When theorists talk about randomization

What happens when an economic theorists/historian talks about randomization? One answer comes from Murat Iyigun when reviewing a new paper by Erik Meyersson on the role of Islamic parties on the education of women in Turkey.

Iyigun brings his knowledge of Turkish history to bear on the issue and makes a convincing case that there were some important institutional changes during the 1990s that make it difficult to argue for a regression discontinuity (RD) design that Meyersson uses. Iyigun makes the claim that

Now this paper is a very fine specimen of its kind, so this is in no way a critique of Erik's work in particular. Instead it is to highlight the pros and cons of the randomization approach to development in general ...

Why Iyigun uses this as a way to attack randomization in general, I can't quite figure out.

Myersson makes the unfortunate statement, which Iyigun latches on to, that his RD design is "as good as random". RD though is not the same as random. It is only a way of assuming an approach that has similar beneficial properties of randomness. If the assumptions behind the RD are disproved, as Iyigun does, then it is not "as good as random".

There are plenty of problems with randomization, but making a straw man out of something unrelated doesn't help the argument.