Here’s a paper by Lee Smolin of the Perimeter Institute (Canada), “A perspective on the landscape problem” (February 16, 2012). Smolin, who credits Richard Dawkins for the inspiration, is a proponent of Darwinian fitness as a way of understanding the origin of universes.
String theory brought the landscape issue into focus but, as we have seen, it was inevitable that as physics progressed we would have encountered the problem of explaining how the universe chose its laws. We can call this the generalized landscape problem. Whether string theory is the right theory of unification or not, it is clear that this general landscape problem must be solved. But as we have seen, this problem can only be solved if we abandon the idea that ultimate explanations in physics are to be given in terms of laws organized according to the Newtonian paradigm, with timeless laws acting on a timeless space of states.
Above all this must apply to whatever theory unifies the different effective theories that make up the landscape. Within string theory the search for this unification has largely proceeded along traditional lines.
Any solution to the landscape problem must then transcend the Newtonian paradigm.
As Wheeler, Dirac and Pierce understood, laws must evolve to be explained. It is likely also that the absolute distinction between laws and states must break down14. Our mandate is then to invent new kinds of theories that answer these challenges, while staying true to the demands that theories make predictions by which they can be falsified. The still open problem of giving string theory or M theory a background independent formulation that would be the setting to resolve the landscape issue should be re-examined in this light. The main lesson which can be drawn from the successes and failures of attempts to resolve the landscape problem surveyed here is that theories which embrace the evolution of laws have a better chance to make falsifiable predictions than do theories which try to hold onto to the notion that law is eternal.