I think I learned something today.
Peter
On Jul 20, 2017, 5:15 PM -0400, Coral Sheldon-Hess <[log in to unmask]>, wrote:
> I agree with all of this, except for the idea of using ranked choice voting
> (aka "instant runoff voting" aka "IRV") option. For reasons that are
> mathematically
> interesting <https://electology.org/approval-voting-versus-irv>, ranked
> choice voting is not the best tool to help us determine the will of the
> community--short version: it's easy, under ranked choice/IRV to
> accidentally betray your favorite and help an option you don't like as well
> to win. Like in plurality voting (what we use for US elections, generally:
> choose a single option from the list), you have to vote strategically,
> rather than being able to vote honestly, or else you might risk your
> least-favorite option winning; and, unlike in plurality, it's sometimes
> hard to calculate the best strategic vote for you to make. Also, properly
> tallying the votes is annoying, for whoever has to do that.
>
> As someone on the committee that might have to tally votes, I am not even a
> little bit interested in trying to implement IRV/ranked choice; it's awful.
> As a member of the community who wants my vote to matter and who wants the
> community to have a fair say about this, I *definitely do not* want to use
> IRV/ranked choice.
>
> *When there are more than two options on a ballot, the better approaches,
> to find out what a community wants, are approval voting
> <https://electology.org/approval-voting> and its slightly more nuanced
> cousin, score voting <https://electology.org/score-voting>*. We already do
> score voting when choosing conference programs, so we know what that looks
> like and can easily put that into the Diebold-o-tron. Approval voting is
> just "check everything that you're happy with"; so, for reasons that are
> probably obvious, it's very easy to calculate and very good at finding the
> option that the most people will be happy with. ("Minimizing Bayesian
> regret," in the lingo.) I imagine getting the Diebold-o-tron to do that
> will not be a problem, either.
>
> Just as another suggestion, because I've been thinking about this for a
> while: we could also treat these questions as independent. I'm giving an
> approval voting-based example, below, but it could be rephrased to work
> with score voting ("give a numeric score to each of the options below") or
> even simple plurality ("choose your favorite option from this list").
>
> 1) Choose all of the options that you find acceptable:
> * Do nothing
> * Find a fiscal sponsor
> * Incorporate as a nonprofit entity
>
> 2) *Assuming the community as a whole wants to go with fiscal sponsorship*,
> choose the options you find acceptable:
> * ALA/LITA
> * DLF/CLIR
> * OLF
> * Other
>
> 3) *Assuming the community as a whole wants to go with incorporating*,
> choose the options you find acceptable:
> * 501c3
> * LLC
> * ???
>
> There's a possibility that we lose some nuance by separating the questions
> out like I have, above. This is worth discussing. I'm just suggesting this
> possibility, in case it's helpful to us, in designing our voting mechanism.
>
> I'm happy to discuss the nuances of voting math in more detail, if anyone
> cares to do so, but, um, no worries if nobody's into that. :)
>
> Best,
> Coral
