Social Justice, Libertarianism

The Cohen Fallacy: Why Cohen’s and Rawls’s Arguments about Institutions Are, Well, Garbage

Chapter 3 of Why Not Capitalism? explains why Cohen’s argument for the intrinsic moral superiority of socialism fails. Even though I agree with Cohen that we should not dumb down the requirements of justice to accommodate ignoble motivations, his argument for socialism rests on a few mistakes, the first of which I call the “Cohen Fallacy”.

Recall that Cohen tries to establish the moral superiority of socialism by having us imagine a camping trip. In the socialist version of the camping trip, he stipulates that everyone is fully motivated by morally good attitudes and ends. In the capitalist version, the campers act like selfish, callous, egoistic jerks. The socialist version is clearly better. Cohen then asks, wouldn’t it be great if we could somehow make the whole world like the socialist camping trip? Most people say yes, but then dispute whether that’s possible. But, as Cohen rightly points out, 1) whether something is desirable doesn’t depend on whether it’s possible, and 2) a society motivated by kindness and generosity is eminently possible. In defense of 1, consider: It would be awesome if I could cure AIDS by snapping my fingers, even though I can’t. In defense of 2, Cohen admonishes us not to confuse “can’t” with “won’t”. People easily could be kind and generous; they just aren’t. People could choose not to take advantage of power or to free ride on each other; they just don’t. Asking people to do the right thing is not like asking them to jump to the moon or to lift 10,000 pounds.

So far, so good, I think. But that doesn’t help Cohen defend socialism. Cohen’s argument is:

  1. The socialist camping trip was better than the capitalist camping trip.
  2. It would be desirable to make the world run like the socialist camping trip—having the world run like the socialist camping trip would be better than the way the world actually is.
  3. Therefore, socialism is intrinsically more desirable than capitalism.

The problem is that even if we grant Cohen’s premises (1 and 2), his conclusion (3) does not follow.

What Cohen did was construct an imaginary microsociety in which everyone is stipulated to have morally perfect motives and to always act upon those motives in the correct way. He then compared this to a version of that society in which people have the flawed mix of motives real people have. He’s right that the version in which everyone is stipulated to be perfectly virtuous was better. But that doesn’t tell us much.

On further inspection, his argument is little more than this:

  1. Socialism with morally perfect people is better than capitalism with flawed people.
  2. Therefore, socialism is better than capitalism.

But 2 doesn’t follow from 1. After all, Cohen’s manipulating 2 variables at once: 1) how virtuous and good people are, and 2) what political-economic regime they live under.

Suppose I repeated Cohen’s style of argument. I asked you to imagine a perfect kingdom ruled by a omnibenevolent, all-wise, perfectly efficacious king, and then compared that to pathological real-life democracies, such as France, the United States, or Greece. I then concluded that monarchy is better than democracy. You’d balk. You might agree that idealized monarchy is better than realistic democracy, but that leaves open whether idealized democracy is better or worse than idealized monarchy, and whether realistic monarchy is better or worse than realistic democracy.

So, dub the Cohen Fallacy the fallacy of making an argument like this:

  1. Socioeconomic regime X with perfect people is better than socioeconomic regime Y with real people.
  2. Therefore, X is intrinsically better than Y.

Bad argument.

Cohen’s Why Not Socialism is in that sense a compelling argument that ideal socialism is superior to non-ideal capitalism. But Cohen doesn’t compare like to like. The relevant comparisons are ideal socialism to ideal capitalism, and real socialism to real capitalism.  Cohen thus fails to consider whether ideal capitalism might be superior to ideal socialism. If Cohen wanted to show socialism is intrinsically superior to capitalism, he would need to compare his socialist camping trip to something like the Mickey Mouse Clubhouse Village. He has not done so.

