Jason Kuznicki | Free Thoughts Blog

This is the fourth and last in a series of posts on the socialist calculation debate. Here are parts one, two, and three.

In previous installments we looked at some of the problems that one might face in trying to plan an entire economy mathematically, with reference to a set of price-optimizing equations.

Let’s grant though that we could solve all of these problems—we have computers to solve the equations; we can generate a good enough approximation of the set of equations itself; and we can solve the problem of pricing capital goods.

Still, the planners don’t have the data they need about consumer preferences to put into the equations. Recall that in part III, Soviet economist Leonid Kantorovich simply took as a given the menu of consumer goods and the quantities that had to be supplied for each. He then developed optimization methods that would allocate labor efficiently. Assuming, again, that the list of goods was correct.

As it turns out, that’s a very big problem.

IV. Consumers’ Preferences

F. A. Hayek’s distinctive contribution to the socialist calculation debate, over and above that of Ludwig von Mises,[1] was essentially to ask where the data might be found to do economic calculation about consumers’ preferences themselves. (And by extension, producers’ preferences.)

Who has that data? The simplest answer is that everyone has a tiny little piece of it. It’s dispersed among all of us, because each person has a list of consumption goods that they may desire to varying degrees in varying circumstances, as various needs and opportunities arise.

Whenever the needs and opportunities align just so, a given consumer good rises to the top of our list. And we act to acquire it. When we do, prices will emerge.

Yet prices only tell us a part of the story — they tell us that buyers and sellers were able to agree at a given time and place. Often, prices can signal an opportunity, as when a supplier discovers that he or she can undercut the current market price and realize a profit.

But prices in themselves say nothing about other possible agreements that might have arisen in other circumstances. They also say very little about the future: As soon as a price occurs, it’s history. The entrepreneur who discovers that he or she can undercut the current market price still has to get to market. That takes time, and when they get there, the market price may have changed. That’s a risk that entrepreneurs have to take. Consumer demand is fickle and inherently hard to predict.

It’s also very likely that you can’t articulate beforehand just what your list of preferred consumer goods really looks like, including how much you would buy of various goods at various prices. No one really can.

Your list also probably changes very rapidly. Every unexpected event alters your preference set to some degree. Every disaster, every unexpected discovery, every windfall, and every loss. Every new product you didn’t know about before. Every old product that disappears from the market. Every single change upsets everything — all in a hierarchy of values that you can’t even begin to articulate in the first place. 

Weirder still: There are items on your “list” that you don’t know about and never will.

It sounds very strange to put it that way, I know. But we have all had an experience that demonstrates it. We’ve all at one time or another walked into a store, discovered a product, and then bought it — all while having known nothing about it in advance.

Here we must start speaking of a “list” of consumer demands only for want of a better term. If it were to be drawn up comprehensively, such a list would include an infinite number of products, past and present, that aren’t generally available on the market, that are unknown to any of us, and that are therefore of unknown subjective value. Until the opportunity arises, and we act, and only then do potential entrepreneurs get a glimpse into consumer demand.

Much like the concept of a “good” — covered in part two — the concept of a consumer demand hierarchy is a conceptual crutch. It’s not a real thing at all. Markets reveal preferences, but in a sense they also make preferences, because consumers choose only among those options that are available to them, and because we can only speak then, in retrospect, about their having acted as if there were a hierarchy of wants.

Consumers have preferences, no question. They act on those preferences. But can they articulate them? Not in the way that we would need to do planning.

As a result, socialist calculation can’t ever really get off the ground. At least not without some kind of incredibly reliable and thus very probably dystopian brain scanning technology. While we’re at it, we’d need a complete knowledge of all upcoming natural disasters, technological changes, fads, and cultural phenomena that will arrive in the near and distant future. This though is an impossibility — if we knew what technologies or cultural developments the future held, we would have them right now, and they wouldn’t be “future” developments at all.

But without them, we can’t reliably model of consumer preferences over time. And without that, we can’t predict the value of capital goods over time either.

V. And Equilibrium, Too.

Recall that socialist calculation proposed to head straight for the Walrasian general equilibrium — a state of efficiency that markets have never actually reached. Even if socialism only got part of the way there, it might still be better than a market. Right?

