From Herb Gintis’ excellent review of the lately-departed G.A. Cohen’s Why Not Socialism?:
Cohen argues that markets are morally offensive institutions that most people would be happy to get rid of if they could figure out some alternative compatible with the standard of living we are accustomed to in advanced market societies. “The market” says Cohen, “is intrinsically repugnant…Every market, even a socialist market, is a system of predation.” (pp. 78,82)
In place of the market, Cohen celebrates the caring and voluntary mutual aid that occurs in small groups of friends (he never mentions family), and believes this can be extended to a community of strangers as well. He calls this “communal reciprocity.” (p. 39)
[…]
Rather than lamenting the incompatibility of socialist community and human nature, Cohen faults our meager social technology; there is simply no known machinery for harnessing natural human generosity. He calls this an “insoluble organizational design problem.” “In my view,” he remarks, “the principal problem that faces the socialist ideal is that we do not know how to design the machinery that would make it run.” (p. 55)
I think there are two problems with Cohen’s argument. First, there is a reason why we lack the organizational institutions that harness human generosity, and it has to do with a side of human nature that Cohen does not recognize. There is a great deal of heterogeneity among people in the degree to which they privilege the personal, including self and family, over the social. Everyday observation, reinforced by a huge body of empirical evidence—see my book, Bounds of Reason (Princeton, 2009) for details—that unless there are safeguards against the free-rider tendencies of the selfish, the natural tendency for the majority to cooperate will be undermined, and cooperation will unravel. Moreover, the larger the group, the harder it is to identify and punish the free-riders, even though most people are willing to incur personal costs to do so. Markets work because they discipline firms, who then discipline workers, thus solving the free-rider problem. Moreover, markets discipline firms by forcing them to compete and therefore reveal to the public exactly what are the limits of the possible in satisfying consumer needs and using technology efficiently. The knowledge of production possibilities unleashed through market competition cannot be revealed in any other way that we know of.
Gintis’ second problem is another good reason also why not.
I’ve read a good number of Cohen’s papers and books. I’ve always enjoyed them, and I’ve always come away feeling he has clarified for me the contours of the debate. The great thing about Cohen was how transparently and unabashedly he angled for the result he wanted to get. But he wasn’t one to pack the conclusion into his premises, so when he would land short of his longed-for conclusion, you could be pretty certain that’s as close as you can get with those particular premises. And you could be pretty sure that if there were some other premises out there both more plausible and more amenable to producing the wanted result, he would have found them and started from there. For many years Cohen was to Anglophone analytic political philosophy something like Ted Kennedy was to American politics: he marked the outer bound of the reasonable left. I think now that contemporary political philosophers have begun to lose their “studied ignorance of standard social and psychological theory, common among philosophers of the mid-Twentieth century,” as Gintis puts it, the bounds are shifting, leaving Cohen’s views well past the edge of a receding tide.
Will Wilkinson 
I don't think that your analysis of contemporary strains of analytic philosophy is quite accurate. I think there is a strong selection bias which leads us to believe that the current crop of intellectuals is arguing from wrong premises or using poor forms of argument for their beliefs.I don't dispute that many people do argue badly, but I dispute that this constitutes a change. The philosophy we read from generations prior usually constitutes the best that group of contemporaries managed to put forward. G.A. Cohen will be read for decades to come. That guy at a conference a few months ago spouting that any deviation from strict metaphysical realism means a dive into total subjectivity about values…not so much.
I think now that contemporary political philosophers have begun to lose their “studied ignorance of standard social and psychological theory, common among philosophers of the mid-Twentieth century,” as Gintis puts it, the bounds are shifting, leaving Cohen’s views well past the edge of a receding tide.And I think you're uniquely well-placed to make that determination, Will.
Have you read his putative take down of Nozick (for analytic philosophers, this means libertarianism by extension) Self-Ownership, Equality and Freedom? It's a great book, but not for reasons Cohen would have wanted. If anything I came away from it more libertarian than I started – self-ownership had never seemed like a more attractive principle to hold.Anyway I think the knock-down counter-argument against his new book is that the best part of a camping trip is when everyone gets to go home.