In chapter 4, I go on to explain why even though the Mickey Mouse Clubhouse Villagers are morally good enough to make socialism work in a nice way, they get additional value from being capitalist. Everything socialism can do capitalism can do better; capitalism can do everything better than socialism. Yes it can. Really, thanks to the Mickey Mouse Clubhouse argument, we bleeding heart libertarians can now regard the lovely folks at Crooked Timber as far right-wingers; we can look down upon them the way they look down upon conservative Republicans.

Now, Cohen also argues–or, more precisely, asserts repeatedly without evidence–that capitalism encourages bad motivations. But, as I point out, that’s a social scientific question, and the evidence shows that Cohen is not only wrong about this, but has it backwards. If anything, capitalism makes us better people by Cohen’s standards.

By the way, John Rawls also commits the Cohen Fallacy, even though he admonishes us to avoid it:

…in both A Theory of Justice and Justice as Fairness: a Restatement, Rawls argues that certain political-economic regimes can instantiate his theory of justice while others cannot. However, if you look carefully, you’ll see that when Rawls discusses his favored regimes, he imagines that people have perfectly just motivations, and he explicitly imagines away many of the problems those regimes would face in the real world. But when Rawls brings up reasons for rejecting his disfavored regimes, he invokes the kinds of problems that would only occur if people had imperfect motivations, the very motivations he asked us to imagine they don’t have.

So, for instance, Rawls knows that in the real world, the social insurance, redistributive, and regulatory institutions he favors would lead to moral hazard and rent seeking. But, he dismisses these concerns, because he is doing “ideal theory,” in which we imagine people have a perfect sense of justice. Yet, not a page later, Rawls says that certain other regimes he dislikes are unjust because they might encourage people of great wealth to use their wealth to buy government power for their own ends. But, by Rawls’s ground rules, we’re suppose to be doing ideal theory, and imagining that people have a perfect sense of justice. And if people had a perfect sense of justice, they would not buy government power for their own selfish ends, since by hypothesis this is unjust. So, like Cohen, Rawls doesn’t play fair. When Rawls argues for one set of institutions over another, he continuously commits the Cohen fallacy. Rawls’s argument for why property-owning democracy is superior to welfare state capitalism thus is a kind of cheating. Rawls sets out some rules for thinking about institutions, and then violates them in order to generate his preferred conclusion.

Oops! Jim Buchanan to Rawls: “Hey, POD will encourage exploitative rent seeking.” Rawls: “I’m doing ideal theory. In ideal theory, people are just. They wouldn’t abuse the system.” Buchanan: “Great, so why not capitalism?” Rawls: “Because people would use their extra money to buy power for themselves.” Buchanan: “Wait, I thought we were doing ideal theory.” Rawls: “Ha, ha, ha. I decide the rules, Jim.”



  • Jameson Graber

    When you put it like that, your argument sounds too devastating to be true. Does Cohen (and does Rawls) really make such an elementary mistake? Now I’ll have to read Cohen’s book (and then yours) to figure out what’s going on.

    • Jason Brennan

      I was worried about these same points. Many of the mistakes I point out just seem obvious. But, after presenting this at the APA and a bunch of universities, plus going through peer-review, so far no one, including Cohen’s fans, has disputed this. In fact, the most common response I’ve gotten from left-wing professors is, “Holy crap, you’re right.”

      • Koen Swinkels

        This puzzles me as well. What does it say about the relevant fields in philosophy that despite such an elementary and fundamental mistake this has been by far the most important and influential book of the past century?

        I’m not even sure whether it would be better if most people simply never noticed the error or if the error is widely noticed but somehow never really talked about much or acknowledged as a crucial problem in Rawls’s argument.

        • Jason Brennan

          Rawlsianism has been a degenerate research program for at least the past decade. There’s a lot to be said for Rawls’s theory of justice, but most of the work written in that area now is just new age scholasticism.

      • Koen Swinkels

        and I’m (pleasantly) surprised that you got that kind of response from left-wing professors (because it shows that they’re open to criticism). Do you have a sense of how strongly your argument changes their beliefs about Rawls and Rawlsianism in the longish run?

