[EM] An interesting real election

[Kevin Venzke]

Hi Aaron,

--- En date de : Lun 31.1.11, Aaron Armitage <eutychus_slept at yahoo.com> a écrit :
> > >1 has a path to 6 at least as strong as 6's path
> to 1,
> > namely 1>3>6, at 
> > >15-11 and 14-11. It
> > >seems a little odd, to me at least, that 6's path
> to 1
> > should benefit 2 
> > >but not 6 itself.
> > 
> > When you say "benefit" do you mean "elect" or
> something
> > more broad? It
> > seems to me election is the only meaningful benefit
> but of
> > course only
> > one candidate can receive it.
> > 
> 
> In this context I mean benefit in the pairwise comparison.
> So in this
> case, using Schulze, 2 receives a pairwise benefit against
> 1 from the
> direct 6 vs. 1 comparison, but 6 itself (himself,
> herself...) does not.
> Actually winning office is the main reason why a candidate
> would care
> about his pairwise comparisons, but not the only one --
> doing well gives
> him a stronger platform for future elections, makes him
> more attractive to
> donors, makes him more likely to be taken seriously, and so
> on.

Well, in that case I think 6 does benefit from beating 1 surely? Better
to be in the Schwartz set than not.

> > >Starting from the top seems the only way of
> ensuring
> > that the path that 
> > >orders the two
> > >candidates relative to each other is the one
> which
> > actually contributes 
> > >to the final outcome.
> > 
> > I don't understand this. Are you saying the Schulze
> outcome
> > in this
> > election is an example where these two things
> differed?
> > 
> 
> Well, yes. 6>1 is part of 2's successful path to 1, even
> though 1 is
> placed over 6 by the 1>3>6 path.

That concept of 1 deserving insulation is specific to Tideman though. If
"the path that orders the two candidates relative to each other" means the
Tideman ranking then certainly you don't have much choice.

> > It's true that 1's path to 6 is better than the
> reverse,
> > but the only
> > method that will never elect the loser of such a
> comparison
> > is Schulze.
> 
> Maybe I'm missing something. I ranked pairs, aren't the
> strongest paths
> locked in before the weaker ones are considered? That is,
> the weakest link
> of the weaker path is only considered after all the links
> of the better
> path are locked in.

Yes, that is true. I just took a guess at what you meant.

What I was saying is that if the beatpath from A to B is stronger than
the beatpath from B to A, then only Schulze will never elect B.

Kevin