Ernie P. wrote:

I think the real question -- and the real problem --  is that if I
remember correctly Joe's proposal requires:
a) objective criteria
b) some voting method for deciding between them

If we have a determinate objective criteria, I don't see why we can't
automate the 'right' solution using explicit calculation or at least
probabilistic techniques.  If indeterminate, then that requires voting
- but how do we get decent voting if we're starting from gerrymandered
districts?

Am I missing something (again)?

Matt responds:

If the objectively measurable optimization goal (such as smallest total road cuts count or perimeters length) with objectively defined constraints (such as per district maximum acceptable population size deviance from ideal uniformity)  defines a NP-complete or NP-hard problem and we cannot expect to find the ultimate solution due to large N then it is sensible to solicit proposals by declaring a competition (maybe offering the winner a cash award).  Ties could be broken by selecting the proposal with the smallest maximum deviation district from population size uniformity and if still tied by the smallest second largest district deviation from district population size uniformity, etc.