Seth Woolley's Blog

Occasional Musings

response to blueoregon kari chisholm on automated redistricting(0)

I was limited to 3000 characters, so I decided to post on my blog and link to this post.

http://blueoregon.com/2011/05/jim-huffman-mathematical-redistricting

Kari, since you're a partisan hack (this is blueoregon after all), it would make sense that you would oppose non-partisan ways of redistricting.  All of them.

However, as a spatial database expert, I can tell you that it is actually not that difficult to do redistricting.  There is only one reason people hate computer redistricting: fear that the lack of political process involved will prevent them from having their political goals included.

In 2001, Bill Bradbury drew maps that were absurdly partisan.  The Republicans thus have a lot of ground to cover to gain a more fair map that distributes voter power more proportionally.

Most of the work on computerized redistricting comes from people who introduce new ways of analyzing districts and you gave a couple mentioned.  I should note that Brian's method is a heuristic method and it depends on how you seed its initial conditions and randomness generator, so it's got a weakness there from the standpoint of completely automated.

But it's not a bad way of doing representative districts that intend to represent geography.

A basic fact is being missed in all this discussion -- what people want to see: a fair election system.  Too many people are ignoring that.

Districts-by-geography are a form of proportional representation -- proportionality not by ideology inherently, but proportionality by location.  There are many advantages to having a house that has this proportionality, but there are countless disadvantages.  For one, the boundaries are not determined by voters, but through a political process.

The goal of election reform should be to find ways to maximize the determination that can be done by voters themselves and to minimize how much determination is done by special interests, monied lobbyists, and entrenched parties.

If we followed that goal we would be maximizing true self-determination in our elections.

Fortunately, there is a reform that is much more important than redistricting -- it is proportional representation by ideology.  And not entrenched parties.  Ideologies.  This is what most students of politics call proportional representation -- and it is most often implemented by means of a party list.

I know your examples are on the federal level, but I think state reform is more possible.

We don't need a party list system -- our Oregon Constitution in Article II, Section 16 outlines the two reforms that could be combined to create a non-partisan house of representatives that truly represent us.  We could leave the Oregon Senate districted by geography -- possibly ignoring "communities related to non-geographic interests" because we will use the House to create a system where communities of ideological interest are best represented.  Right now communities of interest of the ideological stripe cannot be incorporated into a district map unless they are "kinda" close enough to draw a boundary that doesn't look too bad from above.  That's why the problem is so hard to solve -- the election method itself is to blame, not the computers trying to optimize for parameters.

The two election reforms I mention are proportional representation itself and preference voting.  There is a form of preference voting called single-transferable voting that is commonly used in advanced societies like Scotland and Cambridge, MA to allow voters to express their proportional representative interests through the rank ordering of candidates -- and since they are done using preference voting, the spoiler effect that allows strategy plays by entrenched power players is virtually impossible to exploit.  If a minority votes in a block with others of their kind, they will find the outcomes will match what they voted for in the outcome of the election because the votes transfer to concentrate election behind the ideologically similar candidates with the most votes higher in preference.

So all of this tempest could be done by computer if we wanted to.

The beauty of this system is that it actually acts as a true check and balance system -- geographic distribution and proportional distribution can team up to block a bad bill that doesn't satisfy a majority from two different styles of bodies.  The ideological house, I expect, would likely lean more left, but the geographic senate with its communities of interest split up will likely lean more right.  Multipartisanship will be required to get anything through.  The number of viable parties in Oregon would skyrocket as people would feel free to vote their ideologies -- their consciences, rather than hold their noses and vote for Kari's monied candidates.

If we really don't want to make our elections more useful, we could stick to two nearly identical looking geographically divided legislative bodies and then make a computer program that encompasses the statute-based required elements in ORS 188.010.  Showing us pictures of districts not based on the already defined parameters doesn't help us much.

The only thing that should be political about the process is how to decide what coefficients go into the weighting system.

But that is an easy problem, a very easy problem.  I would solve it by asking each sitting legislator to propose the coefficients that fit a given equal power equation -- for example, all multiply to a given number.  They will do this after having been given access to the source code that takes all of the elements of the ORS into account.  The numbers will then simply be geometrically averaged to equal the same power equation.

The ORS defines some interesting aspects, like communities of interest.  I would propose that any communities of interest be decided by simple scientific sampling -- you ask people what communities of interest they feel most associated to (they can vote more than once) and give them five communities to list.  The ones with the most cumulative votes at the end get input in proportion to others multiplied by the aforementioned coefficient.  So somebody who doesn't like communities of interest for one reason (ethnicity) would have to bias against communities of interest for them (say farmers) at the same time.

Everybody gets their say, but the computer outputs the final output in an unbiased manner.  I believe this meets all of your substantive objections, Kari.  Truly, not a bad idea at all.

Seth Woolley's Blog

Leave A Comment

Secret is used for editing your own comment. If subject, secret, and name all are the same as a previous comment, it will be overwritten. Turing is the name of this program (look at the Source Code link on the front page), used to see if you are human.