Passionate about liberty and want a chance to win $4,000? Check out our video contest!

This is part of a series

September 2013

Experiment THIS!: Libertarianism and Thought Experiments

Levatter explains how thought experiments can be a helpful tool in political philosophy, but only if they reach some minimum level of plausibility.

In a May, 2013 Bleeding Hearts Libertarians post, Matt Zwolinski offered what he called his positive thesis about BHL’s world view, what he describes as one of the two “most important beliefs that motivated [him] to start” the BHL blog and that “still drive a lot of [his] thinking today”:

If the left was [sic] right in its belief that libertarian institutions would impoverish the poor, or fail to provide them with sufficient real freedom, or lead to the oppression of socially marginalized groups, and so on, then these would be very good reasons to doubt that libertarian institutions are morally justified[.]

Zwolinski’s reflection here recalls a BHL thought experiment created by Jason Brennan back in March, 2011:

Imagine that a bunch of economists provide compelling evidence that life in a strictly libertarian polity would be a disaster. Imagine that they show conclusively that if people everywhere were to live in a Nozickian minimal state or a Rothbardian anarcho-capitalist civil society, with everyone strictly observing property right rules, that 10% of people would starve, 80% would be near subsistence, and only 10% would prosper.

Brennan concludes the thought experiment by asking, in this situation: “what would you advocate and why?”

Both Brennan’s argument, in “What If You Were Wrong About Economics?” and Zwolinski’s positive thesis in “Getting Over Social Justice” hinge on the fact that since economics is an empirical science, we can freely make thought experiments based on one or another economic conclusion being falsified. As Brennan put it in the paragraph preceding the one above: “Since economics is…empirical, it’s logically possible for it to be wrong. We live in a world where standard neoclassical economics is much closer to the truth than, say, heterodox Marxist economics. But, as a matter of logic, it could have been otherwise… There’s some possibility, and perhaps even some non-zero probability, that most of your empirical economic beliefs could be disconfirmed in the future.” So, Brennan and Zwolinski ask, each in their own way1, why are some libertarians so adamant about universalist libertarian moral principles even if they were to lead to disastrous consequences? Surely, they imply, that’s not reasonable.

Now it’s not as if “hard libertarians”—a term Brennan, in his book Libertarianism: What Everyone Needs to Know, uses to describe traditional modern libertarian thinkers like Rothbard, Nozick, and Rand, and most every other 20th century libertarian luminary—had never before given this potential divergence between moral deontological precepts and consequentialist norms a moment’s thought. For example, almost a quarter century ago Randy Barnett had some cogent things to say in his foreword to the Harvard Journal of Law and Public Policy “Symposium on the Compatibility of Rights and Consequential Analysis.” (see his “Of Chickens and Eggs—The Compatibility of Moral Rights and Consequentialist Analyses,”.) I’ll get back to his view in a bit, but first I want to say a few words about these thought experiments that Brennan, and perhaps Zwolinski as well, seem to believe provide a real dilemma for traditional libertarians. I have no problem with thought experiments per se, but I’m sure professional philosophers would agree with the following very limited claim: thought experiments can be well constructed or they can be poorly constructed.

As Brennan knows, “well constructed” is not synonymous with “realistic.” Brennan, in the summer of 2012 on the BHL blog, took Wendy McElroy to task for opposing “unrealistic” thought experiments. There he argued that because, for example, we can make sense of statements like “Mr. Spock was ethically correct to choose his own death so that many could live in Star Trek II, The Wrath of Khan” and “Voldemort was evil for killing Professor Snape in Harry Potter and the Deathly Hallows” that therefore there is no problem with “unrealistic” thought experiments (since the worlds of Star Trek and Harry Potter are clearly unrealistic).

But what about this unrealistic thought experiment?

In a re-write of the Harry Potter series, wizards are essentially immortal, such that no matter how completely and thoroughly they are killed, they pop back into existence a few seconds later no worse for wear. In this world, how evil was Voldemort for killing Professor Snape?

In such a world, killing a wizard is much like scratching his hand without his permission. Do we have strong ethical intuitions that Voldemort’s actions are evil in such a world, where killing Harry’s parents amounted to much ado about nothing? Even though our intuitions here suggest that wizard murder may not be such a big deal in that world, is there anything we can reasonably thereby conclude about murder in our Muggle world?

