election-methods@mailman.electorama.com

Technical discussion of election methods

View all threads

IRV spoiled-ballots

MO
Michael Ossipoff
Fri, Jan 12, 2024 10:12 PM

There’s no need to regard any IRV ballot as spoiled, or to discard any IRV
ballot:

  1. If a ballot ranks 2 or more candidates at same rank position, then rank
    them in alphabetical order by surname.

The voter doesn’t care about their order, given that s/he ranked them equal.

  1. If the ballot has a candidate at more than 1 rank-position, then average
    (arithmetical mean) those ranks & round-off to nearest integer.

If the average ends in “.5”, then round up.

Round up because there are 2 familiar means: arithmetical & geometric. So
because the arithmetical mean ending in .5 doesn’t work for rounding, then
use the geometric-mean. It’s lower than the arithmetical mean. …so you
round up.

Of course if that would place the candidate at an already-occupied rank,
then implement step #1.

  1. If there are gaps in the ranking, then of course just push them up, like
    beads on a string, to make them consecutive, preserving their order,
    because that’s what the voter implied.

Of course the same applies to any rankings for any rank-count method.
…except of course that Condorcet usually or always allows equal ranking.

There’s no need to regard any IRV ballot as spoiled, or to discard any IRV ballot: 1. If a ballot ranks 2 or more candidates at same rank position, then rank them in alphabetical order by surname. The voter doesn’t care about their order, given that s/he ranked them equal. 2. If the ballot has a candidate at more than 1 rank-position, then average (arithmetical mean) those ranks & round-off to nearest integer. If the average ends in “.5”, then round up. Round up because there are 2 familiar means: arithmetical & geometric. So because the arithmetical mean ending in .5 doesn’t work for rounding, then use the geometric-mean. It’s lower than the arithmetical mean. …so you round up. Of course if that would place the candidate at an already-occupied rank, then implement step #1. 3. If there are gaps in the ranking, then of course just push them up, like beads on a string, to make them consecutive, preserving their order, because that’s what the voter implied. Of course the same applies to any rankings for any rank-count method. …except of course that Condorcet usually or always allows equal ranking.