Dr Micah Altman summarizes redistricting technology

Notes from a guest lecture gives a good summary of current redistricting technology and a bunch of issues around its use and adoption.

Dr Altman mentions my software, and points out a couple limitations:

"Ad-hoc definition of compactness"
It's true. I saw a bunch of measures of compactness out there, and no consensus on what the best one was, so I made up my own. It has advantages and tradeoffs compared with other measures. If comparing many plans crossed with many measures became a key feature in driving adoption, I'd do it. At some point we're probably going to have a big discussion about what the right measure is.

"Does not permit inclusion of districting criteria other than compactness, population, contiguity."
Yup. I'll call that a version-1.0 limitation. It's on my to-do list to consider political-unit contiguity, measuring how many towns and cities and counties get broken by district lines. It may also be a necessary compromise to implement Voting Rights Act 'gerrymandering for good' majority-minory disitricts.


Dr. Altman's talk spends most of its second half recapping and advocating the Public Mapping Project's progress so far and future in making Free/OpenSource web based GIS software for crowd-sourced redistricting. The Open Source world has a long standing maxim that 'many eyes make all bugs shallow', and perhaps that will apply also to the bug of gerrymandering.


  1. Hi Brian,

    Just a note, I included your software in the talk because it is impressive in the scale of data it can handle -- redistricting California at the block level is hard. The limitations I referred to are important for understanding the problem automation can solve, and for practical use in a policy setting (in which many criteria may be in plan) -- but not at all unique to your software's approach.



    1. Thanks for the note!
      I still want to find ways to work on the practical and realpolitk aspects where my system comes up short; or maybe find niches in the process where my data can be most useful as is.