Well, yes. It might be, although that’s not altogether clear from the outset.

It’s illuminating to consider now why markets don’t reach equilibrium.[2] Scientific socialists argued that markets were inherently inefficient — all this groping about, trying to find the right prices and quantities by trial and error, and so much attendant waste. In this they were certainly correct, given the assumptions they were using.

There’s another answer, though, and a much more complete one. As Hayek writes:

[I]n order that all [the plans of various people in an economy] be carried out, it is necessary for them to be based on the expectation of the same set of external events, since, if different people were to base their plans on conflicting expectations, no set of external events could make the execution of all these plans possible. And, second, in a society based on exchange their plans will to a considerable extent provide for actions which require corresponding actions on the part of other individuals. This means that the plans of different individuals must in a special sense be compatible if it is to be even conceivable that they should be able to carry all of them out. Or, to put the same thing in different words, since some of the data on which any one person will base his plans will be the expectation that other people will act in a particular way, it is essential for the compatibility of the different plans that the plans of the one contain exactly those actions which form the data for the plans of the other.

In the traditional treatment of equilibrium analysis part of this difficulty is apparently avoided by the assumption that the data, in the form of demand schedules representing individual tastes and technical facts, are equally given to all individuals and that their acting on the same premises will somehow lead to their plans becoming adapted to each other. That this does not really overcome the difficulty created by the fact that one person’s actions are the other person’s data, and that it involves to some degree circular reasoning, has often been pointed out. What, however, seems so far to have escaped notice is that this whole procedure involves a confusion of a much more general character, of which the point just mentioned is merely a special instance, and which is due to an equivocation of the term “datum.” The data which here are supposed to be objective facts and the same for all people are evidently no longer the same thing as the data which formed the starting-point for the tautological transformations of the Pure Logic of Choice. There “data” meant those facts, and only those facts, which were present in the mind of the acting person, and only this subjective interpretation of the term “datum” made those propositions necessary truths. “Datum” meant given, known, to the person under consideration. But in the transition from the analysis of the action of an individual to the analysis of the situation in a society the concept has undergone an insidious change of meaning. (F.A. Hayek, “Economics and Knowledge,” in Individualism and Economic Order.)

In short, I need to know all of what I know, as well as all of what you know. But I don’t even know all of what I know: My brain’s not big enough to hold and analyze a copy of itself. No one’s is. And much economic activity aims, ultimately, not at acquiring wealth or goods, but at acquiring knowledge about what people might want.

As a result, Austrian economists in particular reject economic equilibrium as a yardstick with which to measure the real world. Still, though, equilibrium gives us an idea of how economic interests attempt to interact with one another, and so — up to that point — it’s useful. But the ways in which we fall short of equilibrium are much more interesting: They are both the reasons for economic action and also the proper study of economists.

VI. Beyond the Market

As we’ve seen, a social order that proposes to surpass the market will have to accomplish many things. First, it will have to solve the problem of economic knowledge outlined above as well or better than existing economies, or perhaps better than what an ideal free-market economy might do.

But if we value individual autonomy, we may also have to take steps to preserve it. It might be found, for example, that we could create autonomous economic production agents — hyperintelligent robots, say — that could go around brain-scanning people and then making stuff and distributing it according to rules of utility maximization.

This process might work better than markets. But we might feel less than okay about it when the robots come and level the house we thought we owned. “It’s more efficient to build a factory here,” they say, “and your compensation will take the form not of a cash payment, but of the greater overall utility to be found in our system.”

If you prefer, the problem can be stated in cartoon form.

The system of private property certainly doesn’t protect individual autonomy perfectly. People routinely experience degrading conditions under every economic system mankind has ever implemented. Still, security of property title does a great deal of good work in this area that we might not want to abandon even if it did realize greater efficiency.

It also seems unlikely that trans-market social orders will be able to dispense with the principle of specialization and gains from exchange. As long as individuals and regions have varying production capacities of different goods and services, comparative advantage will be a real thing, and reassignment of goods — trade, in our world — will be a necessity for maximizing welfare. Perhaps in the technologically unimaginable future that reassignment won’t commonly happen through voluntary exchange at market prices, but it will have to happen somehow.