What is the point of this digression on thought experiments? It is due to my impression Zwolinski and Brennan believe they have traditional libertarians in a dilemma: “You SAY you believe in universal moral principles,2 but what if they led to THIS [followed by pictures of corpses piled high, burned out cities, children on the rack, etc.]? Would you still support them THEN?” But I think the dilemma is flimsier than they appreciate.

Here’s the problem: The economic counterfactual on which Brennan bases his thought experiment is not like “Imagine demand for X is inelastic3 rather than elastic”. It is more like, “Imagine demand curves slope upward4 rather than downward for all goods over all ranges. Then what?!” Such thought experiments, based on such a gross and fundamental divorce from all empirical economic knowledge we have ever experienced, give us little guidance, just as (even though biology, too, is an empirical science) a thought experiment involving the immortality of wizards would give us little insight into appropriate ethical action in our world.

The Star Trek thought experiment above works because, even though we know the fantastical future world of Star Trek is different in many unknowable ways from our world, we see the characters interacting in ways that seem meaningful and understandable to us. They form friendships, loving bonds, command loyalty and affection, develop competitive feelings, seek wisdom and advantage, act heroically and ignominiously, and so on, as we do. So we believe we have intuitions as to their motivations and an understanding of their goals and desires. It seems natural, in other words, to apply to them moral motivations and strictures we would apply to ourselves.

Of course, the more fantastical the world of Star Trek, the more reasonable it becomes to potentially question such judgments. The marriage practices and rituals on Vulcan (the Pon Farr of “Amok Time,” for example), are such that our moral intuitions with respect to marriage may well not apply to Mr. Spock. And, as many Star Trek fans with even a basic understanding of economics have pointed out over the years (see, for example, this essay by Ilya Somin,) it is very difficult to coherently makes sense of the economics of Star Trek spin-offs like The Next Generation, given the post-scarcity society created by replicators and the absence of money as a common medium of exchange. So a thought experiment asking about economics in a Star Trek context would be much more challenging and much less clearly applicable to current economic issues than Star Trek thought experiments about morality.

In a discussion of thought experiments in his Reasons and Persons, Darek Parfit made the following distinction (p 219): “[C]onsider certain imaginary cases…I distinguish two kinds of case. Some cases contravene the laws of nature. I call these deeply impossible. Other cases are merely technically impossible. Does it matter if some imagined case would never be possible? This depends entirely on our question…[D]epending on our question, impossibility may make some thought experiment irrelevant.”

Presumably no one, including Brennan and Zwolinski, really believes they know of any historical instance where what traditional libertarians desire has been closely approximated and yielded anything remotely close to the results of their thought experiments. Is it fair to describe their pure thought experiments as divorced from real world experience in a very fundamental way, deeply impossible in Parfit’s sense?

Consider some well known philosophical thought experiments and see how the Brennan thought experiment compares. Take, say, the uncle Sam thought experiment with which Michael Huemer begins The Problem of Political Authority5 or the transfused violinist in Judith Thomson’s famous essay “A Defense of Abortion”.6

In Huemer’s thought experiment, we know all there is to know about Sam. We know what his actions are, which are very similar to actions taken by government agents. We understand his motivations, which match the motivations of government officials. We appreciate that in all aspects save one—he wears no uniform; he is NOT a government agent—his actions perfectly mimic those of government agents. Were Sam a government agent, his actions would be perfectly acceptable to us, perhaps even praise-worthy. But he is not a government agent, and we are appalled. Thus this thought experiment allows us to focus on the issue of political legitimacy.

Similarly, Thomson’s famous thought experiment creates a person made into a patient without her initial consent. This is done to save the life of another. The medical condition and treatment described are fictional, but the moral issue—may one be forced to help another live?—is clearly analogous to the question of abortion, and pushed people to recognize there was more at issue than “the fetus is innocent.” Thomson’s comatose violinist is innocent as well, but the thought experiment focuses our attention on whether mere innocence is sufficient to justify forcing help from others.

