Syndicate content

Add new comment

Very interesting and useful exposition.
However, even your final conclusion is overly optimistic. CCT and UCTs do do not exist on a smooth continuum, they existing a multi-dimensional design space, some of which is likely to be rugged tather than smooth i.e where small changes may make big performance differences. This point was made by Prichett et al here last year:
Given this, the interesting question then is how to effeciently explore this design space?