[EM] brainstorm'n electoral calculus on acid...

[Jameson Quinn]

2012/2/3 David L Wetzell <wetzelld at gmail.com>

> What if the electoral space goes back and forth between a 2-d space and a
> 1-d space?
> For every election, there's a randomly generated weight given to the 2-ds
> that has some continuity over time.
> Like lets say that the weight given to one dimension at time t is vt and
> the weight to the other is 1-vt and vt is based on an xt variable that goes
> from negative infinity to positive infinity such that
> vt=Exp(xt)/(1+Exp(xt)) and xt= .8*xt-1 + ut and ut has a standard normal
> distribution.
>
> Now, let's postulate a cob-web model of decision-making.  <https://www.google.com/search?aq=0&oq=cob-web+model&sourceid=chrome&ie=UTF-8&q=cobweb+model>
>
> Existing party's candidates make decisions in 2-d space, using vt-1
> weights and some sort of friction that inhibits their ability to reposition
> within each of the two dimensions.
>
> OTOH, a new party's candidates enter into 2-d space anywhere based on the
> new period's weights.
> However, the new party makes its entrance decision based on a cost-benefit
> decision using the incorrect assumption that voters make decisions based on
> the 1-d of this period.
>
> However, voters actually decide based on the weighted average of this
> period and last periods' 1-d positions of candidates relative to them.
>  Let's just use a 2-period moving average for now.  They treat the prior
> period distance of the new party's candidate as the closest corner on the
> 1-d space...
>
> [There also needs to be some expected utility and fixed disutility from
> voting that determines who votes and who does not vote to enable the de
> facto center to be severed from the true center but that feature could be
> introduced later.... ]
>
> More importantly, we need some sort of "money" on the table to justify
> entrance and exit, movement and maybe the merger of parties.
> If I had to choose between endogenous voter-participation and endogenous
> party participation, the latter would be more relevant, since we're talking
> about the desirability of a 2-party vs multi-party system and the no. of
> parties really needs to be endogenous.  So how do we keep the "losers" in
> the game?  Obviously, there's going to be a certain taste for political
> participation that is based on the strength of their support which lets
> them absorb some losses.  [Another twist would be to also have a less
> valuable 3-seat election, using LR Hare, that would give two or three
> parties some additional cash-flow... and which could try out my IRV+ + Am
> forms of PR idea, with other election rules replacing IRV+.]
>
> Finally, we'd need to come up with a way to measure and assess the
> outcomes.
> Here we need a weighted average of the diff between the winner and the
> true center in 1d and the lack of variability of the winner in 2d.
> This allows that nailing the center, or electing the CW, is not the
> end-all-be-all for how to assess election rules.  We could compare and
> contrast the relative performance of election rules in three cases.  The
> first would be where all the weight is on the distance from the true
> center.  The second would be where all of the weight is on the stability of
> the winner in 2 d.  The third would be a mixture of the two, perhaps to be
> progressive 2/3rds on getting closer to the center and 1/3rd on the lack of
> variability.
>
> I think IRV+ will perform well in the mixture assessment.
>

> Any thoughts/suggestions?
>

1. I disagree; I do not think IRV will do well in the scenario you describe.

2. It's too complex. We need toy models that focus on one aspect at a time,
not anything that tries to be realistic. Think macroeconomics 101
(saltwater), where anything that doesn't fit on one graph is put off until
next year.

jq
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20120203/0b60c598/attachment-0004.htm>