I noticed that the spline ignored populated areas. I think there are advantages and disadvantages in doing that, but I liked the commission route as well. muon's suggestion about crowdsourcing is intriguing, but it has the disadvantage of potentially making itself for sale to the highest bidder, whereas a commission of highly-vetted, thoughtful people might be less susceptible to marketing.
If there's a clearly defined set of measurable criteria, then there's little to worry about marketing. Some deep-pocketed interested party could invest in a plan, only to be bested by some netizens in their parents' basements or a couple of science professors playing with numbers in their spare time.
The splitline people have suggested using a contest to determine the best plan.
The objective of splitline is to minimize the (interior) perimeters of districts. The particular algorithm they use, minimizes the perimeter at each split, but iterative decomposition doesn't necessarily lead to the best global solution.
But crowd-sourcing might lead to better solutions, and the simple algorithm would at least lead to a floor.