Friday, December 23, 2016

Assessing democracy in the aftermath of Trump's victory, Part IV: In Valence's name...

In Part III, we talked about cons.  Obvious cons, elaborate cons, and how stupid a person has to be to be taken in by them.  Why?  Oh, no reason...  Not relevant at all to the 2016 election.

Where I'm going with this is something I addressed back in May: the concept of "valence."  In formal models of elections, we like to write out equations for everyone's preferences.  Imagine a single policy dimension:  the left-right dimension.  On the far left, you've got the Bernie Sanders/Jill Stein contingent (we call them "moonbats"), and on the far right, you've got the Ted Cruz/Mike Lee contingent (known as the "wingnuts").  We can assign a number to a voter's preference along that dimension, called an "ideal point," and the closer a voter is to that ideal point, the happier the voter is.  By how much?  That depends on how you write the equation.  In jargon, we write the equation such that the "utility function" is "single-peaked" and "symmetric."  That means having only one ideal point, and the further away from that ideal point the voter gets, the less happy the voter is, but direction doesn't matter.  Five points to the left is just as bad as five points to the right.  Think Goldilocks, but for liberalism/conservatism.

Problem:  there is a lot more to being in public office than just the liberal-conservative dimension.  Many years ago, Donald Stokes wrote a paper about "valence" issues, which are issues about which people agree about the goal, but disagree about how to get there, or who can get us there.  Example:  a good economy.  Game theorists took Stokes's little idea and created "the valence dimension" to add into their models.  So, candidates for office don't just have policy positions, where Ted Cruz is way to the right and Bernie Sanders is way to the left.  They also have a score along the "valence" dimension.

What is this "valence" dimension?  It's sugar and spice and everything nice.  It is good stuff.  Traits that we want everyone to have.  I have my criticisms of how it is used in legislative elections, but we're talking about a presidential election, so let's just go with it.

There are two canonical valence traits:  competence and honesty.  More competence, ceteris paribus, is good.  More honesty, ceteris paribus, is good.

So, rather than simply wanting a specific amount of competence or honesty, voters just want more and more and more, the greedy bastards.  In math jargon, they have "monotonically increasing" preferences over these traits.  So, in addition to wanting policy as close as possible to their ideal points, they also want a candidate with as much valence as possible.

How does the math work out, then?  When faced with an ideologically distant but competent and honest candidate versus an ideologically proximate but dishonest fuckwit, the rational thing to do is often to vote for the former.  That's the point of the valence models.  That's why the models were created.  Why?  Ideological positions are, by definition, contestable and debatable.  How much objective damage can be done by widespread corruption?  How much objective damage can be done by true incompetence?

Doesn't that kind of depend on a bunch of stuff?  I guess we've got more to cover!

In the meantime, um, here's Walter White...

No comments:

Post a Comment