As a blueprint for the future that’s not much to go on, I know. It doesn’t point to scientific socialism, or to much of anything else. Future social orders may end up being very different from our own, but I don’t think it likely that they will move beyond markets in one form or another. And sure, they may tell themselves they have surpassed the market, but we know what that looks like.

Notes

[1] It’s wrong to claim, as some have done, that Ludwig von Mises discounted or neglected the problem of incentives in socialist societies. Although focused most of his attention on certain aspects of the calculation problem, in the very same article that began this series we also read:

The problem of responsibility and initiative in socialist enterprises is closely connected with that of economic calculation. It is now universally agreed that the exclusion of free initiative and individual responsibility, on which the successes of private enterprise depend, constitutes the most serious menace to socialist economic organization… Since we are in a position to survey decades of State and socialist endeavor, it is now generally recognized that there is no internal pressure to reform and improvement of production in socialist undertakings, that they cannot be adjusted to the changing conditions of demand, and that in a word they are a dead limb in the economic organism.

As I so often find with Mises, that’s very prescient for a guy writing in the 1920s.

[2] Given that the concept of “equilibrium” depends on the concept of “goods,” a concept that we have already shown to face severe limits, what follows may be true for more than one reason.

This is the third in a series of posts on the socialist calculation debate. Parts one and two appeared earlier.

Many early socialists denied that their utopia even needed economic calculation: The habits and morals of mankind would simply change, they declared, and each of us would come to feel in our very souls what we should do, in whatever economic circumstances might arise. That was heady stuff, but not at all practical. As a result, some of the more hard-nosed socialist theorists turned to calculation. But it should be remembered that the theory of economic calculation always far outstripped the practice.

Even in the Soviet Union, and even despite talented economists like Leonid Kantorovich, mathematical planning was rarely more than window dressing on what amounted to an elaborate, politically driven wish list.

Planners bickered, formed factions, falsified, cheated, stole, and – when all else failed – they allegedly lifted consumer prices directly from the Sears Catalog. But they didn’t calculate, whether in money prices, or shadow prices, or hours of labor, or anything else. In some ways this is a stronger indictment of the Soviet system than even the existence of the gulag: It shows the Soviets weren’t eating their own dog food.

One may wonder, then, how much of the socialist calculation debate amounts to the good guys being completely gulled by the other side. When real socialists do not calculate, how can we call it “socialist” calculation?[1]

As I’ve stressed before, though, we study socialist calculation because it is a kind of outline or a shadow to the market economy. Socialist calculation attempted to obtain consciously all of those things that markets tend toward through unplanned human action, through the so-called invisible hand. Examining socialist calculation makes the actions of the invisible hand more visible.

One such action is finding the value of capital goods. Ludwig von Mises was the first to stress the peculiar difficulty of this task under socialism.

Mises noted that consumption goods would be distributed in a planned society according to whatever criteria its leaders thought proper: presumably these criteria would be very egalitarian, perhaps with special consideration to individuals’ needs, although in practice a planned society could use whatever criteria its leaders wished.

Yet these distributions, whatever they were, would have great difficulty taking into account consumers’ varied and ever-changing preferences: We can’t give everyone a set of dentures and expect them all to be made equally happy.

Exchange would probably arise, regulated or not, with money or not, and it would likely be beneficial even in a planned society – but only for consumption goods. Capital goods, remember, would never be allowed to go to market. They’re owned collectively.

But how does one determine which capital goods go to what use? Whenever possible, we will want them to produce the consumer goods that have the highest consumption values, of course. But which kinds of capital goods should be preferred, and how are they to be used, and when? And how do we know when to stop using them here and start using them there? When a raw material goes through several stages of production, with choices of different later end products and/or intermediate processes, how do we know where to direct the intermediate goods – if not with price signals from the (sadly non-existent) capital goods markets?

