There is a section on "Exit and Voice" which discusses this on www.radicalsocialentreps.org/theory as well as a practical entrepreneurial proposal called "Free Cities" here: http://www.radicalsocialentreps.org/theory/free-cities/
Thanks for your thoughtful comment.
Sounds like Bayesian vs. A/B testing. e.g. http://blog.custora.com/2012/05/a-bayesian-approach-to-ab-te... or http://www.mynaweb.com/blog/2011/09/13/myna-vs-ab.html Something that's not repeatable but works in an open system full of people, not experimental objects
Organizations like Groupon, LivingSocial, Airbnb, Kickstarter and (around where I live) The Boston Society of Spontaneity are all encouraging people to get out and make more random, meaningful connections. I look forward to the day when these organizations, ideas like thepoint.com, and the concept of something like a "sanctuary city" [1] are more commonplace and acceptable.
For the example of a society I think a good metric would be: "How many people enjoy living in this particular society and don't want to move away?" We wouldn't need to make any big changes right away because we could start experimenting with what we have already: cities and towns.
How about this: "Cities and towns can create some of their own laws (but only certain types of laws that have been pre-approved by the state and federal courts). These laws will supersede state and federal laws."
I think this would give countries the opportunity to experiment with the effects of theoretically harmless legislation (e.g. cannabis legalization, polyamorous marriages, the invention of new types of corporations) that might make people uncomfortable if they were instantly implemented on a larger scale.
There would probably be a long period of adjustment as cities and states got used to their new relationship, but I think in the end it could bring smaller communities a lot closer together and make places in general a lot more interesting and unique.