[EM] IRV & Spoiler Effect

[Paul Kislanko]

Eric Gorr asked:
If this is true, no ranked ballot method can be free from the spoiler 
effect, but do we not generally claim that the better Condorcet 
methods are spoiler free?

Yes, that has been claimed, but in the very first example of "pure"
Condorcet I saw, there was a spoiler, and in the examples that
demonstrate the difference between methods, you can always find "the
spoiler."

The only difference in the methods is whether the "spoiler" is a
candidate who by plurality has a higher score than the CW or a lower
score than the CW.

This is sort of the essence of Arrow's proof - if you satisfy the other
four criteria then there must be a "dictator". The normal English
interpretation of that is "somebody who tells everybody how to vote",
but all the logic says is "someone who uniquely determines the outcome
of the election". That is not someone who stands for the election
necessarily ("the spoiler") but instead it is the (bloc of) voter(s) who
selected the candidate who "spoiled" it in a position such that a
majority of voters are disappointed with the outcome.

Anyway, I am not qualified to debate the subject, but its been proven
that no ranked ballot method can meet all four of Arrow's criteria, so
any that claim to meet all but the Dictator Criterion must necessarily
fail that one.

Most EM advocates of one method or another choose to weaken one of the
other three, which is certainly "fair" as long as it is acknowledged,
but when you do that then there can be a "spoiler" in it (relate it back
to whether the system is "strategy-free", which is a different
impossibility theorem I'm too lazy to look up).

Again, I'm not qualified to analyze any particular method and do not
wish to get yelled at again for my ignorance, but if all the theorems
that have been proven are indeed true then the claim that any particular
ranked ballot method cannot be "spoiled" by either a voting or
nominating strategy is not.

Sorry, I'll go away again.