[conlang_learners] As for me, no auxlangs

[Sai Emrys]

On Fri, Jun 12, 2009 at 5:36 PM, Jim Henry<jimhenry1973 at gmail.com> wrote:
> Do you have any suggestions about what voting systems have yield that
> kind of information, or about ways to analyze the ballots in an IRV
> system to get that information from them?

Not really. It's a kinda tough question, and not what voting systems
are intended for.

Voting is for when you have a single thing that needs to be decided,
whereas what you are asking here is how many outcome slots there ought
to be.

Given any specified number of outcome slots, IRV will fulfill them
pretty decently (although I should point out that one of the funner
parts of my discrete math class was proving that every voting system
is inherently flawed, i.e. gives a clearly undesirable result in some
situations - cf. http://www.fairvote.org/?page=1688).

I don't think there's an objective way to determine what the number of
slots ought to be, though. The closest metric I can think of is to
measure e.g. average voter satisfaction (as measured by their votes -
i.e. 1 if you get your top pick, .75 if you get your next, etc,
depending on # rankings you have), and balance that in various
scenarios (1-slot, 2-slot, etc) against more subjective factors, like
minimum # people needed to have critical mass, group cohesion, etc.

It's entirely unclear, in any case; I think inherently so. As
ethnobiologists would put it, you can choose the number of ethnicity
in the world and relatively easily determine the genetic profile of
each - but you can't determine the number of ethnicities from the raw
data, 'cause there is no such objective category there to be found.

- Sai