[EM] A simple thought experiment.

Jameson Quinn jameson.quinn at gmail.com
Wed May 29 14:05:37 PDT 2013


2013/5/29 David L Wetzell <wetzelld at gmail.com>

> Let's agree tentatively that Alternatives to IRV don't really outshine IRV
> unless the number of competitive candidates exceeds 4.
>

I think that's false (the number should be 2.5), but OK, let's see where
you're going with this.

 Then, if we used a 1+ the output of a poisson dist'n to model the number
> of competitive candidates in an election, it could shed some light on the
> debate.
>
> Let's say that since Nate Silver tended to give Obama 2:1 odds over Romney
> that we tend to have 1.5 competitive candidates in our current system so a
> good starting parameter is 1 + the outcome of a poisson distribution with a
> mean of .5.
>
> Then,
> =1-poisson(3, .5, 1)=.002, which would account for how most activists are
> non-plussed by the debate over which single-winner election rule is the
> best and willing to go with IRV since 1-poisson(1, .5, 1)=.01 and IRV wd
> thereby reliably allow the number of candidates who aren't competitive but
> able to make a difference to rise considerably.
>
> But if IRV's use tends to increase the mean number of candidates say to 2
> then 1-poisson(3, 1, 1)=.019, and 1-poisson(1, 1, 1)=.26.  Here, it seems
> that
> that IRV will still suffice most of the time and even when C is larger the
> loss from a non-CW candidate winning wouldn't be that bad if (s)he isn't
> that far off from the CW candidate.
>
> But let's say instead we raised the mean to 2.5 then 1-poisson(3, 1.5,
> 1)=.066, and 1-poisson(1, 1.5, 1)=.44.  We'd still have all the advantages
> with none of the down-falls in almost half of the time and in less than 10%
> of the time would there be a situation where IRV could be significantly
> improved upon.
>
> But even if the number of competitive candidates did show potential to
> balloon, as suggested by simulations and consuitations with experts, then
> there'd be scope for switching from IRV to another election rule.  But in
> the  current situation, C tends to be small and it's more important to
> improve the quality than the quantity of competitive candidates, which can
> be done by expanding significantly the number of non-competitive but
> earnest candidates on the ballot and in the public square.
>
> Also, the bottom line is that when you're advocating for a change in which
> single-winner election rule alternative ought to be used, it's not right to
> dump the burden of proof on IRV advocates.  The amount of time spent
> marketing IRV already is a sunk cost and so the burden of proof for
> switching ought to lie on the challengers not the defenders of the status
> quo progressive electoral alternative to fptp.
>

As to the burden of proof, I agree with you. In my opinion, that burden has
been met. It's still good to avoid infighting, and I'd still vote for IRV
out of solidarity, but if I were writing a constitution and I had to choose
between, say, IRV and real runoffs, I'd choose the latter in a blink.

>
> dlw
>
> ----
> Election-Methods mailing list - see http://electorama.com/em for list info
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20130529/7a0db88a/attachment-0004.htm>


More information about the Election-Methods mailing list