I should say first, that I have found Hawking's earlier work to be useful (ie: Hawking-Unruh radiation). However, I do not agree with the attitude to science represented by his recent attempt, with colleagues Hartle and Hertog, to make 11-dimensional string theory compatible with a very abstract version of reality, by adding even further complexity to string theory (they add a so-called Escher space). See: arxiv:1205.3807v2 (Accelerated expansion from negative Lambda, Hartle et al., 2012).
To make my point I can start with the first statement they make in their abstract: "Wave functions specifying a quantum state of the universe must satisfy the constraints of general relativity". The implication is that general relativity (GR) is the truth, and everything else must be measured against it. Well, whether or not GR is the final truth, and I am pretty sure it is not, what should have been said is something like: theories specifying the state of the universe must satisfy the constraints of observational data. This seems like nit-picking, but is not. It is fundamental that the data must always come before the theories.
GR only agrees with galaxy rotation data if, typically, 10 times as much matter as can be seen is added ad-hoc in an unphysical way to the galaxies to make it fit. This seems to me to be an example of theorists adding complexity to a theory (adding dark matter and the new physics needed for it) in a desperate attempt to save it (like Ptolemy's epicycles were invented to help geocentric theories fit the observed planetary ephemerides). Sure, the idea of dark matter worked in the case of Neptune, whose existence was postulated to fit the motion of Uranus to Newtonian gravity, but this was the addition of a small amount of normal matter in the plausible shape of a planet, whereas dark matter needs the addition of 10 times as much matter as is seen, in a new form and with new physics to go with it to explain its bizarre halo-like configuration.
The physicist's focus should always be on the data and not the theory, which is why I am so keen to critise the first sentence of Hartle et al.. In contrast, MiHsC is not a finished theory yet, but I have developed it from the bottom up, by looking at messy anomalous observations and disregarding most of the existing top-level theories. As a result, MiHsC is simple and can explain cosmic acceleration, dwarf and disc galaxy rotation (I've just submitted a paper on this), the flyby anomalies, the Pioneer anomaly (although so can Turyshev et al.'s complex thermal model), and the Tajmar effect (albeit unrepeated experimentally), without the need for adjustable parameters, extra dimensions, or Escher space.