I'm surprised that Gintis did not mention his own research that bears on this point: “The market” says Cohen, “is intrinsically repugnant…Every market, even a socialist market, is a system of predation.”Some of the research is here. The finding is that:”group-level differences in economic organization and the degree of market integration explain a substantial portion of the behavioral variation across societies: the higher the degree of market integration and the higher the payoffs to cooperation, the greater the level of cooperation in experimental games.”That is, the dominance of markets may actually help make people more cooperative with strangers (and possibly more tolerant and fair-minded as well).
“Rather than lamenting the incompatibility of socialist community and human nature, Cohen faults our meager social technology; there is simply no known machinery for harnessing natural human generosity. He calls this an “insoluble organizational design problem.” “In my view,” he remarks, “the principal problem that faces the socialist ideal is that we do not know how to design the machinery that would make it run.””It's funny, when some scientists discover that their theories fail because of certain constants of reality, they decide that must mean their theories are wrong. But for Cohen, it's the fault of those realities.It's like saying that my theory that 2+3=6 would be accurate, if only all the 2's could be more like 3's. I suppose it's unfair, since you can theoretically change human behavior in this fashion, but even if you could make everyone maximally generous, there's still an information barrier — it doesn't matter how generous I am if I can't simultaneously grasp the needs/desires of all those people I need to be generous towards. If only Cohen could've come up with some system of valuing things/services that we could agree upon, so that people could accurately judge how important it was to the other, potentially distant person, so that they could generously give their labor in exchange for similarly valued gifts.
*Why Not Socialism?* is the first thing by Cohen I've read, and I liked it. More than anything it reminds me of that chapter in Hayek's *The Fatal Conceit* about the extended order. Both Hayek (elsewhere) and Cohen approvingly cite Mandeville's satire of the marketplace: it really is making private vice into public virtue. Cohen differs from Hayek only in his emphasis on the word *vice*. Both of them think the best socialist schemes we have won't work, and that the social technology for harnessing our self-interest is very impressive. But Cohen holds out hope that one day social technology for harnessing fellow-feeling will one day catch up.This is going to seem glib, but Hayek never got to see open source software, wikipedia, reputation systems, and other social tech. Sure, none of those really approach the sophistication of fractional reserve banking and currency exchange. But you know, the first glimmerings of barter probably weren't very impressive either. The most Gintis is willing to say is that, “The knowledge of production possibilities unleashed through market competition cannot be revealed in any other way that we know of.” That doesn't make Cohen seem crazy for remaining agnostic about the possibility of alternative production schemes that rely less on “greed and fear.”I'm curious where you come down on this, Will. Do you think it would be nice if we hit on the social technology that would allow for a socialist egalitarian scheme that was comparable to a market economy in its productivity? Any guess on how likely we are to see something like that?
Btw, Gintis's Amazon reviews are awesome, and I've been subscribed to them since you first blogged about them.
…he wasn’t one to pack the conclusion into his premises…You mean like assuming that politicians are perfect substitutes for each other, thereby eliminating the need for a complicated effort to analyze the actual costs of political assassination that might (or might not!) lead to a conclusion that politicians aren't, in fact, overprotected?I guess baking the conclusion into the premise is fine as long as you're inclined to agree with the particular conclusion at hand. It's also kind of fun watching your standards juke and jive on a semi-weekly basis.(BTW, my favorite part of that Frey paper referenced above is that it is published by the Institute for Empirical Research in Economics. Who says economists don't have a sense of humor?)
Politicians don't have to be perfect substitutes for that argument to work. It's enough for them to be approximate substitutes. Sometimes the substitute will be worse, sometimes he'll be better. On average perhaps he'll be a little worse, but it's enough to recognize that the cost to the nation of losing a president in terms of presidential effectiveness is only the difference between his effectiveness and that of the next guy in line. Plus: it's no fair looking at that difference now! If we protected presidents less and accepted a larger chance of assassination, the political process would tend to select for more presidential VPs than it does today, when the role is essentially ceremonial. Which would shrink the difference further.