How does the Brennan thought experiment compare? Poorly, I hope to now show. I appreciate that Brennan’s general answer to people (especially non-philosophers) who reject his thought experiment, and by implication Zwolinski’s positive thesis, is that they do so because they are not sophisticated as to the workings of thought experiments in modern philosophy. (As he put it on the BHL blog on March 16, 2011, “It can prevent them from being able to do philosophy at even an introductory undergraduate level.”) But I demur from that view. Thought experiments can be well constructed or they can be poorly constructed. Good thought experiments make relatively clear and focused changes in well understood contexts. Huemer’s uncle Sam is an individual taking clear and specific actions that are universally recognized as acceptable if he is an agent of the government, but in the context of the thought experiment he is merely acting as an individual. Thomson’s violinist is helpless and innocent, like a fetus, and the question of whether or not someone can thus be forced to assist for 9 months is brought to the fore. The choices offered in Nozick’s description of Newcomb’s Problem7—the contents of the boxes and the conditions by which the contents are determined—are clear and unambiguous, if perhaps confusingly self-referential.

What is offered in Brennan’s thought experiment is, instead, unclear and hopelessly confused. Why is 90% of the populace immiserated? Is trade impossible despite the claim of free markets? Is this a society where only a few have even minimal skills, a complete world of Zero Marginal Productivity (ZMP)? Do the 10% of the populace that are well off have no incentive to become even better off by mutually beneficial interactions with the 90%? Is this a world where the law of comparative advantage is not in play? Is this a world where trade is not mutually beneficial? A world where specialization of labor does not improve productivity? To highlight the problem: This is a literally unimaginable world.

One clear piece of evidence the Brennan thought experiment is not fruitful is that it is too easily multiplied.

Imagine that a bunch of political scientists provide compelling evidence that life in a Rawlsian liberal democratic polity would be a disaster. Imagine that they show conclusively that if people everywhere were to live in a democratic system following basic Rawlsian principles of justice , with everyone strictly observing the difference principle, that, paradoxically, 10% of people would starve, 80% would be near subsistence, and only 10% would prosper.” Does this thought experiment add anything to the question of whether or not Rawlsian arguments are reasonable?

Imagine that a bunch of theologians provide compelling evidence that life in a strictly secular polity would be a disaster. Imagine that they show conclusively that if people everywhere were to not devote their lives to the Church, with everyone strictly observing tithing rules and the Ten Commandments, that 10% of people would starve, 80% would be near subsistence, and only 10% would prosper.” Does this thought experiment add anything to the question of whether or not secularism is reasonable? Would it become relevant if I said the purpose of this exercise is to see how your moral commitments interact with your empirical beliefs about how secularism actually works.

Granted, in these thought experiments it is unclear just how liberal democratic polities or strict secularism could lead to such horrid results, just as in Brennan’s thought experiment it is unclear just how free market economics could lead to such horrid results. Nonetheless, it must be admitted that political science and the sociology of religion are not deducible from the laws of logic. In this sense they are, like economics, empirical. Thus a thought experiment can assume they reach certain results, even if they in fact don’t. Truth be told, for most people it is fairly easy to imagine a non-religious world would lead to disaster. On the other hand, it is hard to imagine what it even means to say that as a rule voluntary trade is not mutually beneficial.

It is, to say the least, unclear what if any limiting principle informs Brennan’s “if a field of study is empirical, form a thought experiment to see what happens when fundamental connections or conclusions of the field are denied” approach. Reductios seem too obvious. The following appears to be a clearly constructed analogy to Brennan’s thought experiment:

Agricultural management is an empirical science. Imagine a world where property held severally leads to crop destruction while property held in common leads to bountiful yields without much effort. Are traditional libertarians still committed in such a situation to demanding several property? If not, does that tell us that traditional libertarianism fails in the real world? To paraphrase Brennan, “If agricultural scientists were to provide compelling empirical evidence that you are wrong about how farming works, and that a libertarian society composed completely of several property would be a humanitarian disaster, what would you advocate and why? In particular, would you think it would be morally right for states to mandate communal farming? How much and why?”

It may be replied that the problem I describe is not really an agricultural management problem, but an economic problem. That doesn’t help eliminate the reductio. It just brings us full circle. After all, as Brennan stressed at the outset, economics, too, is an empirical science. So imagine a thought experiment that posited studies which proved the putative tragedy of the commons was wrong, that careful quantitative examination showed people took better care of land held in common than land held severally. Would any BHL philosopher then ask, “Would hard libertarians still call for full private ownership of land if only communal farming led to bountiful crop yields?” This is, after all, just one specific application of Brennan’s more general thought experiment. Note as one specifies Brennan’s thought experiment it becomes increasingly clear that it is incoherent.

