- one must distinguish the 1PI and the Wilsonian effective actions more carefully: the 1PI action describes the whole field content, and you get the same action regardless of the point around which you expand
- on the other hand, the Wilsonian effective action contains the low-energy degrees of freedom that depend on the point in the configuration space - and this is the type of the effective action that one obtains in string theory
- Tom re-uses his statements that one cannot think about the bubble of another vacuum (such as a de Sitter bubble) as an excitation of the original vacuum, and interprets the Guth-Farhi results in this fashion - one cannot verify the existence of a dS bubble inside the bottle because the external observer sees that it is surrounded by a black hole
I would agree that this "disconnected feature" of the vacua prevents us from making a scientific claim about the probability measure of different ones. He argues that the "democratic" measure on the space of vacua is unjustified, and a more precise measure cannot be defined - and I agree with these statements. On the other hand, if the large number of stringy vacua exist (just imagine that!), we may perhaps be living in either of them, regardless of the existence of cosmological solutions (such as the eternal inflation) that interpolate between them. Of course that I tend to agree that even in this case, the full rules of string theory are more likely to pick some "special" vacua rather than the numerous ones, but we don't have any proof either way. In order to kill the current versions of the landscape idea completely, one would have to find a general enough and serious problem with the construction of the KKLT-like vacua.
Nima Arkani-Hamed, who has been converted to a landscape supporter much like many other people have been converted to Christianity ;-) (but otherwise he studies it much more scientifically than other landscapers!), argues very correctly that the anthropic idea is either colossally correct or colossally wrong. It is a bifurcation point that can direct physics in the next 10 years in vastly different directions, and one cannot decide which answer is correct by cheap attacks.