I think I put this book on my get-at-library list thanks to a post by Ilya Somin at the Volokh Conspiracy blog hosted by Reason.
It's a good book, I'm sure. Only problem is that it was way above my senior dilettante level at many spots in the discussion. Example (page 92, leading off a section titled "Intertemporal Trade-Offs and Time-Discounting Inconsistencies"):
The standard neoclassical model of time discounting is exponential. This means that an outcome at time t is evaluated now (t = 0) as δtu(x), where δ is a constant discount factor and u(x) is an undated utility function defined on outcomes.
Now I almost get that, because I see the t up there in the exponent. But pretty much I have to nod and read on.
So caveat lector, this is a very high-level academic discussion. I swear, I looked at every page.
The authors, Mario Rizzo and Glen Whitman, have undertaken to refute the underpinnings and conclusions of economists and psychologists who wish to influence public policy to encourage the citizenry into better behavior. You know, like your parents did. Specifically, your dad, hence "paternalism". Sometimes it's even dubbed "libertarian paternalism", in order to imply that such policies are "not really" coercive: they're simply designed to encourage you to make better choices, according to your own values and goals. Sin taxes, making retirement saving vehicles opt-out instead of opt-in, government warnings on potentially unhealthy products, are just three examples.
The leading book in this field is the best-selling Nudge by Thaler and Sunstein. (Which I really should read to get at least some of the other side of this discussion.) The authors disdain the marketspeak of "nudge". Why not "push" or "shove"? Or an entirely different metaphor (see the cover): puppet and puppeteer?
The authors note paternalism's underpinnings in neoclassical choice theory, whose mathematical axioms were posited in the 1940s by John von Neumann. They note that those axioms, while plausible, were primarily developed to allow well-behaved "utility functions", which go on to be useful in heavy economic theories. But they were also seized upon to give a rigorous definition of "rationality". Which gave rise to a regular cottage industry of psychologists revealing how people, in various ways, departed from that definition, and were hence labeled as "irrational".
And obviously those "irrational" folks need help from the rational. For their own good.
The authors point out that rationality is much fuzzier in practice than in theory. The thought processes deemed "biased" by the shrinks can often make sense outside the artificial von Neumann axioms. So there may be more holes in the foundations of paternalism than its advocates would admit.
There are a number of other pitfalls: paternalistic public policies are designed by human beings, not von Neumann machines; they are subject to the same biases and fallacies as the folks they want to "nudge". Worse, they have little skin in the game; they have minimal incentives to design optimal policies even if they could.
And they probably can't, for reasons popularized by F. Hayek decades ago: they lack the requisite knowledge, which is highly dynamic, scattered and tacit throughout society.
And finally, there are slippery-slope issues: suppose"libertarian" nudge A "works" (for some criteria of "working"). Won't that encourage policy makers to get even better results via slightly less libertarian nudge B?
Or suppose nudge A does not "work". Given what you know about bureaucrats, do they just give up at that point? Or do they say, "well, the actual problem was that we were too libertarian"?
As I type, I see failures all around of paternalistic policies. Most notably COVID-related ones: a large fraction of citizens aren't vaccinated, despite being incessantly "nudged" to do so. And there's the whole drug prohibition thing… You'd think all those smart nudgers whould have figured out how to get better success by now.
All in all, a fine book which (unfortunately) won't sell as well as Nudge. I can't do justice to all its arguments here.