Mises writes:

One may anticipate the nature of the future socialist society. There will be hundreds and thousands of factories in operation. Very few of these will be producing wares ready for use; in the majority of cases what will be manufactured will be unfinished goods and production goods. All these concerns will be interrelated. Every good will go through a whole series of stages before it is ready for use. In the ceaseless toil and moil of this process, however, the administration will be without any means of testing their bearings. It will never be able to determine whether a given good has not been kept for a superfluous length of time in the necessary processes of production, or whether work and material have not been wasted in its completion. How will it be able to decide whether this or that method of production is the more profitable? At best it will only be able to compare the quality and quantity of the consumable end product produced, but will in the rarest cases be in a position to compare the expenses entailed in production. It will know, or think it knows, the ends to be achieved by economic organization, and will have to regulate its activities accordingly, i.e. it will have to attain those ends with the least expense. It will have to make its computations with a view to finding the cheapest way. This computation will naturally have to be a value computation. It is eminently clear, and requires no further proof, that it cannot be of a technical character, and that it cannot be based upon the objective use value of goods and services.

Because capital is heterogenous—better at some things and worse at others—a mere reference to the exchange values of consumption goods is not enough to tell us where different capital goods should go.

The Soviets did make some theoretical efforts here. In particular, Leonid Kantorovich proposed mathematical techniques for optimizing the allocation of labor and capital goods to best fulfill a given set of production quotas. Inputs would be “priced” in terms of the gained or lost amount of product that would be had in pursuing one course of production rather than another, in a process known as shadow pricing.

His approach, though, is inapplicable to an economy for two reasons that have nothing to do with math. (Mathematicians, who understand such things better than I do, agree that his math is just fine as far as it goes, and it’s now an important part of many different optimization techniques in engineering.)

First, Kantorovich held that capital goods’ value was to be determined by the value of the labor that they saved, and by no other criteria. Like all good communists, Kantorovich was ultimately committed to the labor theory of value, at the very least for determining the efficiency of capital goods. (But you can’t please everyone: That he did not use the labor theory of value in directly formulating all of his shadow prices rendered Kantorovich a dangerous heretic to some Soviet commentators!)

Still, as Mises and many others have argued, the labor theory of value simply doesn’t work. One hour’s labor varies greatly in its productive value, not just as a function of the capital goods that back it up – a problem we could solve by looking recursively at the labor required to make the capital goods that amplify a laborer’s productivity – but also as a function of skill, serendipity, workers’ enjoyment or lack thereof in labor, and many other factors.

Nor does labor find anything like an egalitarian reward in the eyes of consumers. One may spend many hours in very difficult labor, all to no valuable end. Or one many stumble onto a million-dollar discovery in a moment of idle reflection over a beer. There just isn’t any correlation between labor and subjective value. In a slightly more just world, there might be. But not in ours. This is also why mainstream economics has rejected the labor theory of value.

Second, Kantorovich’s method was limited by the fact that it took the schedule of consumer goods to be produced as if it were a given. Which in Soviet society, it was. Given how? Given by the political and economic authorities. And not, in other words, by actual consumer demand.

And yet the search for what consumers really want is one of the biggest parts of the information discovery process performed by the market. One may very well optimize for labor under a given set of output constraints. But the discovery of the proper constraints themselves – that is, the discovery of consumers’ real demand schedules – is a process that, in a free market, unfolds in coordination with the allocation of labor and capital goods. At best, Kantorovich solved half the problem, but only by pretending that the other half didn’t exist.

As one reviewer put it:

First and most important, the planner takes as given by “the political and economic” authority and specified social needs the aims of society in the form of final demands. The shadow prices and standard effectiveness are not “regulators” of the economy but only means to the effective realization of goals… In general, the planners start and end with the final demands determined outside the planning system.[2]

Emphasis added. And yet it is the very function of an economy as we now understand it to discover such goals, not to presuppose them or to conjure them out of thin air. Further, as we will see in the next post, there are never really any “final” goals at all, only provisional ones.

Socialism is not actually the reason why the socialist calculation debate is important.

What we term socialism today is only a family relation to the thing that the socialist calculation debate addressed, and I’m not all that interested in refuting (yet again!) a form of social organization that has only properly speaking existed on paper–namely, a calculating socialism that earnestly tries to solve the optimization problem.

In real-world socialist economies, and excluding those that aspire to full communism, the state segment of the economy is perhaps best thought of as a very large firm that enjoys–and suffers from–a number of exceptional privileges. That firm’s size causes it many difficulties. Its monopolization of certain markets brings the usual consequences of monopoly. And so on.