“markets may actually help make people more cooperative with strangers”Of course they do. Unless you live somewhere very homogeneous or live in a multi-multi-racial family, most of your interactions with people who look, act, dress, think different from you are going to be in the economic sphere (restaurants, retail, bars, etc.).
That's interesting, because I came away from Anarchy, State and Utopia much less libertarian than when I started it. Maybe if I read them both at the same time I'll achieve enlightenment.
That's not really surprising is it? Markets depend on very high degrees of trust and cooperation – for example that people will repay their debts, show up to work, generally respect property rights, and so on. Its pretty obvious, really, that economies where market institutions are weak – Russia, for instance, or Southern Italy- have lower levels of trust and cooperation.The interesting question is which way the causality runs – do market institutions engender trust, or does trust have to be widespread for market institutions to get started?
When I read feminist academics, I still see a lot of Freudian and Psychodynamic theories being kicked around as though they were authoritative. As a psych. grad student, I can safely say this is not mainstream psychological theory.The feminist academic stuff I read are tremendously loopy.
The respect you show for Cohen is surprising. There were engineers in the 1920s, 30s and 40s who said flight faster than the speed of sound was impossible. Cohen's philippics against the market are no different. We may praise these engineers for challenging uncertain belief, but in retrospect they are morons and we forget them. I count Cohen among them. Not only does he think the sound barrier is unbreakable–much worse he's afraid of flying. Having been a graduate student at Oxford, I can tell you he was not a gentleman to those outside his inner circle. He would joke with his students by wearing a Stalin mask. And, worse for All Souls College, Oxford, and political philosophy in general, he created a patronage system to create disciples who never strayed too far from the, I mean, his party line.
Gintis isn't a libertarian. He's an egalitarian. He's just saying Cohen's not making a great argument for the egalitarian side, and that Gintis's egalitarianism is better than Cohen's. I don't know anything about Cohen, but I think Gintis is on the right track with his strong reciprocity approach.
Stuff like open source works for reasons that really show off the elegance of microeconomics as an analytical tool. So, the first thing to realize is that software is not homogeneous. There isn't a software market; there's lots of markets for different kinds of software, such as operating systems, word processors, text editors, spreadsheets, databases, music players, web browsers, compilers, and so on. The second thing is that the marginal cost of software is very close to zero — the cost of making a copy of a piece of software is the cost of transmitting it over a network. The third thing is that when you have a heterogeneous market, the phenomena of complement and substitute come into play. So, Firefox and Internet Explorer are substitutes — if you have one, your demand for the other one is reduced. But a web server and a web browser are complementary goods. If the total number of web servers demanded goes up, then demand for web browsers will also go up, because the number of interesting sites has gone up (we hope!). Now, connect this phenomenon of complementarity with the low cost of reproduction. If you are a supplier of software, you know demand for your product will increase, if the price of its complements goes down — because when price falls, demand goes up. Now, since the cost of reproduction is basically zero, you can invest a lump sum in software to produce a complement good, and give it away for free, in order to maximize the demand increase from cheap complements. Now, since this is a locally rational decision, and you've got lots of suppliers of software, you can get a runaway effect in which lots of kinds of software become free. In reality, things are a bit more complicated — open source projects usually start because of the professional ethos and generosity of some engineer. But when they start to get larger, the dynamic I outlined tends to play an important role in securing the funding for continued development.
I should add that this analysis is probably due to either Yochai Benkler or Hal Varian.
That's a pretty interesting analysis, but I'm not sure its really that complicated. Commercial software companies, even those of us who make software with very high value to our customers, are constantly fighting to protect our margins. IP “leaks” out over time, so where right now only maybe 20 people know how to write a given algorithm, even if its never published, in a few years those guys will have gone to other companies and spread the knowledge around. Ultimately the possible sales price of certain things – multitasking Unix kernels, for instance – fall to zero, because the cost of reproduction is near-zero and the engineering involved is well-known.
