[EM] Condorcet How?

[Kevin Venzke]

Hi Juho,

--- En date de : Ven 9.4.10, Juho <juho4880 at yahoo.co.uk> a écrit :
> >>> I more or less despise the election of A in
> this
> >> scenario:
> >>> 
> >>> 49 A
> >>> 24 B
> >>> 27 C>B
> >>> 
> >>> I believe the possibility of this outcome is
> a
> >> disincentive for a candidate
> >>> like C to run.
> >> 
> >> The story behind these votes seems to be that C
> is
> >> ideologically close to B and its nomination makes
> A win
> >> instead of B. If C is ideologically close to B why
> don't the
> >> "B" voters vote "B>C" (and make C win)?
> > 
> > Because the story is that C is not considered an
> established candidate
> > (or a frontrunner), and the B voters (if they actually
> prefer C, which
> > is not something I mean to suggest) will not
> condescend to rank C.
> > Depending on the method, they could hand the election
> to C by ranking C,
> > and/or alternatively compel C supporters to not vote
> sincerely.
> 
> Ok, but the story is not quite flawless yet.
> 
> Case 1: "B" voters did not notice C. B had 51 supporters
> before the nomination of C. All that heard of C changed
> their opinion to "C>B". If C is so obviously better than
> B then also some of the "A" voters should have voted
> "A>C" (or "A>C" if they feel the other way around for
> some reason). But it seems that also they did not notice C.
> It is also strange that someone can get 27% support (clearly
> more support than B) and still be unknown to so many of the
> voters.
> 
> Case 2: "B" voters did notice C and liked C but didn't want
> to vote for C. In this case the votes are not sincere but
> twisted for some psychological reasons. (Also some part of
> "A" voters should have ranked B or C if they noticed C.)

I don't get why people assume that B voters secretly do like C. If I
vote for a left-wing candidate, that doesn't mean I like all left-wing
candidates better than all right-wing candidates.

Yes, you are right: I am proposing that voters backing more established
candidates may insincerely decline to give lower preferences to other
candidates. But this isn't simply because they are mean. It benefits
candidates like B if the nomination of candidates like C risks wrecking
the result from the perspective of C voters.

> It seems that the role of non-informed voters and/or
> non-sincere/mutinous voters and the surprising emergence of
> candidate C have a strong role in the story, and all the
> blame (of not following the story (and intended outcome)
> that was discussed above) can not be put on margins alone.

I don't understand that way of thinking. Uncertainty about the strength
of a candidate is going to exist everywhere. Self-interested strategy
is going to exist everywhere. I guess you could excuse margins from
messing up here, but what's the point?

I'm aware you can find scenarios where WV offers a strategic incentive
where margins doesn't. I don't think I've seen theories about general
patterns of this, or any particularly relevant and alarming scenarios,
etc.

> > The best scenario is for C to be able to run, be voted
> for sincerely, and
> > either win, or not affect the outcome at all.
> 
> Yes, that would be ideal.
> 
> > 
> >> If C is
> >> ideologically far from B then most "B" voters may
> vote as
> >> they do now, but then maybe "A" voters should vote
> "A>B"
> >> (and make B win) (since C seems to be closer to B
> than A).
> >> It is hard to find an explanation to sincere votes
> like
> >> this. Or are these maybe strategic votes, e.g. so
> that
> >> sincere "A>B" voters have decided to vote "A"?
> > 
> > The A voters will not vote A>B because A and B are
> considered the
> > frontrunners. (i.e. that is the scenario that concerns
> me.)
> 
> Yes but also this means that "A" voters are badly
> misinformed when none of them have heard of C or believe
> that C is weak although C has 27% first place support
> (clearly more than B) and all that heard of C "on the B
> side" voted for C.

You keep saying 27 is clearly more than 24. I don't think it is outside
the realm of possibility that a candidate expected to place third
would actually place second.

You can't really conclude A voters are misinformed, not from the votes
anyway. A voters are going to win it, and voting A>B turns B into the 
CW. Notice that C is not really very close to winning.

> >>> Fundamentally, if I said only one thing: I
> don't
> >> believe that the margins
> >>> ranking of defeat strength (resulting from
> its
> >> treatment of unranked
> >>> candidates) is in agreement with what voters
> would
> >> expect and want.
> >> 
> >> What would you consider to be a better approach
> than
> >> margins for sincere votes? (winning votes has also
> scenarios
> >> that may be questioned)
> > 
> > Maybe Range? I guess I don't see that question as
> interesting or useful.
> 
> Yes, Range could be considered more ideal at least for some
> uses (when one wants to maximize the sum of utilities)
> (there are also other alternatives like minimizing harm to
> some and seeking majority). The problem of course is that it
> is so difficult to collect sincere ratings. I think the
> question is important since performance with sincere votes
> (= elect the correct winner) is important.

It is difficult to collect sincere ratings. But doesn't your question
assume we can get them?

By the way, I do want to maximize the sum of utilities. I just don't
think you can be so direct as to ask for them.

I wonder how the idea of performance given sincere votes applies to WV
and margins for you. If everyone ranks all the candidates, they'd be
the same method. If people are truncating, I do prefer WV for the reason
I've already said, that WV treatment of truncation is better (closer to
what is expected) because it places more weight on the "least abstained"
contests.

You have probably explained before why you think margins is more
intuitive than WV even given *sincere* votes. I don't remember what the
summary of this is though.

> >> (Note btw that in another mail I just mentioned
> also the
> >> possibility of allowing B and C to formally team
> up so that
> >> defeats within their team would not be considered
> as severe
> >> as defeats between A and the team.)
> > 
> > This doesn't interest me much (same as Forest's
> suggestion of pre-
> > election agreements in DYN) because I'm not interested
> in the case that
> > a single party (for all practical purposes) nominates
> two candidates.
> > I don't think that will normally happen or be
> desirable under any
> > method. What I'm concerned with is the viability of a
> "third" option and
> > voting for the third option.
> 
> I don't believe such strong methods like teaming would be
> generally/typically needed. But if there are problems then
> teaming may be a good way to overcome them (they may e.g.
> quite efficiently cut out any strategy considerations). I
> also understand that in some environments all "agreements
> between parties" may be considered to represent too much
> some "spirit of old times".
> 
> In summary, maybe the "third" party explanation is more
> what you are after than the "not heard of C" explanation. In
> the example "B" and "A" voters were maybe from some old
> established parties and they did not want to recognize the
> emergence of a "third" new strong candidate.

Yes, that's right.

> I believe such
> problems may well be temporary. If one adopts a new method
> that allows also "third" candidates to run in a meaningful
> way and with real chances to win if they have strong enough
> support, then I'd guess the attitude and problems of
> ignoring them and not ranking them could fade out very soon.
> The two main contenders and voters that support them can not
> pretend any more that other candidates do not exist (one can
> e.g. not play down a candidate with 27% of the first
> preference votes).

That would be nice, but even if true, it is just a mitigation of the
problem, not a reason to refrain from eliminating the problem.

Kevin Venzke