JG replies intercalated after RB comments:

JG original> > In an STV-PR election (a.k.a. RCV), the voter's second and any subsequent preferences are contingency choices, to be used only in the contingency that the voter's first choice candidate cannot be elected (because of lack of support) or has already been elected to represent a full quota of voters (and so does not need the additional support).

RB> But James, we know that that is not always the case.  Burlington 2009 and now, Alaska 2022, are counter-examples that disprove that.

RB> In Burlington 2009, Kurt Wright voters were promised (as we all were promised) that if their first-choice cannot win, their second-choice vote is counted.  Wright was defeated and those voters' second-choice votes were not counted.  Had their second-choice votes been counted, a different candidate for mayor would have been elected.

But who made that "promise"?  That "promise" is totally invalid because that is not the way STV-PR (RCV) works.

RB> Yesterday, the Alaska Division of Elections released the Cast Vote Records, and now we know this was repeated in Alaska last month.  Sarah Palin voters were promised (as we all were promised) that if their first-choice cannot win, their second-choice vote is counted.  Palin was defeated and those voters' second-choice votes were not counted.  Had their second-choice votes been counted, a different candidate for U.S. Congress would have been elected.

Again, who made that "promise"?  That "promise" is totally invalid because that is not the way STV-PR (RCV) works.

RB> These are cold hard facts supported by the public record that cannot be disputed.

They may be "cold hard facts", but they are based on a serious misunderstanding of how STV-PR works, so they are not facts at all.

JG> > Where a voter does not mark a preference against every candidate, that voter is telling the Returning Officer that he or she has no preference among the unmarked candidates,

RB> But that voter is expressing on their ballot that he or she does have a preference of any marked candidate over any of the unmarked candidates.

Yes, that is true, but I didn't bother stating that in my original message.

RB> That is equivalent to that all unmarked candidates are tied for last-choice preference on that ballot.

No, that is not true.  No further preference means just what I said, i.e. "I do not wish to express any further preference".

JG> > and that if any choice has to be made among those candidates, he or she is happy to leave that decision to those voters who do have preferences among those candidates.  Unmarked preferences mean nothing more than "I have opted out at this point and leave any further decisions to others".

RB> No.  That's not true.  Unmarked preferences mean "I prefer these candidates less than I prefer any candidates that I have marked".

Your contradiction of my statement is illogical.  My original statement is fully in agreement with your statement "I prefer these candidates less than I prefer any candidates that I have marked".

Your error comes from implying more about an unmarked preference than can be justified.  "No further preference" means "No preference AMONG the unmarked candidates".  No-one is entitled to imply or input that it means any more.

James Gilmour

On 09.09.2022 18:42, robert bristow-johnson wrote:

As I understand it, James considers an IRV ballot to be a kind of
program instruction, so that e.g. voting A>B>C is a way of telling the
voting method "I want my ballot to count towards A until he's
eliminated; then I want my ballot to count towards B until/unless he's
eliminated, etc".

From such a perspective there is no failure because the voters gave the
method certain instructions, and the method obeyed these instructions,
and the winner was who was elected since that's what the procedure says.

If I understand that right, then there's no promise of a vote counting
towards B if A can't win, because B could be eliminated before A and so
the ballot skips directly from A to C. Which leads to Condorcet failure,
nonmonotonicity, and so on.

Since IRV passes both LNHarm and LNHelp, the procedural interpretation
does happen to coincide with the (method-independent) idea that unranked
candidates should be considered to rank below every explicitly ranked
candidate, because that's the effect not ranking candidates has on the
method. But if the method were different, the procedural interpretation
would also come to a different conclusion.

(However, I don't think the procedural interpretation is particularly
common; at least it isn't how IRV has been sold.)

-km

Kristofer
If the last statement in your message below is correct, about how IRV been sold, all I can say is that those responsible have been
engaged in serious misselling.

My description, well amplified by you, of how IRV works and is intended to work, is exactly as you set out below.  That is because
IRV, and STV more generally, is a contingency voting system.  It always has been, and it always will be.

Yes, that means IRV and STV-PR fail Condorcet and are not monotonic.  But STV does comply with LNHarm, a criterion that is important
to real voters in real elections.

James Gilmour

Kristofer Munsterhjelm  > Sent: 09 September 2022 21:16

telling the voting method "I want my ballot to count towards A until he's eliminated; then I want my ballot to count towards B
until/unless he's eliminated, etc".

instructions, and the winner was who was elected since that's what the procedure says.

that unranked candidates should be considered to rank below every explicitly ranked candidate, because that's the effect not ranking
candidates has on the method. But if the method were different, the procedural interpretation would also come to a different
conclusion.

