[EM] Fix Philly Districts (Warren Smith)

[Jameson Quinn]

Here's a compactness measure I haven't seen proposed before:

Compose a district boundary out of line segments. The "thickness" of a line
segment is the closest distance of any voter to any point on that line
segment. Minimize the total of the length of segments divided by their
thickness.

It is, of course, not an easy measure to maximize. In particular, small
perturbations of boundaries could give big changes, and the measure would go
infinite if any boundary crossed over a voter. However, I suspect that
"simulated annealing" would do a decent job. But it is easier than travel
time to calculate, and more sensitive to population distribution than
splitline or purely perimeter-based districting. At any rate, you'd probably
have to do it with a contest to find the best answer.

JQ
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20110907/170b14d8/attachment-0004.htm>