These are related to, but different from, the problems faced by a calculating socialism. The Austrian School finds the calculation debate interesting not primarily for its applications to real-world socialism, but because the debate helps explain what markets do, and what any social system must also do if it proposes to do better.

Consider what was to me the most arresting passage of this essay by Cosma Shalizi. Though very long, I encourage you all to read it:

That planning is not a viable alternative to capitalism (as opposed to a tool within it) should disturb even capitalism’s most ardent partisans. It means that their system faces no competition, nor even any plausible threat of competition. Those partisans themselves should be able to say what will happen then: the masters of the system will be tempted, and more than tempted, to claim more and more of what it produces as monopoly rents. This does not end happily.[1]

The possibility of a mathematically modeled socialism may be one of the closest things that the mixed-market economy has had to a competitor. It’s a paltry competitor, but even it tells us a lot. Which brings us to the second key facet of the socialist calculation debate.

II. What’s a Good?

As I mentioned in part I, finding equilibrium prices in Walras’s model requires an additional simultaneous equation for each additional good in the market. Supply and demand schedules for all goods are connected together in one way or another, and economists want to optimize for all of them at once. But how many goods are there, exactly?

This is not merely a technical question–not, in other words, just one of counting the different things for sale in all the stores. It is also to a high degree a question of subjective judgment, and it reveals some of the conceptual simplifications that make economic modeling both very powerful and yet distinctly limited.

Is a car a good? Yes, of course. But are all cars alike, such that we should think of them as only one type of good? Surely not. How many different types of cars shall we consider? Before, that is, we say “Ehh, close enough”?

Is a car in Florida the same good as a car in Alaska? Should it be considered as such? As a technical matter, and in modern capitalist economies, cars are often made to different specifications depending on the weather conditions they will likely meet. Pianos and lots of other goods are made this way too. And that’s just on the consumer side–capital goods likewise must be adapted to their environments.

Even a gram of gold in New York is a different good, in a certain sense, from a gram of gold in London. Economists conceptualize a “good” as being part of a uniform class of things, but goods are always differentiated somehow, even if it’s only by position. So when an economist solves an optimization problem for all of the goods in an economy, he or she has simplified tremendously–not only by creating fixed, a priori categories of goods, but also by ignoring the problem of their distribution.

(Shalizi concludes, by the way, that even the purely mathematical part of the socialist calculation problem is unworkable: Once we consider transportation, the difficulty goes from supercomputer to supernatural. That may or may not be so. I’m not a computer engineer, only a historian of ideas. But either way, the claim “this math problem will never ever be crackable” is not one I’d like to hang a social theory on!)

And there’s more. That’s because it’s always possible that individual actors will find themselves in new circumstances, or simply have a new idea, and a previously unappreciated difference within a lumped-together category of goods will suddenly open up. When that happens, something very interesting takes place–what we had all along conceptualized as one good now turns out to have been two. And all of the equations must now be redone.

This sounds weird, but it’s actually quite ordinary. It’s called “discovery” or “innovation.” As F.A. Hayek put it:

That the price-fixing process will be confined to establishing uniform prices for classes of goods and that therefore distinctions based on the special circumstances of time, place, and quality will find no expression in prices is probably obvious. Without some such simplification, the number of different commodities for which separate prices would have to be fixed would be practically infinite. This means, however, that the managers of production will have no inducement, and even no real possibility, to make use of special opportunities, special bargains, and all the little advantages offered by their special local conditions, since all these things could not enter into their calculations… [I]t would never be practicable to incur extra costs to remedy a sudden scarcity quickly, since a local or temporary scarcity could not affect prices until the official machinery had acted.

Although it disrupts the process of central planning, this type of local, spontaneous repurposing is still something we probably want to have in an economy. Sometimes increasing our well-being means not approaching a given economic equilibrium, but rather abandoning it in favor of a better one. And yet such opportunities for innovation are never knowable ex ante. They can never be subject to a plan.

