Is OPRA a Deterministic or Probabilistic Problem?

An almost universal pair of comments I get from readers so far has been a combination of two observations:

1. Wow, there is so much (social-political and mathematical) complexity to Hands-Off Redistricting!

2. Isn’t there a way to have a computer figure out the best Optimal Proximity arrangement?  Why does it have to be crowd-sourced?

(BTW, I wish everyone could see everyone else’s comments- they are uniformly great so far!  Would you all please consider Following the blog and commenting online??)

As to #1: yes there is a tremendous amount to think about and do.  So to avoid brain damage and to keep from being intimidated by the enormity of it all, I’m going to focus on the mathematical aspects of it for now.

As to #2: I don’t know, and I’m not mathematically sophisticated enough to even know how to determine this.  I think I understand some of the basic issues, though.  And I know I have to know the answer to this before any of the political issues can even be considered.

Here is what I think the basic issue on topic #2 is.  I hope that the way I will describe it will so offend someone who knows that he/she is talking about that they will be forced to intervene.

OPRA, as a computational problem, is likely either Deterministic or Probabilistic in nature.  If it is Deterministic, you could feed it the Census Tract boundary, population, and contiguity information, and it would automatically come out with the solution that best satisfies the 5  rules.  If it is Probabilistic in nature, it would mean that computation would employ hunches, heuristics, pseudo-randomized scenarios, etc. to quickly arrive at very good scenarios, but never be able to “prove” that any scenario was the best solution.  There’s nothing at all wrong with Probabilistic problem solving.  It is used extensively, for example, in computer graphics and is foundational to much of our current statistical analysis.  Nate Silver‘s celebrated success predicting the 2012 elections hinged on his extensive use of probabilistic scenario generation.

There are three interesting inter-related issues here:

  • OPRA may or may not be the “best” way to implement Hands-Off Redistricting
  • The political approach and acceptability of Hands-Off Redistricting will change, depending on whether it is Deterministic or Probabilistic, and therefore
  • Whether OPRA — or any alternative approach — is Deterministic or Probabilistic may at least partially determine whether it is the “best” way to implement Hands-Off Redistricting!

I shouldn’t over-emphasize the third point.  The “best” approach to de-politicizing district boundary determination should win.  But I think the second point will be important down the road.

If OPRA is Deterministic in nature, by the time anyone is going to consider its being implemented, there will likely already be a solution computed and a map to look at.  OPRA won’t seem like an approach, it will seem like a proposed redistricting map, and the reasonable question anyone is going to ask is whether the OPRA map is better or worse than the current state of affairs or other competing hand-drawn solutions.  If OPRA is Probabilistic in nature, it may be that through several preliminary runs there appears to be an OPRA Map emerging, but nobody will really be sure that this is what the map will look like once it is opened up to crowd sourcing.

In either case, the argument is going to have to be made that a fair Hands-Off approach is just superior to any manual, politicized boundary drawing by its very nature.  But the path seems clearer when it is known whether OPRA is in the Deterministic bucket or not.

Can anyone reading this help, or have an opinion on the mathematical aspects of this?  If you are knowledgeable on the topic please comment, or if you know someone who is likely to know, please send them a link to the blog and ask them to chime in.

This entry was posted in Boundary mathematics, Politics of Redistricting and tagged , , , . Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *