Actually, I should clarify that first part a little.

Let's consider civic governance.

Right now, we have lots of towns, all of which have different laws. People have the right to move between towns whenever they want. Yes, the federal government places many restrictions on what towns can or can't do, but there's still a wide space of possible regulation regimes towns can go with.

How is this different from the situation you're recommending?

To begin with, the towns came into existence through a process less likely to produce optimal design than the process by which proprietary communities came into existence--the point of my piece. Insofar as you accept that argument, the size of proprietary communities gives at least a little evidence on the optimal size for such institutions—imperfect because they are not producing all of the same services. That evidence suggests that even most local governments are way above optimal size. Beyond that, I'm under a local government under a county government under a state government under a national government, with size and costs of moving increasing as you go up that hierarchy.

That said, competition among governments at various levels does limit how bad a job they can do.

And what kind of legal or cultural changes would be necessary to move from here to the situation you're recommending (other than decreasing the amount of federal law so as to increase the space in which town law can vary)?

I'm an anarchist, so the full change required is substantial. Expanding on that requires a book, not, I think, an email conversation. The second edition can be downloaded for free from my web site, and if you are interested I can send you the third edition (or if you like Kindles, it's available in that form from Amazon).

{He ended up doing a friendly but critical review of my book on his blog}

That document was written about four years ago, and some of my views have changed in the interim. I'm not interested in too lengthy a discussion, and my economic knowledge isn't good enough to respond to some of the more technical things you're saying, but just to answer a few of your main points:

I don't disagree with you about the capitalist trucks - I feel like 1.2 makes almost the same point, and I discuss this further in this blog post - http://slatestarcodex.com/2014/06/07/archipelago-and-atomic-communitarianism/ . But a few caveats.

First, this works exactly once, while housing is first being built. That is, if I want to avoid having something bad within a one mile radius, there is no way this is going to be do-able in New York City, which already exists, is already inhabited, and is already divided into tiny mostly-single-building fiefdoms.

Second, even though this should happen in real life it almost never does. I have looked high and low for apartment buildings that promise draconian noise restrictions - let alone ones that promise draconian noise restrictions plus have a third-party certification that they comply - and they just don't exist. That is the absolute simplest test case of this kind of thing.

You are assuming that there are enough people who share that particular taste to make it worth producing such a product. I'm not sure to what extent producing it is made more difficult by restrictions in the legal system on the ability of a landlord to evict tenants—that would depend on where you were looking.

I would argue that even gated communities don't really meet this standard - the active ingredient there seems to be filtering out the usually-more-rowdy poor people, through high land prices. Why aren't there "gated communities" for poor people, where the houses may not be the nicest, but the poor people pay a (small) premium to live somewhere where they know more rules will be enforced and they're not going to have to hang out with drug dealers and grafitti-sprayers all the time? If the answer is "poor people can't pay enough money to support voluntary fiefdom rule enforcement", then this is a big problem with using voluntary fiefdom rule enforcement to maintain quality of life.

Third, the difficulty of implementing this solution scales more than linearly with size. Gated communities of a few hundred homes are pretty easy to fiefdomify. If air pollution is a function of number of factories in a fifty-mile radius, you're going to need individual capitalists to have fifty mile urban fiefdoms, which is stretching credulity.

Yes. As I tried to make clear in the chapter I sent you, I agree that market failure problems mean that a pure market system does a suboptimal job of dealing with some problems. The point of that chapter is that having a government tends to make the problem worse, not better, since the conditions that lead to market failure are a special case in the private market, the norm in the political market. I'm not a utopian. In Machinery I describe situations in which my ideal institutions (a competitive market for law and rights enforcement) could produce unlibertarian law.

I'm not arguing for "fiefdoms," incidentally—my preferred private legal system is non-territorial. I'm pointing out the difference between two somewhat similar institutions, one produced on the private and one on the political market.

Fourth, in practice a lot of house-getting is based on "somewhere cheap enough I can afford, close to the one good job I can get, which is up for sale in the two weeks I have between jobs which is the longest period I can afford not to be working". This isn't really conducive to intentional communities.

It's a whole lot more conducive to producing optimality by market mechanisms than by political mechanisms, where your opportunity for choice is much less. And I think you are underestimating the range of alternatives available to most people, not in the context of a week but of years. Populations do shift.

Compare also to the claim "All US states have pretty good governance, because it is easy to migrate to a state with better governance." I'm not very sure what's wrong with this argument (possibly transaction costs?) but something certainly is.

That depends on how bad you think things would be absent that constraint.

I'm unclear what you mean to do by talking about how "in a simplified model (no externalities, public good problems, etc.) you really can treat the world that way". I agree that the only problem with that model is externalities, public goods, etc, which is why I'm focusing on them.

If you have never studied economics you might find it worth doing. You are already spending at least forty-eight hours a day studying things, judged by output measures, so why not take it up to fifty? I can recommend a book.

My point is that interdependence by itself doesn't prevent the simple libertarian model, in which each person controls his own life, from making sense. Absent economic theory it seems to prevent it, since I depend for what I do on interactions with lots of other people. Prices and property solve that problem because, in first approximation (ignoring the market failure problems), what I sell things for exactly measures their value to those who get them (price=Marginal value), the price I buy them for exactly measures the cost to the producer (price=marginal cost), so I am bearing the net costs of my life on others, receiving the net benefits of my life for others.

Just as if each person was entirely self sufficient, but in a complicated and interdependent world.

Beyond the first approximation the conclusion breaks down, for the sort of reasons you point out. It would in a self sufficient world too--the deer I shoot might be one you would have shot tomorrow. But considered as a mechanism for getting the right things to happen, for giving actors incentives compatible with the general welfare, it comes much closer than the political alternative, where individual actors bear almost none of the costs of their acts and receive almost none of the benefits.

A good deal of your argument consists of pointing out that there might be, probably have been, situations where particular government regulations did net good, which is true. But we don't have the option of only giving governments power to do good. In choosing among possible institutional structures, the question is whether, on net, government power makes us better off or worse off. You offer an a priori argument (market failure) for why the laissez-faire market that libertarians support sometimes produces a suboptimal outcome, one worse than what could be produced by a wise, benevolent, and all powerful philosopher king, which is correct. I offer an a priori argument for why the same logic that implies that also implies that the outcome produced by the political market will typically be farther from the optimal.

The non a priori argument is hard, since it involves comparing alternative realities only one of which is there to be looked at. If I am correct in thinking that a modern stateless society with reasonably good rights protection is practical, obviously a debatable claim, I suggest that all of the sorts of benefits you describe, added up, come to less than the utility cost of immigration restrictions. Maybe even less than the utility cost of the War on Drugs.

I'm sure you're aware that the share of income/wealth held by the bottom 90% is declining precipitiously compared to that held by the top 10%, especially in comparison to past eras. By my understanding this trend accelerated around the time trickle-down economics started.

I don't know at what time you think "trickle-down economics" started, or even what you think it is. I believe the term was invented about eighty years ago by people on the left and attributed to people on the right.

I understand that correlation is not necessarily causation, but you have to admit that as far as empirical results go it's hard to imagine ones that would have supported the "trickle-down doesn't work" hypothesis more conclusively.

I assume the positive hypothesis is that rising incomes for high income people result in higher incomes for other people. You can't reject that by observing that incomes became more unequal. The question is whether incomes in lower parts of the distribution would have been higher or lower if something had prevented the rise in incomes at the higher part, which you have no evidence on.

Let me suggest an alternative explanation of the pattern, the one that Charles Murray has been implying. On his evidence, people in the upper part of the income distribution continue to display roughly the same pattern of behavior that most of the population displayed in the post WWII period. Most people get married. Most stay married for long periods of time. Children are mostly raised by two parents. Most people get jobs, attempt to maintain careers, save money. But in the lower part of the distribution the pattern has changed radically, away from those examples. So we have a functional social structure at the high end, disfunctional at the low end.

I don't know if it's right—for unrelated reasons I don't entirely trust Murray. But it makes more causal sense than "the rich got lots richer, the poor got only a little richer, and that must be because the rich were hogging all the money that otherwise would have gone to the poor," which doesn't have any obvious causal structure at all.

Let me offer a third explanation, based on data. From the end of WWII to the beginning of the War on Poverty, the poverty rate, definition held constant, fell sharply. Since the War on Poverty got fully funded and operating, the poverty rate, definition held constant, has been roughly fixed, going up and down with general economic conditions. That suggests that the expansion of the welfare state had the opposite of its intended purpose. It was supposed to get people out of poverty, to make them self-sustaining. It actually made poverty a little less unpleasant, and so somewhat reduced the pressure to struggle out of it. As Murray describes in Losing Ground, the original purpose proved undoable, so was abandoned in favor of what was doable.

And, for a fourth explanation, having little to do with either side of political controversy, perhaps what happened was that as more and more low skilled jobs became doable by machinery, the premium to high intelligence, education, various related characteristics, rose. Putting claims in terms of wealth, especially if that's limited to things like stocks and bonds and doesn't include housing and pension funds, focuses attention on the very top of the distribution. My guess is that if you look at income, it's not the multi-billionaires who drive the pattern but the doctors and lawyers and techies, of whom there are a lot more. But that's a guess--I haven't looked.

Regarding public vs. private schools, see http://www.nytimes.com/2006/07/15/education/15report.html?_r=0 . I have yet to hear of any study that finds private schools to do significantly better once student characteristics are adjusted out.

My point wasn't about outcomes but about input. I believe your asserted fact, that private schools spend more per pupil than public, is the opposite of the truth. I don't know what the data are on outcomes. As it happens, I'm critical of the whole conventional model of K-12 schooling, for reasons I've discussed on my blog and elsewhere, which is part of why our children were unschooled, first in a small and very unconventional private school and then at home.

Regarding seatbelts and Peltzman effect, I find Levitt's arguments convincing. See http://freakonomics.com/2006/12/09/the-difference-between-theoretically-possible-and-important/

Peltzman analyzed data. His conclusion, for a particular set of mandatory safety requirements, was that they increased the accident rate by about as much as they decreased the deaths per accident, leaving the total death rate about the same.

Whether that's true for seatbelts in particular I don't know. My point was that I suspected the figure you gave was based on the same error that Peltzman pointed out—ignoring the effect on the accident rate.

The final bit by Levitt is an old idea of Gordon Tullock's. He had two versions. One was a dagger set into the center of the steering wheel. The other was to use the same detectors used for air bags, but wired to a hand grenade instead.

And Levitt has the economic implication wrong. The Peltzman effect doesn't imply that one shouldn't wear a seatbelt or that seatbelts don't provide benefits--merely that people take some of the benefit as faster driving and other things they value that make accidents more likely. Tullock wasn't recommending his solution, he was using it as a reductio ad absurdum of the idea of minimizing traffic deaths as the ultimate objective.