[EM] Suppose, for a moment, there were never any cycles...

Kevin Venzke stepjak at yahoo.fr
Thu Jan 26 20:50:04 PST 2023


Hi Robert,

> > On 01/26/2023 7:30 PM EST Kevin Venzke <stepjak at yahoo.fr> wrote:
> > ...
> > > > The point was that the aspect of the analogy you called disingenuous, isn't likely to make
> > > > much difference. If FPP elects a majority favorite 95% of the time then its observable
> > > > Condorcet efficiency is going to be at least 95% no matter how many ranks you add to the
> > > > ballots,
> > (adding back the rest of the quote)
> > > > because a voted majority favorite is always the voted CW, and adding ranks doesn't
> > > > change the strategy.
> >
> > > oh, no, no.  That's false.  If you add more ranking levels (than just one), the apparent CW
> > > may very well be a different candidate.  That is because voters will rank some candidates
> > > at different levels whereas in FPTP they were all at one level, specifically unranked.
> >
> > When there's a majority favorite, I don't agree with you (I can't see a reason for the
> > voters' inclination to vote strategically to change simply because ranks are added), but in
> > any case it's probably more valuable to understand why I'm using this argument than for you
> > to necessarily accept it.
> >
> > In some ways it's unfortunate to have to seek an analogy in FPP, due to its ballot format.
> > But FPP does provide us with a striking demonstration of how a seemingly very poor method
> > can still manage to avoid presenting one disaster after another.
> 
> To be clear (still not sure I am understanding you on everything) I don't want to equate
> First-Past-The-Post with using a ranked-ballot and counting only the vote for the
> highest-ranked candidate.  Now, when you have FPTP and only one ranking level other than
> unranked, the plurality winner of that election is also the CW from what information in the
> ballot data we know.  But that does not mean that this CW would be the same CW if the
> number of ranking levels were increased to, say, the number of candidates on the ballot.
> 
> When the number of ranking levels is much less than the number of candidates, we know that
> there are ballots that get exhausted early that would not be if the number of ranking
> levels was increased.  FPTP is the ranked-ballot with extreme ballot truncation.  FPTP is
> "everybody gotta bullet-vote".

I get you, I'm not sure I disagree with any of that. What I want to note is that if the FPP
elections always had a majority favorite, additional ranks shouldn't be able to change the
apparent CW. You would have to hypothesize that adding more ranks could make voters less
strategic, or something, to explain why the previous majority favorite might lose their
status.

Kevin
votingmethods.net


More information about the Election-Methods mailing list