I response to the recent posts, I continue to assert that the best model has the legislature write the criteria by which districts are drawn, then turn the process over to an independent body to execute the criteria and produce a set of different plans consistent with the criteria. Finally the legislature is given the final decision on which of the independent plans are adopted.
I think that's a very reasonable compromise. However, my biggest concern would be with the criteria being set by the legislature. For example, what would prevent a legislature from requiring that two incumbents cannot be placed in the same district? I think allowing the legislature to set criteria grants too much deference. Deference to the legislature has effectively ruined Florida's redistricting reform.
If you're talking about something closer to Iowa's setup, I would have a very hard time opposing it. As I understand it, election and incumbent data is barred from consideration. I think those are absolute requirements for a fair redistricting scheme.
My suggestion is more like IA than FL. The FL plan was drawn by the legislature under the guidance of their law and the result was not unlike MI which did the same thing starting in the 2000 cycle. It is interesting to note that the MI criteria in 2000 and 2010 were the same ones used from 1970-1990, but back then they were in the hands of an independent mapper due to the divided government at the time. Only when they passed to legislative mappers did the power to gerrymander even with the statutes become apparent.
The issue of whether or not political criteria should be considered is debated and different states creating independent commissions have turned up on different sides of the issue. Most would agree that criteria that are based on individual candidates should not be used. That sentiment isn't as true when one wants to test a plan for partisan fairness or responsiveness to electoral swings through a reasonable number of competitive districts. AZ explicitly includes political data to make those tests, but CA excludes that data. I contend that intelligent mappers and map readers will know the general political impact of a plan, even if the data isn't there to confirm it, so why not put the information out there for all to see.