It’s always a little worrying when this happens, but sometimes I find myself very much agreeing with at least parts of what Lubos Motl has to say. For example see this recent posting to sci.physics.strings. In it Motl argues that
“I think it is premature to try to construct this major framework that would explain the character of vacuum selection and very early cosmology in string theory”
“So my belief is that we will have to understand the nonperturbative structure of the stringy arena using some new universal definition of string theory – a definition that is both non-perturbative (reaches the strongly coupled regions) as well as background-universal (reaches the geometries and non-geometries that are different from the starting one), and only afterwards, we will be able to start answering the stringy cosmological questions in a better context. Without this new tool, everything is just vague guesswork.
In my opinion, the research of string cosmology; stringy inflation; de Sitter space in string theory; scattering in backgrounds with non-standard causal diagrams; and all similar things that have essentially be started by the observation of accelerating Universe back in 1998 – has led to a very small number of intriguing results. There is almost nothing non-trivial and mathematically intriguing going on here; there is as much output as much input we insert. It remains a combination of phenomenology and speculations where conjectures can rarely be clearly ruled out.”
I’ve never really understood why there are fields of “string phenomenology” and “string cosmology” when the theory is still in a state that it can’t reliably calculate anything. While I think it is wishful thinking to believe that if you understood string theory better it would reproduce the real world, at least Motl’s is a consistent scientific position.
Motl is also a fierce opponent of the “anthropic” arguments that have become popular among string theorists. For the latest example of anthropic argumentation, see this posting at Jacques Distler’s weblog.