What does this mean for economics? In the final analysis, the supply of a “good” is not really a class of things at all. It’s a conceptual crutch that allows economists to treat many ultimately very different things as if they were all alike. Why do they do that? Because that way, they can tell some very important stories about the economy–things like the supply and demand curves of Economics 101.

Those things are important as a first step to understanding what people do when they produce, exchange, and consume. But they aren’t the whole story. They are rather the backdrop before which the story takes place.

The socialist calculation debate is an important episode in the history of economic thought in part because it revealed some of the previously underappreciated things that markets accomplish. It did so by supposing the existence of a socialist planning agency that would have to do all of those things instead - a thing that many socialists of course wanted, but that supporters of the free market have also found very useful to examine. It turned out that the planning agency would face a large number of previously unanticipated problems. Many of them now appear insoluble.

Not that we haven’t tried. On the contrary, lots of work has been done here. That work illuminates what markets actually accomplish (not always what we thought they did!), what economic science can know about them (not as much as we’d like!), and what institutions, if any, might ever be able to improve on them.[1] There are many facets to the socialist calculation debate, and in what follows I’ll talk about only some of the more relevant ones.

One problem in talking about this fascinating episode in the history of economic thought is that each of its lines of argument is often conflated with the others, and, at times, an author raising objections to the prospect of socialist calculation will veer from one problem to another with little in the way of transition. I’ve tried hard to sort out the threads as I see them, but it’s very possible that someone more well-read in Austrian economics will come along and disagree.[2]

Anyway, let’s begin looking at objections.

I. The Math Is Too Hard.

This is by far the weakest objection to the prospect of socialist calculation. It’s also one that I still seem to find in all kinds of places, and still offered as if it mattered. It doesn’t, and it’s not a terribly important objection anymore, but it’s necessary to go through it to get to the interesting stuff.

By the early 20th century, economists had shown that in theory, a centrally directed socialist allocation of resources could match that of a market economy that was in a Walrasian general equilibrium.

What’s a Walrasian general equilibrium? The economist Léon Walras had earlier shown mathematically that, under certain not obviously problematic assumptions it was possible for the markets in many different goods to clear simultaneously – that is, a multi-good market could theoretically adjust all of its prices so that supply and demand were in balance for all goods at the same time. Producers could theoretically make the exact amounts needed to fill consumer demand, and consumers who were willing to pay the market price could always get what they wanted. Most importantly, Walras showed that nothing about the simultaneous existence of many different goods in his model would get the way of the process, and the existence money wouldn’t either.[3]

General equilibrium has obvious appeal, and yet real-world economies never seem to reach it. Walras himself emphasized that markets only approximated his theoretical equilibrium through a process he called tâtonnement–literally, “groping.” And they were therefore not terribly efficient when compared to the mathematical ideal. Which I think is true.

Scientific socialists, in part inspired by Enrico Barone, proposed to do better.[4] Solving the right system of equations would in theory give planners the optimal allocations of all capital and consumption goods, allowing them to reach Walrasian equilibrium directly, even as markets could only grope about in the dark.

In the early twentieth century, theoretical economists – even folks like Lionel Robbins and F.A. Hayek – conceded that all of the above was true. But, they said in part, the math was just too hard. It’ll take forever to solve all those equations.[5]

But forever is a very long time, and today we have computers that can most certainly crack problems like this one. Even, as some have claimed, for an economy with millions of goods.

In 1993, Allin Cottrell and W. Paul Cockshott examined the calculations needed to very closely approximate market-clearing prices for all goods in the Soviet economy circa 1983. For this economy of around 10 million goods, and using only a commercially available supercomputer of mid–1980s vintage, they determined that market-clearing prices, denominated in hours of unskilled labor, could be roughly arrived at in just 17 minutes.

That’s a pretty impressive result. So impressive, in fact, that one gets the feeling that something else must almost certainly be going on here.

And indeed, something else is going on here, but what it is will have to wait for the next post in the series.

Matt Zwolinski says we should scrap the non-aggression principle: It gives easy answers in some places, but it stumbles in others. Better to just replace it, he argues:

[T]he NAP’s plausibility is superficial. It is, of course, common sense to think that aggression is a bad thing. But it is far from common sense to think that its badness is absolute, such that the wrongness of aggression always trumps any other possible consideration of justice or political morality. There is a vast difference between a strong but defeasible presumption against the justice of aggression, and an absolute, universal prohibition. As Bryan Caplan has said, if you can’t think of counterexamples to the latter, you’re not trying hard enough.