        • Jason Brennan

          I don’t attack Rawls’s or Cohen’s theories of justice, just their theories of institutions. So, I’m just getting them to admit that a morally perfect society would be capitalist. But this book leaves open what to do in a world full of rotten, immoral people.

          • ThaomasH

            It may be that not even a society of morally perfect beings could not make “socialism” work and that only a society of morally perfect beings could make “capitalism” work. In which case maybe we need to forget about isms and work on deregulating taxis and taxing CO2 emissions and lots of other things to make what we have work better.

      • Sean II

        “In fact, the most common response I’ve gotten from left-wing professors is, ‘Holy crap, you’re right.'”

        Given ’em a few months, and they’ll have cooked up a sophistic refutation for you. I mean…what else are they going to do, change? Hahahahaha!

        Taking nothing away from your credit as the guy who found and exposed those mistakes, the truth is as you say – they should’ve been obvious. So the question is: How did they go unnoticed for so long?

        Answer: Why shouldn’t they go unnoticed? Cohen hit for the home team, so who cares if he corked up his bat?

        The usual pose of left-wing intellectuals is that they started as intellectuals, and as a result of their immense learning became left-wing. But moments like this suggest the opposite is more common. Call it the Hans Landa theory of anti-capitalism: “You don’t like markets. You don’t really know why you don’t like them. All you know is, you find them repulsive.”

        • zjohn

          The opposite is indeed true.

      • assman35

        I thought it was obvious that Rawls work is utter garbage. As is Kant’s and most other philosophers. I am surprised you are surprised. I always felt that to be a philosopher you had to basically pretend not to notice vast flaws in logic and huge holes.

  • TracyW

    People easily could be kind and generous; they just aren’t.

    But that’s not the whole problem with socialism. As Hayek would point out, there’s still the calculation problem, as society becomes more complex we must act on things for which we don’t know a lot. Let’s say, for example, that I’m a farmer and I genuinely want to feed people only out of the good of my heart. Different crops however have different nutritional benefits, different appeals to taste, and require different resources to grow. Furthermore which crops are the best to grow depends on what other people are doing, eg if other farmers can grow wheat much more efficiently than me it may make sense to concentrate my efforts elsewhere. Thus, prices.

    • Jason Brennan

      Yes, that’s right. Cohen actually accepts the Calculation Problem, but he thinks we don’t know for sure that it rules out market socialism. (He thinks Roemer’s work in defense of market socialism *might* be right.)

    • AP2

      That’s in the post:

      > In chapter 4, I go on to explain why even though the Mickey Mouse
      Clubhouse Villagers are morally good enough to make socialism work in a
      nice way, they get additional value from being capitalist. Everything
      socialism can do capitalism can do better; capitalism can do everything
      better than socialism.

  • jtlevy

    “in both A Theory of Justice and Justice as Fairness: a Restatement, Rawls argues that certain political-economic regimes can instantiate his theory of justice while others cannot. However, if you look carefully, you’ll see that when Rawls discusses his favored regimes, he imagines that people have perfectly just motivations, and he explicitly imagines away many of the problems those regimes would face in the real world. But when Rawls brings up reasons for rejecting his disfavored regimes, he invokes the kinds of problems that would only occur if people had imperfect motivations, the very motivations he asked us to imagine they don’t have.”

    Yes, yes, yes, dammit. Exactly right.

    • Jason Brennan

      It seems like such an obvious mistake that I can’t help but wonder if it’s intellectually dishonest.

      • murali284

        Rawls has got lots of moving parts in his theory. And towards his later life, I understand that he was surrounded by people who were to the left of him. I think we should just attribute it to him not paying close enough attention to those parts of his already unwealdy theory. I’m surprised someone like Nozick or Lomasky didn’t catch it before.

