A snippet of his argument:
If politicians’ behavior is determined by the vested interests to which they are beholden, economists’ advocacy of policy reforms is bound to fall on deaf ears. The more complete our social science, the more irrelevant our policy analysis.
...By endogenizing politicians’ behavior, political economy disempowers policy analysts. It is as if physicists came up with theories that explained not only natural phenomena, but also determined which bridges and buildings engineers would build. There would then scarcely be any need for engineering schools.
Almost ironically, this stance seems to be driven by his own vested interest. "If vested interests drive politics, there would be no need for the type of courses that I teach. I know that the courses I teach have value, so it cannot be the case that vested interests drive politics!" It cannot be because it must not be.
If we start from the assumption that what economists teach is both true and useful, this is a valid point, but those assumptions are not verified a priori: it is perfectly plausible that what economists say is true but of no use at all in the realm of politics. It would be unfortunate, but possible. It's premature to rule out policy irrelevance.
Of course, as Dr. Rodrik rightly notes, there are many examples of politicians responding to ideas rather than solely to vested interests. I agree that models which solely focus on the venality of elected officials are missing part of the picture. But are ideas alone enough to fill that gap in? He goes on to say:
There are three ways in which ideas shape interests. First, ideas determine how political elites define themselves and the objectives they pursue...This is all true, as far as it goes. But where do the voters factor into this account? If voters are rationally ignorant or, even worse, actively irrational in their policy preferences, the ideas that politicians respond to may not be good ones.
Second, ideas determine political actors’ views about how the world works...
Most important from the perspective of policy analysis, ideas determine the strategies that political actors believe they can pursue.
Suppose that tomorrow there was suddenly a 100% consensus among economists that open immigration would boost national GDP and make the general public better off, and that every politician was made aware of this fact. Would that be enough to overcome the public's anti-foreign bias, which makes open immigration a quick ticket to the unemployment line for elected officials? Probably not.
Tellingly, Dr. Rodrik's examples of policy revolution due to the role of ideas - China and South Korea - are/were not particularly democratic states. So bringing good ideas to the ruling elite can have great results, but only if you can bypass all the terrible ideas brought by the general public, which is not an option available to us living in a Western democracy.
In this respect, the "vested interest" story is more optimistic than the "ideas" story. Even if the steel industry gets its way and extracts some rents from the general public through higher prices, they also care about keeping the overall economy functioning. If politicians respond to ideas, they're probably bad and wrongheaded, which is much more likely to crush our economic output.
Arguably, the best social role for political economists is advising the vested interests in how to rent-seek in the most inoffensive way possible. It's less noble-sounding, but a much more direct route to policy relevance. There is still a role for teaching good economic principles to students, but expecting that to trickle-up into better policy is optimistic in the extreme. Given the choice I'd rather speak to the captains of industry, rent-seekers though they might be.