[EM] Voting paradoxes article

[Bart Ingles]

David Catchpole wrote:
> 
> Much of Saari's work regards positional voting schema, of which
> (obviously) Borda is the best. He uses weird logic to say Borda is better
> than Condorcet in his articles by saying something like "the absence of a
> Condorcet answer corresponds with voting paradoxes, and Borda gives
> solutions in these cases, so obviously Borda is better." I don't think
> he's done much work on approval (not sure though...). I'd prefer to think
> that as time progresses we get back to the generalities of Sen, Arrow,
> Gibbard and Satterthwaite, because it is so apparent that ignoring some
> electoral systems allows us to become biased, as Saari does and as
> approval-voterers do, and as Condorcetists do, etc. etc.

I suppose if you want to base your method on ordinal position, Borda is
best (at least with sincere voting);  if you are interested in
preference intensity levels, you will favor approval;  if you regard all
choices between any pair of candidates to be equally significant, no
matter how ambivalent the chooser, some Condorcet method would be
preferred.

Not sure I understand the justification for using ordinal position the
way Borda does, though.  If you consider it a proxy for utility level,
you could say it was similar to an average ratings system.  The assumed
utility values would be correct on average, but so is a watch stopped at
6:00.

> Personal whinge- multiple-"seat" elections barely get a look in, which is
> a shame because some interesting parallels with Condorcet arise which one
> day I will actually do work on (hopefully).

Just some musings on the subject would be interesting.  I would think
that you could identify Condorcet winners within factions if there is no
cross-voting, but that it could get messy otherwise.