X is the approval cutoff on all ballots ... inclusive
>> only on the
>> > > ballots whose transferred vote went to X on the final IRV round.
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20220617/88c7de8c/attachment.html>