      • assman35

        “It seems like such an obvious mistake that I can’t help but wonder if it’s intellectually dishonest.”
        Huh? If its intellectually dishonest then most of philosophy is the same. You are required to throw away half your brain to do philosophy.


    Haven’t you just shown that capitalism with perfectly just individuals could be awesome, just as well as socialism with perfectly just individuals could be awesome? I could stipulate, as you have with capitalism, that there are defining characteristics of socialism that one could point to claim such is why socialism might be deemed better (e.g., a close-knittedness you don’t find in capitalism), not that it is in some independent way, better.

    • Jason Brennan

      You’ll have to read the book for that. I ended up arguing that one reason why capitalism is best is that for some people socialism is best. The Mickey Mouse Clubhouse Village allows Smurf communes within its borders, so long as the Smurfs aren’t going to force everyone to adopt Smurf communism.

      • Raul Rodriguez

        This reminds me of something I heard Walter Block once say. He said, “It’s not a debate between capitalism and socialism, it’s a debate between voluntary action and coercive action.” He pointed out that there are plenty of voluntary socialist type structures within any capitalist system, starting with the family.

  • Alvaro

    Not to mention, in the socialist camping trip your intentions don’t matter. You “share” or they help themselves to your stuff anyways.

  • adrianratnapala

    I would like to hear the Mickey Mouse argument. Maybe we can have a Mickey Mouse course for explaining it.

    But I think it is not fair to assume capitalists angels will be as nice as socialist angels. For example socialists can demand: “from each according to his ability, to each according to his need”. Demanding the same from capitalist angels just spoils all the fun.

    For a fair comparison, the capitalists should have no more than Ayn Rand style virtue. Selfish, but be honest in business and willing to accept misfortune in good grace. That is we should be comparing the Smurfs to the folk of Galt’s Gulch.

    • Jameson Graber

      You’ve put your finger on one of the key problems in this debate. If we’re going to do ideal theory, we have to agree on what it would mean for people to be “ideal.” Then we have to imagine them submitting to some system. The problem with your comment is that you propose, apparently just as Cohen does, to compare two different “ideal” people. The real question is, given a group of people already satisfying an agreed upon set of ideal moral criteria, which system–socialism or capitalism–would they be more likely to submit to?

      • Jason Brennan

        Jameson, that’s exactly what I do. I give Cohen all of his morality, but then argue that his moral system and his idea of what’s ideal privileges capitalism over communism.

        Ayn Rand constructs a utopia that Cohen and I both find repugnant. The Mickey Mouse Club utopia is one against which Cohen can mount no principled moral objection.

    • Jason Brennan

      Nah. Same morality, different institutions. Ayn Rand defends the morality of capitalism by dumbing down morality. I defend the morality of capitalism by ramping it up. If anything, I see some of Cohen’s arguments as dumbing morality down too much.

      The problem Cohen has is that he identifies capitalism with selfishness. I argue in the book there’s no reason, conceptually or empirically, to do that.

      • adrianratnapala

        The debate between socialism and capitalism is partly (or mostly) a clash fundamental moral intuitions. Capitalists rightly smirk we demand less unnatural benevolence from people, this makes capitalism both more achievable and nobler (because it leaves people freer). However, in the Utopia game game it is Pyrrhic to win only when the
        capitalist agents submit to as many restrictions as the socialist ones.

        There is wiggle room (which your book might fall well inside). You could assume that capitalutopians have not just the Randian virtues, but are also fairly nice in the way real people are. But we can’t go to the same extremes as the socialists. For example, they can stipulate that people feel as altruistic towards the state as they do to their own friends; but I find that creepy, and don’t want to see it in any model of capitalism. Or in the real world.

  • David Gordon

    Harold Demsetz ‘s “Nirvana Fallacy” is very relevant here.

    • Jason Brennan

      The thing is, Cohen has a good response to Demsetz. He’s trying to explain what we *should* do, even if he knows we won’t do it.

