I thought this was kind of silly for a few reasons:
1. His "trial & error" idea and the "god complex" idea are distinct; the clumping together of these two ideas seemed to have been intended for attention-grabbing than for actual synthesis. It seemed like the old issue of finding a solution first and then fabricating a problem.
2. I'm pretty sure he's conflated trial-and-error with screening...
3. Despite his attempts to claim that it is not obvious that screening & directed evolution are common tactics in industry & research, I remain unconvinced. Yes, screening works in certain design scenarios, in pharmaceutics, etc. etc. Yet he failed to apply it adequately to situations in which people are not already using this strategy. For instance, he used as example a politician campaigning about health care reform. The fact that we do have countless different health care systems around the world and that they learn from each other is an example of screening. Yet it is not feasible for each area to try out multiple health care systems at once; when one actually has to use trial-and-error, as is the case in institution design, rather than the kind of low-investment screening that he's talking about, rational design becomes much more important. You see this in chemical biology all the time; when you're just screening phages or yeast or easily synthesised small molecules, of course it makes sense to just do a high-throughput screen rather than actually spend time trying to figure out the details, since the rationalisation can happen retrospectively. But if you just have enough resources to do things one at a time, you'd better take the time to make sure the one you're trying at least has a good shot. And the politician is the one trying to convince you that his idea, rather than his opponents, has a better shot.