Friday, December 02, 2005

Congressional Redistricting

I see that the Texas congressional redistricting of a few years ago is in the news again.

Some time ago I came up with what I believe to be the perfect way to draw congressional districts:

  • Choose, at random, a voting precinct anywhere in the state.

  • Choose, at random, any precinct contiguous to the first. Compute the total population. Choose another one contiguous to the set consisting of the first two. Recompute the total population.

  • Continue as above until the total population reaches or exceeds that specified for congressional districts, within some tolerance. This is District 1.

  • Choose, at random, an as-yet-unchosen voting precinct and repeat the process for District 2. Repeat again for District 3, etc.

  • After all precincts have been assigned to congressional districts, perform an annealing step: using contiguity as the primary cost function and population as a secondary cost function, trade precincts back and forth until all of the districts are contiguous and more or less the same size, population-wise.

A third-year computer-science student could write this as a class
project.

(But: who will bell the cat?)

No comments: