[EM] Sparring over AV vs IRV at Least of All Evils...

David L Wetzell wetzelld at gmail.com
Fri Feb 3 07:54:36 PST 2012

>> *dlw:JMKeynes* stated that creative thinking begins as a "grey, *fuzzy*,
>> woolly *monster*" in one's head..... my idea is still under development
>> and it very well may remain a heuristic due to the phenomena of theglobal underdetermination of science.<https://www.google.com/search?sourceid=chrome&ie=UTF-8&q=underdetermination+global+science+duhem#sclient=psy-ab&hl=en&source=hp&q=global+underdetermination+science+quine&pbx=1&oq=global+underdetermination+science+quine&aq=f&aqi=&aql=&gs_sm=e&gs_upl=16143l16942l1l17830l5l5l0l0l0l0l146l501l4.1l5l0&bav=on.2,or.r_gc.r_pw.r_cp.,cf.osb&fp=8bfec2db87b5ba25&biw=1490&bih=905>
>> JQ: Certainly, nothing you've said so far is unreasonable for that; a
>>> one-dimensional spectrum with an unknown shift between elections, voters
>>> with some bias towards the top two parties from the last election, and
>>> parties with some cost for mobility, especially with an inability to switch
>>> positions, trying to maximize their win percentage with the least movement.
>>> I believe that in such a model, IRV would be closer to Plurality (ie, bad)
>>> than it would be to any other good system (approval with reasonable voter
>>> strategy, Condorcet, MJ, Range, or SODA). I hope we could make some kind of
>>> a bet, and that my expected winnings plus my curiosity would be enough to
>>> make it worth my while to actually do the test.
>> You'd need to model a random walk of the center that alters all of the
>> positions and the difficulties of mobility of parties.  It's easier for a
>> new party to locate anywhere, but then it's gotta take time to gain
>> voter-support.  I think thereby that voter-participation (or lack thereof)
>> cd also be a relevant factor and the use of a 2-d model.
> It sounds to me as if you're leaving yourself an out. When you're making a
> model to explain a basic behavior, you need to be able to reduce it to the
> essentials. If one of those essentials is something that can't work in
> silicio, OK. But the epicycles you're talking about are the kind of thing
> you add to make your model fit reality better, not to get basic behaviors
> like center-following.

dlw: I know.  I'm still grappling w. the fucking fuzzy monster in my

> We both agree that parties will not follow the center well under
> plurality, and that they will under some other system(s) (such as honest
> Range). I'm saying IRV will be more like plurality, you're saying it will
> be more like the good systems. We should not have to build more than a
> simple model to test this. Until we build that model, reasserting our
> positions verbally has no value.

dlw: I'm saying that it's possible that closely following the center might
be a curse in disguise(since people can be fickle and it takes time to
bring about serious changes).  We need to bridge following the center and
not following the center and IRV+ might be what does that...

>> I'm also guessing from my fuzzy monster that both the distance of the de
>> facto center from the true center and the stability of the de facto center
>> are going to be worthy of consideration, not unlike how mean-variance
>> models are used to evaluate stocks.
>>> Otherwise, I'm tired of debating this back and forth. I'm an empiricist,
>>> not a platonist. (Middlebrow?)
>> I'm not a gambler and not very peculiarly endowed at this point.  (I was
>> a prof of Econ in Idaho, but apparently my car started leaking
>> carbon-monoxide close to the beginning of my time in ID, and that took a
>> toll on me.  I needed some time off and when I started looking for work
>> again it was the great recession.  That was when I began to study electoral
>> reform seriously and it is why I'm currently a tutor  (and a writer of
>> political science fiction apparently).)
> OK, understood. So that means we have to try even harder to make the model
> simpler; simple enough so that my curiosity is enough to get me to test it
> out.

I'm game if others are.  I'd like to do some digging in the lit to see if
others have done something similar... I passed along to Dale Sheldon Hess
an article that modeled something like this.

I think 2 dims matter because there's more scope for epicycles and because
it seems relevant to the US's so-called democracy.  I don't think we need
to model participation in a  very complicated manner, but it is also
relevant for the US and it does help to capture how the de facto mean can
shift away from the true mean.  And I think 3rd parties are better at
moving to a new point on the graph, but there's gotta be something that
keeps them from automatically garnering all the votes close to them...
 This probably needs to be integrated with the model of voter


>> So I hope we can try doing something like this... as a way to push the
>> envelope from static models that presume parties are able to position
>> themselves where they choose.
>> dlw
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20120203/14d02e7d/attachment-0004.htm>

More information about the Election-Methods mailing list