[EM] Technical nomenclature regarding voting systems.

[Kristofer Munsterhjelm]

On 16.02.2022 08:48, robert bristow-johnson wrote:
> In my paper:
> 
> The Failure of Instant Runoff Voting to accomplish the very purpose
> for which it was adopted: An object lesson in Burlington Vermont
> 
>  https://drive.google.com/file/d/1jIhFQfEoxSdyRz5SqEjZotbVDx4xshwM/view 
> 
> I wanted to boil down the Burlington 2009 IRV failure to these two principles:
> 
>    (1) One person, one vote
>    (2) Majority rule
> 
> and these three desirable properties of elections:
> 
>    (3) Avoid spoiler effect
>    (4) Disincentivize tactical voting
>    (5) Precinct summability
> 
> I believe that (1) is called the "Principle of Anonymity" in 
> scholarly circles, that (2) (as stated in the paper) is the 
> "Condorcet Criterion", and that (3) is "Independence of Irrelevant 
> Alternatives" (I am of the school that says *any* loser that changes
> the winner is a spoiler, not just losers that have no chance of
> winning.)

There are two "shouldn't matter who you are" criteria. There's
anonymity, which means that there are no privileged voters; and there's
neutrality, which means that there are no privileged candidates. For
instance, the Ranked Pairs version where you set a tiebreaker first
violates neutrality, because the candidate who's first on the tiebreaker
has an advantage simply due to who he is. A method that lets the chair's
vote count double violates anonymity.

Strictly speaking, IRV passes both of these. I think you need something
that's more about the relative power of a vote, so that you could
capture the problem of IRV ignoring some votes' later preferences while
accepting others'. But I don't know of any term that captures this well.

> (5) is simply about Process Transparency, an important principle of
> fair and honest elections.
> 
> Is there a better or a more scholarly label for Property (4)?  In my
> paper I describe (4) as:
> 
> 4. Voters should not be called upon to do “tactical voting”. Voters
> should feel free to simply vote their conscience and vote for the
> candidates they like best, without worrying about whom that they think
> is most electable... They should not have to sacrifice their vote for
> their favorite choice because they are concerned about “wasting” their
> vote and helping elect the candidate they loathe. Voters should be able
> to “Vote their hopes rather than vote their fears”.
> 
> Is there a technical name for (4)?  Would it be "Favorite Betrayal"?

Favorite betrayal is one kind of protection from strategic voting. It
basically says "you never need to do a compromising strategy". There are
others, like in Plurality or IRV where you never need to do a burial
strategy.
It's not possible for a deterministic ranked voting method to be
completely free of the need to use strategy, and the reason is actually
related to IIA.

So a method that passes IIA more often should be less susceptible to
strategy. And James Green-Armytage found that making a method Condorcet
compliant never increases the number of times that strategy pays off (if
that method is of a form that a coordinated majority can always force an
outcome).

"Voting your hopes rather than your fears" sounds like immunity to
compromising. If that's right, then favorite betrayal is indeed what you
want. Condorcet methods can't pass this criterion, but some methods come
very close. (And every Condorcet method passes it when there's a honest
CW and the strategists can't engineer a cycle.)

But for the more general concept, I don't think there is one. You could
say "strategy-proofness" or the method being strategy-proof -- that's
the closest you can get, I think.

-km