[EM] Kevin's SFC objection

[Michael Ossipoff]

Kevin had said:

> > Hi, By the way, you can also ensure a method satisfies SFC by having it 
> > comply with this votes-only criterion: "If more than half of the voters 
> > rank > A over B, but there is no majority of the voters ranking some 
>third > candidate over A, then B doesn't win." > >
I replied

> > First, does Kevin’s criterion apply only to rank methods? My criteria > 
>apply to all methods.

Kevin replies:

I wrote the above to help clarify the implications of SFC on at least rank 
ballot methods.

I reply now:

Ok, so you were just offering a pedagogical help for Warren. Warren had 
offered your definition as a better wording for SFC.

Kevin continues:

I won't get into again how I interpret criteria so that they apply to all 
methods

I reply:

Yes, it’s probably best that you don’t. Because, last time you did, posted 
your rough sketch of what sort of criteria system you proposed,  and then I 
asked you for:

1. A precise definition of your proposed criteria system

2. A demonstration that uses that definition to show that Approval and 
another RV version pass or fail SFC.

And your answer?: ….silence

Your proposed criteria system was a sketch that never took precise form that 
could be used to show that Approval or RV meet or fail SFC.






When asked why he feels that SFC isn’t about a strategy guarantee, Kevin 
says:

I've told you why in the past actually. I don't think it is a very *useful* 
guarantee, since you can't know in advance whether you qualify for it

I reply:

That seems an odd objection, because the whole reason why there can be 
strategy dilemma, and the reason why we need a better method than Plurality, 
is because one _doesn’t_ have complete information, doesn’t know what the 
situation is. Are there criteria with which a voter knows for sure that the 
criterion’s premise applies in a particular election, and that s/he is one 
of the voters to whom the criterion makes its guarantee?

Anyway, you do know that, if there is a CW, you prefer that CW to the 
candidates to whom you prefer the CW. All that’s additionally needed is a 
majority who prefer the CW to those same candidates. But it’s hardly unusual 
for the CW to be preferred to other candidates by a majority. So if you 
prefer the CW to Y, then most likely a majority does too--because that’s the 
CW.

So I guess you aren’t being entirely clear with us about in what sense you 
feel that the voter doesn’t know if s/he “qualifies for” SFC’s guarantee.

Note that it isn’t necessary to know who the CW is. Whoever the CW is, no 
one worse (in your estimation) will win, if (as is typically the case) a 
majority prefer the CW to him/her. For that, you and the rest of that 
majority need do nothing other than vote sincerely.

It’s true that you don’t know for sure that falsification won’t occur on a 
scale sufficient to change the election outcome--but that’s unlikely to 
happen, for reasons that I went over in a posting within the past week.

Someone could object: “You’re saying that most likely there won’t be enough 
indifference to keep the CW from having a majority over Y, but if, elsewhere 
too,  there’s no truncation, Pairwise-count methods don’t need SFC.” But the 
benefit of SFC is for a majority. That isn’t an unreasonable premise 
stipulation. A group smaller than a majority typically, with any method, 
can’t really expect to get its way. For instance, if it’s Nader, Hillary 
Clinton and McCain, and Nader + Clinton don’t have a majority--so McCain has 
a majority--there’s not a lot that you can expect. The Lesser-of-2-Evils 
problem is about what has to be done to enforce majority wishes.

And, as I said, for some other candidate Y, there typically is a majority 
preferring the CW to Y.

Kevin continues:







, and if you did know that you qualify for it, you'd also know that you 
don't need to use it, since your candidate should win no matter what you do 
with your lower rankings.


I reply:

Here I must admit that I have no idea what Kevin means. My candidate? SFC 
isn’t about my candidate or your candidate winning. It’s about some 
candidate Y _not_ winning, when a majority prefer the CW to him/her and vote 
sincerely, when no one falsifies (or at least falsification doesn’t occur on 
a scale sufficient to change the election outcome).

“If you know you qualify for [SFC’s guarantee]…” Does that mean “If you know 
the CW is preferred by a majority to the candidates whom you like less than 
the CW…”? If so, then Kevin is mistaken, because even if you know that, 
there is no guarantee that one of those less-preferred candidates won’t win, 
if the method doesn’t meet SFC. With non-complying methods, truncation by 
voters outside that majority can cause one of the less-preferred candidates 
to win. With IRV, Borda, etc., it doesn’t even require truncation.

Kevin continues:

Well, the way I understand CC and "pairwise count method," MMPO and MDDA are 
pairwise count methods that satisfy SFC but fail CC.

I reply:

Any kind of method would do. Thank you. If those methods meet SFC and fail 
CC, that means that it isn’t possible to say with assurance that SFC is 
stronger than CC. I much prefer SFC’s guarantee to that of CC because CC 
only gives its guarantee if everyone votes sincerely, whereas with SFC only 
that beneficiary majority need vote sincerely, and everyone else is only 
stipulated to not falsify.

Mike Ossipoff