My sense is that both Zwolinski and Murray Rothbard, the NAP’s key exponent, may be asking too much of their models. Rothbard for his part doesn’t seem to notice when the model isn’t delivering. That’s why he gives what are in my view some unconscionable recommendations about children. Zwolinski is right to reject them:

It’s one thing to say that aggression against others is wrong. It’s quite another to say that it’s the only thing that’s wrong – or the only wrong that is properly subject to prevention or rectification by force. But taken to its consistent extreme, as Murray Rothbard took it, the NAP implies that there is nothing wrong with allowing your three year-old son to starve to death, so long as you do not forcibly prevent him from obtaining food on his own. Or, at least, it implies that it would be wrong for others to, say, trespass on your property in order to give the child you’re deliberately starving a piece of bread. This, I think, is a fairly devastating reductio of the view that positive duties may never be coercively enforced. That it was Rothbard himself who presented the reductio, without, apparently, realizing the absurdity into which he had walked, rather boggles the mind.

Partly as a result, Zwolinski seems to think we need to give up on the NAP as well. First he recommends a “defeasible presumption”—probably the right approach, I’d say—but in his conclusion, he seems to aim at something much larger. Though not necessarily better:

Libertarians are ingenious folk. And I have no doubt that, given sufficient time, they can think up a host of ways to tweak, tinker, and contextualize the NAP in a way that makes some progress in dealing with the problems I have raised in this essay. But there comes a point where adding another layer of epicycles to one’s theory seems no longer to be the best way to proceed. There comes a point where what you need is not another refinement to the definition of “aggression” but a radical paradigm shift in which we put aside the idea that non-aggression is the sole, immovable center of the moral universe. Libertarianism needs its own Copernican Revolution.

We are ingenious folk! But it’s not clear we need anything as clever, or large, as a Copernican Revolution.

Since we’re on the subject, let’s talk physics. To model how a billiard ball rolls on a table, one makes simplifying assumptions—the table is a perfectly flat plane, the ball is a perfect sphere, the mass is distributed equally, and the like. One does these things not because the simplified model is perfectly true to life, but for three related reasons: (1) the model is often close enough; (2) the math is vastly easier; and (3) you can use it to say things that are more or less true about billiard balls anywhere: here, in Spokane, and—with only minor adjustments—on Mars.

Moral and political philosophy should be like that. They should make simplifying assumptions. They must, if they are to do anything more than reference isolated cases without any extensive explanatory power. And the ability to extend to additional cases is the very reason we do philosophy, at least as a practical matter.

This is why Michael Huemer’s treatment of the non-aggression axiom seems like a significant step forward to me. To Huemer, the non-aggression principle is not really an axiom at all; it’s a strong but rebuttable presumption. That may seem like a weakening to a lot of folks, but if so they should consider the radicalism of Huemer’s conclusions as evidence that he’s not going weak in the knees.

By contrast, I have to say Zwolinski’s objections sound as if he is indicting libertarianism for saying that the billiard ball was a sphere, while only he knows, sadly, that spheres have nothing to teach us about billiards. Granted, the first was false, but so is the second.

Spheres are a good solid start to the problem, and we ought not to feel too badly about needing refinements from time to time. The word for these refinements is not epicycles, however. It’s engineering.

I think this analogy points the way toward solving many of the conundrums Matt poses in his post. Many of his difficulties, including children, pollution, and the need for a clear definition of property, are all a part of building models that meet particular real-world circumstances. Of course abstract theory has much to tell us. And of course it can’t tell us everything.

In one case, though, I will commend to him a useful solution that doesn’t fit the above. His fourth conundrum holds that fraud is unrelated to force, and that a libertarian society might find itself embarrassingly unable to punish fraud. I disagree.

Fraud is wrong because it, like physical force, defeats the will without convincing the intellect. As such:

[Force and fraud] are both impediments to reflective self-rule, and any political theory that values reflective self-rule is permitted to find them similar problems. This is why both are properly censured in libertarian political thought… While there remain many salient differences between force and fraud, the likeness is not imaginary or arbitrary, at least not any more so than are concepts like “will” and “intellect.”