More generally: “Imagine the entire social world is different in numerous and unspecified ways than all evidence has ever demonstrated it to be” is not a helpful premise with which to begin a thought experiment. It is a rather major change from all experience and it is thus harder to trust our intuitions, and harder to analyze the implications of our intuitions. How would such a counterfactual lead us to any pertinent or helpful conclusions? Should libertarians be devastated to find our principles of political economy work only in the actual world we live in and fail in some other logically possible worlds that are not even close to ours?

To paraphrase Barnett in his 1989 foreword, granted that most people who claim, “Liberty, though the heavens should fall” say this in part because they think it highly unlikely the heavens will in fact fall, they are right to think that. It’s not a coincidence. Historically, if only 10% of people prospered in market economies while 90% were immiserated, nearly the whole of libertarian study would not have developed. Thus, the wedge Brennan’s thought experiment attempts to push between moral deontological principles and consequentialist norms is quite soft, crumbles easily when dissected, and is more than a little cheesy… a wedge of Camembert, not steel.

The answer to Brennan’s question is: “In the unimaginable world you posit, no sensible person would any longer advocate libertarianism. But this deeply impossible thought experiment does not provide even one iota of a reason to deviate, in this world, from universal deontological principles supporting a ‘hard’ libertarian polity.”

Here’s a thought experiment: Would Brennan and Zwolinski agree with my analysis if they would die otherwise? “As a matter of logic, it could” happen, after all. It IS a logical possibility, by which it is meant merely the claim doesn’t contain a self-contradiction. As Brennan might put it, “There’s some possibility, and perhaps even some non-zero probability” that they would die if they were to disagree with my analysis. What follows from that?



  1. As Brennan put it in the above essay: “The purpose of this exercise is to see how your moral commitments interact with your empirical beliefs about how economies actually work.”
  2. See Zwolinski criticizing the well-known Non-Aggression Principle, or NAP.
  3. Elasticity and inelasticity refer to the responsiveness of the demand curve to changes in price. If a small change in price leads to a large change in demand, the curve is elastic; if a large change in price leads to only a small, or no, change in demand, the curve is inelastic. Changes in elasticity with changing circumstances are common and easily imagined.
  4. It is a fundamental observation and basic assumption of standard microeconomics that the relationship of the demand for a good and the price of the good is such that more is demanded as the price drops. This yields a classic “downward sloping” demand curve. For a demand curve to slope upwards, demand would have to increase as price increases. For this to happen for all goods over all price ranges is clearly impossible.
  5. Huemer imagines a single individual who engages in actions analogous to those deemed morally appropriate for government agents—demanding people contribute to morally worthy goals like helping the poor; policing his neighborhood and then demanding his neighbors pay him for that work, and locking them up in his basement if they refuse. The thought experiment brings out the fact that the single individual’s actions are routinely condemned as coercive by virtually everyone, while the analogous actions, performed by government officials, are not merely accepted but deemed laudatory by many.
  6. Thomson’s famous thought experiment involves waking up to find you’ve been taken while asleep and medically connected, via tubing and semi-permeable membranes, to a comatose adult violinist. The doctors explain that you were the perfect match, and your remaining connected to this innocent adult—who has fallen into a coma through not fault of his own—for 9 months will allow him to survive and return to his former baseline state of good health, while disconnecting you from him will kill him. There is minimal to no risk to you, medically, just the 9 months of inconvenience.
  7. Newcomb’s Problem is a conundrum that focuses on two principles of choice, the Expected Utility principle and the Dominance principle. Initially created by William Newcomb of Lawrence Livermore Labs, it was made famous by Robert Nozick’s description in a festschrift to philosopher Carl Hempel. The two principles of choice usually lead to the same decision, but Newcomb created a situation where the two principles lead to divergent choices, and the problem is to justify which principle trumps, even while large numbers of reasonable people feel strongly the other principle is paramount. More details can be found at the link, an excellent summary discussion by philosopher David Schmidtz, but the point here is that the conditions of the thought experiment are well defined.

This is part of a series