[EM] Independence of Pareto-Dominated Alternatives--clarification of what I meant

[Michael Ossipoff]

I'm not saying that Independence of Pareto-Dominated Alternatives,
Independence of Smith-Dominated Alternatives, and Local IIAC aren't
important.

I'm just saying that they aren't important operationally. They aren't
important other than as embarrassment-criteria, vulnerabilities to
criticism in an enactment-campaign.

But, as embarrassment-criteria, as is ordinary, un-weakened, IIAC.

(But of course we don't talk about that one, because Condorcet methods
fail it, and some of us want to propose Condorcet methods)

To better apply these statements to the text that I replied to before,
I'll re-quote some of it, and re-reply to it:


Kristofer said:

Basically, one that gives good results while being resistant to
tinkering by the parties (who have greater capacity to coordinate
strategy than do the voters, and more to lose under the new regime),
and not giving weird results or having weird result dynamics that
could be used to discredit the method.

[endquote]

...unless it's a weird result of Condorcet &/or MJ, in which case it's ok :-)


Kristofer continues:

In practice, that means: is cloneproof, passes independence of as much
as possible (independence of Smith-dominated alternatives, say)

[endquote]

...but not ordinary, un-weakened IIAC, because you want to propose Condorcet

"As much as possible"? Exactly. As much as possible while still
letting-through the methods that you want to propose and advocate.

Kristofer continues:

, and is monotone.

[endquote]

But you conveniently forgive failure of a stronger monotonicity
criterion, Participation, because you want to propose Condorcet and
MJ.


Kristofer continued:

River would be even better than Ranked Pairs, since River passes
independence of Pareto-dominated alternatives and RP doesn't.

[endquote]

But you conveniently disregard the fact that all of your "strong"
Condorcet methods, such as Beatpath, Ranked-Pairs, River, Kemeny, and
VoteFair, have the (easily-avoidable) chicken dilemma, and fail FBC,
Later-No-Help (I've described the strategic importance of those two
criteria), Mono-Add-Top, Participation, Consistency, and IIAC.

While haggling over the difference between IIAC and Independence of
Pareto-Dominated Alternatives, you don't notice the gross, serious and
major strategy-need problems resulting from chicken dilemma, FBC, and,
to a somewhat lesser extent Later-No-Help.

...and the fatal criticism vulnerability resulting from failure of the
embarrassment criteria listed in two paragraphs before this paragraph.

Approval and Score meet all of the abovementioned criteria, though
they have the chicken dilemma. Chicken dilemma is a nuisance, but
isn't difficult to deal with. But, for rank-balloting, with which
chicken dilemma is easily gotten rid of, the chicken dilemma nuisance
is inexcusable.

Symmetrical ICT doesn't have the chicken dilemma, and meets all of the
abovementioned criteria except for the embarrassment criteria
Participation, Mono-Add-Top, Consistency and IIAC and some of its
weakenings (failure which, as I said, would probably be fatal to any
enactment proposal of any Condorcet version, or MJ).

Kristofer said:

We put strong in quote marks because I know others may disagree with
my priorities. FairVote obviously doesn't consider the "having weird
result dynamics" part important as long as the strangeness can't be
exploited by deliberate strategy.

[endquote]

Kristofer said:

Enactment opponents won't agree, and will make sure that the voting
public hear all about those "weird dynamics" embarrassment criterion
failures.

The second is resistance to noise and strategy. Independence of clones
fit here, as well as independence of X (Smith-dominated alternatives,
Pareto-dominated alternatives, weak IIA).

[endquote]

But not Independence from Losers (Usually called Independence from
Irrelevant Alternatives (IIAC) )

That doesn't "fit here", because Kristofer wants to propose Condorcet.
.
Kristofer said:

The third is quality of the outcome under honesty, according to some
metric or desired logic. It's hard to say which metric one should
pick, unfortunately, and for Ranked Pairs (and Schulze), there's
probably no simple metric that the method optimizes. Furthermore, the
logic one uses for rated methods probably wouldn't directly fit onto
rank methods (because utilities are either unknown or not applicable).

[endquote]

For Approval and Score, their winner-choosing rule is clearly about a
good measure of the right winner: The candidate approved by the most
people. The candidate liked and trusted by the most people. And even
if voting is strategic, the winner is the candidate considered
better-than-expectation by the most people, meaning that the winner is
the one whose election pleasantly surprises the most people, the
candidate whose election confirms the most voters' optimistic hopes.

What about Symmetrical ICT? It chooses the (legitimately-defined) CW.
You've discussed the natural justification of that, when people are
free of disincentive to rank sincerely. When there isn't exactly one
unbeaten candidate, there is obvious rightness in choosing the
unbeaten candidate who is top-ranked (marked as a favorite) by the
most people. When no one is unbeaten, there obvious rightness in
electing the candidate who is top-ranked by the most people.

Mike Ossipoff