NIM, PUB and Cognitive Paternalism

Please allow me to think aloud. Just don't be disappointed if I don't get anywhere . . .
A few posts ago I coined a term: Neutral Institutional Monism (NIM). NIM is the thesis that there are institutions and there are institutions. The primary explanatory distinction is not coercive vs. non-coercive, or state vs. market, but between stable patterns of mutually advantageous interdependent action vs. unstable or mutually harmful patterns. This cuts across the coercive/non-coervice distinction.
Undergirding NIM is what I've called the principle of behavioral uniformity, but which I'll call instead the principle of uniform behavior (PUB), for the sake of a mnemonyic acronym, since PUB is the hard nugget of wisdom at the core of Public Choice theory. PUB says that there is human psychology and then there is human psychology. Whether a person is embedded in a market institution or a state institution will not affect how she will tend to represent her prospects, represent their relative costs, or act on these representations. Psychology is everywhere the same. When behavior differs, it is not due to underlying differences in cognitive or motivational structure, but due to the way alternatives and their prices are represented. People are smart and good when it pays to be smart and good. People are dumb and corrupt when it pays to be dumb and corrupt.
Now, whether “it pays” is a function and consequence of the structure of institutions. Not all institutions are formal. If your community has accepted a norm according to which women who show their ankles will be treated as if they are not there—will be denied any acknowledgement or human contact—then, other things equal, it will not pay for women to show their ankles. This norm is a kind of institution. If we want to go a level deeper, many norms lead to stable patterns of action because they have been fully “internalized,” leading people to impose negative sanctions upon themselves through shame and guilt, even when their norm violations are undetected. Norm enforcement is cheapest when people can't help but punish themselves.*
Anyway, the point I want to get to from NIM and PUB concerns the justification of paternalistic intervention on the basis of the fact that actual human psychology diverges from some ideal or other of rationality. Many of these paternalist arguments suffer from violations of both NIM and PUB.
First, the Fallacy of Asymmetric Idealization. The cognitive paternalist (that's what I call them) is very good about getting into the nitty gritty of actual cognition, its foibles and imperfections, and how the real deal differs from idealizations of rationality. Great! And, then they fail completely to get into the nitty gritty of actual political institutions, their foibles and imperfections, and how they differ from idealizations of collective action. In effect they say this:

In the real world, the mind works like this, and this can lead to all kinds of problems. And in an extremely unrealistic abstract model of government action, its agents can easily and objectively identify problems and act to effectively solve them. So, let's have the ideal government solve the problems of nonideal cognition.

It sounds stupid when you put it that way, doesn't it? The trick is figuring out how to work with real minds, using real governments!
NIM asks, so how are you going to do it? Most real government institutions are at least as kludgey and means-ends inconsistent as real minds are. “Silly, your sock won't open that can of spinach! So try your pillow instead, because a model exists in which pillows are can-openers.”
The NIM “what is this wonderful institution of which you speak” question is heightened by PUB. Psychology is general. So if we've got nonideal psychology, then we've got nonideal psychology, including every single agent of the government. So the cognitive paternalist proposal really depends on a supressed premise that there is some possible institution of government that can solve problems of nonideal psychology, but which is itself unaffected (or only mildy affected) by problems of nonideal psychology.
NIM also suggests that we just slow the hell down before deciding we need a government institution to solve some kind of perceived problem. We might. But we might not. NIM asks us to compare all our possible institutional options.
Consider these combinations with nonideal cognition fixed:
(1) nonideal cognition + ideal markets = no paternalism
(2) nonideal cognition + nonideal markets + ideal goverment = paternalism
(3) nonideal cognition + nonideal markets + nonideal government = ???
Our world, the world of NIM, is (3). Whether nonideal cognition could justify paternalism depends on what our market and government institutions can in fact accomplish.
Let's call an institution that corrects for some nonideal aspect of cognition an ameloriative institution. An ameliorative market is structured so as to raise the prices of nonideal cognition, resulting in less of it. An ameliorative policy is one the government implements to change the prices of nonideal cognition, through coercion, resulting in less of it.
OK. Suppose there is the possibility for an ameliorative market, but the market needs some changes in regulation, legal code, etc., in order to function, and this is something the government has to do. (For example, a Hanson-style futures market in ideas might be ameliorative in some ways, but it cannot exist the US's present regulatory climate.) Now, if the government is of sufficient quality to implement an ameliorative market, then coercive ameliorative policies will be unecessary, or at least less necessary. But if the government is not of sufficient quality to implement ameliorative markets, then it is probably not in a position to apply effectively ameliorative coercion, either.
And . . . that's all I've got for now.