Comment Measuring gerrymandering (Score 1) 66
If I had to make an objective measure of whether a district looks prone to gerrymandering, I'd start by applying measures of eccentricity to the border of each district of a state and taking the geometric mean. Some of these measures are ratios of the area of some shape related to but larger and rounder than the border to the interior area of the district. This could be the district's convex hull (convex = 1.0), the quarter perimeter squared (square = 1.0), or the area of a circumscribed circle (circle = 1.0).
Then I'd see how the mean for each state correlates with the actual measure of gerrymandering, which is difference between the two major parties' seats-to-votes ratio in the state's legislature. I'm aware of the Goodhart-Campbell law stating that measures like this tend to become gamed as optimization targets, thereby losing much of their value as measures.