Well, let's talk about what RCV is intended to accomplish.  And then frame how it's intended to work from that.

You're insisting that it's intended to work according to Hare STV rules because it's intended to work that way.  It's kinda a circular argument.

I'm saying that RCV is intended to work in such a way that accomplishes some well-defined goals.

I agree with that.  But it failed to work as a contingency voting system in Burlington 2009 and Alaska 2022.

But Hare is not always that.  However, outside a cycle (which has never happened yet in the U.S. that we know of) Condorcet always has been and always will be.

The reason Hare works so good is that 99.6% of the time it elects the Condorcet winner.  The 0.4% of the time when it doesn't elect the Condorcet winner, the election must be a spoiled election.  It fails.

I haven't met a single pedestrian voter say, "I'm sure glad we have RCV because it satisfies Later No Harm."  Not one.

I get from regular people who like RCV is that they believe that it doesn't force you to evaluate the lesser of evils and vote for that lesser of evils  because your worrying that the greater of evils may be elected.

I get from regular people who like RCV is that they believe it frees them to vote their hopes instead of their fears.

I get from regular people who like RCV is that they believe it prevents the spoiler effect.  That they don't have to worry about a Nader spoiling the election for Gore.

I get from regular people who like RCV is that they believe in levels the playing field between the major party candidates and minor party or independent candidates.  Voters are not discouraged from voting for the candidate they really like out of fear of assisting the candidate they loathe to be elected.

I get from regular people who like RCV is that they believe that if their favorite candidate (marked as first-choice) can't, won't, doesn't get elected, their second-choice vote is counted.

I get from regular people who like RCV is that they believe that it guarantees electing the majority candidate, even when there are 3 or more candidates in the race.  People believe that because RCV boils the election down to two candidate, in which there is always a majority, unless they tie.

That is basically how it is marketed.

--

r b-j . _ . _ . _ . _ rbj@audioimagination.com

"Imagination is more important than knowledge."

.
.
.

On 9/9/22 23:54, James Gilmour wrote:

You need borh LNHarm and LNHelp to provide immunity to burial, no?

If you have only LNHarm, then it's possible that an X>W>A voter can turn
the winner from W to X by instead voting X>A>W, thus helping X. And if
you have only LNHelp, then it's possible that an X-first voter harms X
by voting X>W>A but doesn't harm X by voting X>A>W.

If only one of the LNHs is needed, then Descending Acquiescing
Coalitions or Descending Solid Coalitions will get you mutual majority
and the LNH in question -- and monotonicity.

-km

On 09/09/2022 10:31, Kristofer Munsterhjelm wrote:

I don’t agree with Kristofer's statement. A voter Veronica goes to the
ballot box with a set of opinions. These include judgements ("I prefer A
to B") and ignorance ("I know nothing about D"). If she is honest, she
will fill in the ballot in the way which best represents her opinions.
She may do so erroneously if she doesn't understand the system or lazily
if it's easier to say one thing than another,
   The only legitimate meaning of a ballot is the set of opinions it
might honestly express. If Veronica likes A, hates C, has mixed feelings
about B and knows nothing about D, E, and F, then she will fill in her
ballot paper in a certain way; and complete ignorance of D, E and F has
to be accepted as a possible meaning of a ballot filled in in this way.
If you insist on understanding her as disliking D, E, and F, you distort
the meaning of her vote. So how is she meant to vote, given her
opinions, if the ballot paper allows no form of tie besides truncation?
She could vote A>B>C, or A>B, or just A, but none of these is
interpreted by Kristofer in a way that's faithful to her opinions.
   CJC

On 9/10/22 10:48, Colin Champion wrote:

I'm not taking a position, I'm just saying that that's what the "equal
last" interpretation implies, and contrasting that to an interpretation
where unranked means no opinion (kitten duel example).

I'm not entirely sure which I personally think is the better
first-principles interpretation. Practically speaking, I think Forest is
right when he says:

That is, counting unknowns as ranked below everybody else is a way to
prevent a completely unknown candidate from winning, similar to the
convention of filling in unrated candidates as zero in Range. This
suggests that if honest performance is what we care about, unranked
should be considered "no opinion".

On the other hand, there's something intuitively appealing about the
Plurality criterion:

"If A is ranked first on more ballots than B is ranked at all, then A's
probability of winning should be no lower than B's",

which is about a sort of cold start problem: if the data indicates that
some people like A, but the data is less clear about B, then that should
be taken to be in favor of A.

If complexity is no issue, there might be a soft quorum[1] analog that
both satisfies Plurality and doesn't presume too much about the ranking
of unknowns. But I don't know what it would look like.

-km

[1] For range voting: https://rangevoting.org/PuzzDLaplace.html