      I agree with Cohen on this point. I think that justice clearly requires that we live in a Nozick meets Mickey Mouse anarcho-capitalist meta-utopia. If people did the right thing, as they could, that’s what they would do. But, given that people are jerks, we then have to ask what institutions work the best in light of how awful people are. And it might be that different institutions work better.

      • David Gordon

        I see;so Cohen has then inverted the Nirvana fallacy. He has failed to take into account ideal capitalism, when he needed to do so to make his case.

  • Chris

    Hey Jason, there’s a typo in your self-quotation. It should be “supposed” and not “suppose”.

    Hope that didn’t make its way into the print version!

    • Jason Brennan

      Luckily, I’m drawing from the manuscript.

  • Ron Jawls

    This might be a bit unfair to Rawls. Rawls points out that the dispositions people have in society are molded by the institutions they’re raised under. Different institutions inculcate different dispositions. The point of part three of TJ is to show that the sense of justice would develop and thus the disposition to comply with what justice requires would naturally arise in a society characterized by justice as fairness (i.e., in a property-owning democracy or liberal democratic socialism). Can’t the Rawlsian say that Rawls is asymmetrically idealizing, but not arbitrarily? We have reason to believe that property-owning democracies would be characterized by citizens of a certain kind of motivational makeup, whereas capitalist institutions wouldn’t inculcate that same motivational makeup, thus requiring an asymmetry in the analysis.

    Note, this is actually in line with what Buchanan says about the employment of homo economicus in comparative institutional analysis. In a paper w/ Geoffrey Brennan (“The Normative Purpose of the Economic Science”) he says that we’re justified in assuming a different motivational makeup in the political arena if we can offer an explanation of why specific institutions lead to this motivational change. Rawls can say that there’s something about property-owning democracies that lead to this change, something absent in capitalist institutions.

    • Jason Brennan

      Yes, I realize Rawls and Cohen could both try to make this move, but I call this the “Other Cohen Fallacy” and then complain that they’re again being ideologically irresponsible. The problem is that it’s not an armchair exercise, and the empirical work suggests they’re mistaken.

      • POD person

        “The empirical work suggests they’re mistaken.”

        How can this be if there are, and have been, no property-owning democracies?

        • Jason Brennan

          Yes, and no capitalist societies either.

          • POD person

            You’ve lost me. What then is the “empirical work” in question?

  • CFV

    I think you are misreading Cohen and, as a consequence, there is no “Cohen fallacy”. Cohen does not compare ideal to real. Instead, he devices a micro setting in which socialism is CLEARLY superior to capitalism and then proceeds to ask: why not socialism on a larger scale?

    Maybe there is a micro setting in which anarcho-capitalism is clearly superior to socialism (your Mickey Mouse Clubhouse village) and yet some complexities of modern societies prevents us from having anarcho-capitalism on a large scale (for example, the problem of national defense. See David Friedman The Machinery of Freedom, Chapter 34)

    • Jason Brennan

      I don’t see how that’s a disagreement with me. You are describing my argument.

  • mary

    Am from UK I give thank to a great doctor who help me out of my
    illness I was very sick I thank god who use this man to help me, it
    started when i travel to Florida for visit there I meet a lady not
    knowing to me that she HIV positive I really like the lady because she
    was beautiful I always see her every moment I close my eye I went to
    tell her how I feel for her but I don’t know if she know that she HIV I
    went to bed with her I contacted the virus too when I get home for month
    my doctor come to check on me and he discover that I have HIV he was
    shock and tell me I was so confused and so surprised to hear that I was
    taking HIV drug to cure it for good a 2year I decided to look for cure
    them I meet this post on internet I contacted him for help… well
    DR.ODUMODU proving to be a great man and he heal me.. According to him
    he said is the power of his gods well I thank god am back again if you
    need cure for your HIV…you can still contact him or his email or number I
    promise he is 100% he cure any virus like T-VIRUS HIV AIOTA-VIRUS,
    SMALLPOX, HEPATITIS B if you have this virus or friend or relative
    contact him now (

  • Swami Cat

    As I was reading this I kept thinking of *the problem of cooperation.* This is, of course a central problem in sociology, game theory and evolutionary psychology.

