An interesting summary of development economics. The overriding theme seems to be controlling for effects. Randomization can account for some effects but often the results are applied far outside the scope of the trial, and people are intelligent so sometimes the randomization is not effective.
What they achieve in the real world is good. But I can't see the scientific merit. Doesn't the actual work lie in making organisations accept change?
Banerjee and Duflo wrote a great book, Poor Economics, if you're interested in their work on development economics.