I spent some time on the ensemble theories, so I thought some final thoughts were needed.
The 2009 Ellis/Smolin paper mentioned above can be found here at "the Weak Anthropic Principle and the Landscape of String Theory". The essence is that a positive (increasing over time) cosmological constant is much less likely, in such a large ensemble using the anthropic principle as a selection effect, all other things being equal, than a similarly negligible negative (decreasing over time) cosmological constant. So one of the few times String Theory's Landscape can be used to make a prediction, it doesn't agree with observation, which does show a positive cosmological constant.
Lee Smolin's book criticizing String Theory is THE TROUBLE WITH PHYSICS and Peter Woit's is NOT EVEN WRONG. Leonard Susskind's THE COSMIC LANDSCAPE vigorously defends the Landscape and anthropic selection for much of the fine-tuning.
Each side claims their solution is better by Occam's Razor. Critics of String Theory says it multiplies entities a quintillionfold, but defenders of String Theory (and many other enemble theories) often say it is simpler in the number of categories needed, thereby satisfying Occam's Razor....for instance, that many-worlds, by eliminating a complicated collapse of possibilities by an observer, is simpler than the Copenhagen Interpretation. String Theory makes similar claims. Proof remains...elusive, though.
Next time: SMOLIN'S SELECTION