    Rawls and Cohen seem to ignore the problem when convenient.

  • JW Ogden

    To me that is not even socialism. In socialism the people as a whole own everything and decide as a whole how everything is used. Everyone doing the good is capitalism with all very, very good people. A church picnic where everyone brings what they want is not socialism.

  • Phil Lines

    Jason, can you please do a takedown of Stefan Molyneux? It would surely serve your goal of minimizing cartoonish or hooliganish libertarians, and hopefully equipping them with better arguments. His followers make libertarians look awful, and he also ‘splits’ people into good, clever people who agree with him and bad, stupid people who disagree with him, as you noted some do in a previous post. Thanks!

    • Sean II

      If you think Stefan is disgracing the movement now, just wait until Donnie Darko burns his house down and they find god-knows-what in the wreckage.

  • jtkennedy

    Presciently, David Friedman addressed the Cohen Fallacy in 1973:


    Most varieties of socialism implicitly assume unanimous agreement on goals. Everyone works for the glory of the nation, the common good, or whatever, and everyone agrees, at least in some general sense, on what that goal means. The economic problem, traditionally defined as the problem of allocating limited resources to diverse ends, does not exist; economics is reduced to the “engineering” problem of how best to use the available resources to achieve the common end.

    The organization of a capitalist society implicitly assumes that different people have different ends and that the institutions of the society must allow for that difference.

    This is one of the things behind the socialist claim that capitalism emphasizes competition whereas socialism emphasizes cooperation; it is one of the reasons why socialism seems, in the abstract, to be such an attractive system. If we all have different ends, we are, in a certain sense, in conflict with each other; each of us wishes to have the limited resources available used for his ends. The institution of private property allows for cooperation within that competition; we trade with each other in order that each may best use his resources to his ends, but the fundamental conflict of ends remains. Does this mean that socialism is better? No more than the desirability of sunny weather means that women should always wear bikinis or that men should never carry umbrellas.

    There is a difference between what institutions allow and what they require. If in a capitalist society everyone is convinced of the desirability of one common goal, there is nothing in the structure of capitalist institutions to prevent them from cooperating to attain it. Capitalism allows for a conflict of ends; it does not require it.

    Socialism does not allow for it. This does not mean that if we set up socialist institutions everyone will instantly have the same ends. The experiment has been tried; they do not. It means rather that a socialist society will work only if people do have the same ends. If they do not it will collapse or, worse, develop, as did the Soviet Union, into a monstrous parody of socialist ideals.

    The experiment has been done many times on a more modest scale in this country. Communes that survive start with a common end, whether provided by a strong religion or a charismatic leader. Others do not.

    I have encountered precisely the same error among libertarians who prefer limited government to anarcho-capitalism. Limited government, they say, can guarantee uniform justice based on objective principles. Under anarcho-capitalism, the law varies from place to place and person to person, according to the irrational desires and beliefs of the different customers that different protection and arbitration agencies must serve.

    This argument assumes that the limited government is set up by a population most or all of whose members believe in the same just principles of law. Given such a population, anarcho-capitalism will produce that same uniform, just law; there will be no market for any other. But just as capitalism can accommodate to a diversity of individual ends, so anarcho-capitalism can accommodate to a diversity of individual judgments about justice.

    An ideal objectivist society with a limited government is superior to an anarcho-capitalist society in precisely the same sense that an ideal socialist society is superior to a capitalist society. Socialism does better with perfect people than capitalism does with imperfect people; limited government does better with perfect people than anarcho-capitalism with imperfect. And it is better to wear a bikini with the sun shining than a raincoat when it is raining. That is no argument against carrying an umbrella.”