I agree with everything in this post. Yet, I find it fascinating how devo economists like David characterize the impact evaluation workflow as being complicated. It's telling of the fact that we are still the pencil and paper academics who are just setting a timid foot in the real world.
It's clear that human society has been able to create hierarchies that handle vastly more complicated processes, where the stakes are much higher. Like: space exploration. Nuclear plants. Armies. I could go on. Heck, roads are tough to build and involve serious principal agent problems (final quality is not fully observable); yet many societies have been able to solve that, even in a market environment (which is not the case of the other examples).
A potential answer to that is that doing business in developing countries is difficult for everyone, which is a fair point.
It does seem quite necessary to recruit more qualified people to manage the intermediate steps of the process and be able to translate the idea that David has into the study that David wants. As a grad student in econ, my sense is that this path (full-time research manager) is considered some sort of failure or at best an intermediate step, because of the contempt of many academics for this kind of work (as a full-time job). Otherwise, there would be career RAs, and that would be great! Higher salaries may help, but so would an environment which gives credit for this kind of work, and recognizes that it is important and challenging. This is not going to happen if we insist on sharply separating people into those with the idea and those who "merely" executed.
Hopefully this is already changing.
One more thing which seems relevant: standardization of methods. And that's already under way (JPAL, IPA, etc.).
Still, compared to other systems in the modern world, running evaluations seems like peanuts! =)