It’s okay to use force to combat fraud, just as it’s okay to use force to combat a physical aggressor. It may not even be too much to say that both force and fraud are species of some larger genus, albeit one that lacks a name in our language. Both acts attempt to subvert individual atuonomy, and both may therefore be treated alike, at least if we do value individual autonomy as our highest political end.

I’d add that we libertarians are a good deal more consistent than most about the censure here—we wouldn’t make exceptions and allow the initiation of force simply because some people call themselves “the state.” But the intuitions that force and fraud are almost always wrong, and that they are wrong in similar ways, are both very common. They are not at all weird, and, used properly, they constitute a strong argument for our side. They ought not to be rejected.

Ayn Rand was a brilliant author, an influential polemicist, and a great thinker. I don’t care if the academic mainstream thinks less of me for saying so. That’s so much the worse for them.

But she wasn’t always right. Foreign policy is one of the areas where she stumbled, at least on one occasion. Her essay “The Roots of War” in Capitalism: The Unknown Ideal is magisterial, and I love it in particular for the following:

The trader and the warrior have been fundamental antagonists throughout history. Trade does not flourish on battlefields, factories do not produce under bombardments, profits do not grow on rubble. Capitalism is a society of traders—for which it has been denounced by every would-be gunman who regards trade as “selfish” and conquest as “noble.”

Truer or more important words have never been written about human history. I find it difficult, however, to accept the following quotation, also from her:

Anyone who wants to invade a dictatorship or semi-dictatorship is morally justified in doing so, because he is doing no worse than what that country has accepted as its social system.

I recognize no right whatsoever for dictatorships to exist. But I think that Rand erred here in two ways.

First, it is perverse to hold that people suffering under a dictatorship have “accepted” the dictator’s evil acts, such that anything we do to them in the course of a war is no better than what they are in effect doing to themselves.

Many of a dictator’s subjects doubtless are responsible.  Not just the dictator and his cronies, but also his enablers in politics, the press, the academy, the business world, and elsewhere. But others surely wish the dictator were gone. If anything, they probably wish it more strongly than we do. Lumping these people together with their domestic persecutors is indefensible. 

Note the collective noun, “that country,” and ponder its moral implications. It’s not the sort of thinking I would expect from an individualist like Ayn Rand. Indeed, coming from her, I find it prima facie evidence that this answer – first given in an interview – wasn’t the product of deep reflection.  It’s on such thin foundations that some have even built a foreign policy approach that Jennifer Louise Burns has likened, embarrassingly, to neoconservatism. All of which I ultimately find hard to reconcile with “The Roots of War.”

Many in a dictatorship would obviously prefer a free, liberal polity. Yet achieving that goal may well be impossible, and we do not ordinarily blame or punish people for failing to do the impossible, or for not achieving it on a schedule of our devising. Nor can we discount their lives to nothing; they are not sacrificial animals any more than we are. That’s the reason why acts of war must survive an exceptionally difficult moral test.

Rand’s second mistake was to neglect another relevant group of people: the citizens of the liberating country. I agree as a matter of principle that it is always legitimate to overthrow a dictator. Dictators also need to know that when we are attacked, we will destroy them, as we should, wherever they may be located geographically.

Yet for better or worse, I am also a student of history. And in the history of our country, there has not been one single armed conflict of any note in which Americans’ liberties did not contract. War has always brought restrictions on travel, immigration, trade, free expression, habeas corpus, and the judiciary as a whole. It can still mean conscription, the closest thing to slavery that we tolerate under law. Here or abroad, it can mean displacement, dispossession, internment, and torture.

These are not accidents or coincidences. They are the very stuff of war; they are what war is. As James Madison put it, “Of all the enemies to public liberty, war is, perhaps, the most to be dreaded, because it comprises and develops the germ of every other.”

If I could push a button and depose a dictator without any other consequences, then of course I would do it. But that is not and has never been a choice in the real world. There the question is much more complicated, and experience has shown that very often we must decline the offer, not because we are indifferent to liberty, but because we value it so much.