There's definitely something to that, but writing a Unix kernel, for example, was quite literally a student exercise (in Andy Tanenbaum's book) when Linus Torvalds started writing Linux, but Unix kernels still weren't free, or even low cost. I tend to chalk this up to the fact that markets aren't perfect, and that the knowledge discovery aspect of markets really does require for people to try things out and convince other people that potential business models are actually feasible. But I do think this story of complements explains why firms like IBM and Sun were two of the biggest boosters for free software: they made their money from complements to software (services and hardware, respectively). Once it became clear that the theoretical story worked empirically, they could make the business case to themselves very easily. As an aside, since we're talking about Gintis, the success of the GPL probably does have something to do with tickling the reciprocity norms he's always going on about.
Well, it'd be nice to see some, um, empirical research into these questions, because as it stands now the conclusion really is baked into the assumption. I mean, Frey doesn't even wave a hand lazily in the direction of the caveats you've just offered, or any other for that matter, and here's Will saying that Frey proves that bringing weapons to political events is a great idea.I'm open to actual evidence/analysis that shows that politicians are overprotected, but my attitude towards packing the conclusion of an argument into the premise is less, shall we say, situational than Will's appears to be. Seriously, he stops just shy of accusing Akerlof and Shiller of intellectual dishonesty because they omit data the he thinks is important from their analysis and hops straight from there into praising Frey's completely data free assertions.It's not really a big deal in the larger scheme of things, but it does make Will's praise of Cohen here essentially content free.
“We may praise these engineers for challenging uncertain belief”Why would we do that? They didn't challenge a claim, they made absolute statements without sufficient support to justify them.We don't need to develop (have developed) supersonic flight to say that they're morons. The moment we detect that their assertions outstrip their arguments is when we're free to do that.
How would you suggest one do said empirical research? Pick a bunch of politicians, randomly decide which ones to assassinate, and measure the outcome after doing so? :-)There's a weird observer bias issue here which is that as long as the populace is deluded into thinking it matters who is in the White House there will be a huge cost to assassination in the form of lost productivity from people wailing and gnashing their teeth and worrying about What It All Means and how we will recover from Our National Tragedy whenever it happens. Whereas if assassinations were something that happened routinely this cost would completely disappear. “President Jones killed, replaced by President Smith” could be a small human interest story, like a ballplayer retiring or getting traded to a different team. Or like some event triggering CEO or CFO succession in a company whose products you occasionally buy. I don't know how one should analyze that – a situation where something clearly doesn't matter much except to the degree that people *believe* that it matters, which they do, so it does. Hoever one does, I'm pretty sure that factor far outweighs any other cost. Certainly the difference in expected policy results between one guy and the next guyfrom the same party with the same support staff and the same advisors can't be all that large.
Look, these theoretical imponderables are great and all, but nobody making the argument has actually produced any numbers concerning how much we actually spend as even a starting point for discussion. That puts this entire exercise firmly into BS/wanking territory without even considering the stuff that's more difficult to analyze. If you or Will or Frey think we're spending too much, well, how much is that, exactly?And that's not even what really bothers me here. Frey has drawn some pretty pictures suitable basically for a barroom napkin (nothing wrong with that!), but Will, in a bid to commit the largest fallacy of misplaced concreteness ever (on a dare, maybe?), extrapolates from this that allowing guns at political events is a good idea. He then has the balls to accuse those who disagree with him of not thinking clearly. His opinions on the finer points of rhetoric and his pious defenses of intellectual honesty can take a long walk on a short pier as far as I'm concerned.And another point on substitution: I was pretty sure, after the past 8 years, that no one agrees with Nader anymore that it just really doesn't matter who's in charge, but apparently he's still got some followers on this one. I'm also pretty sure that I don't want the government run by people who are either deathly afraid or don't really care that they're likely to get whacked any day now. YMMV, of course.
So the problem's that friends give each other gifts, but they don't give strangers gifts. Problem solved: Take the hatred that German Gentiles feel for German Jews, and make German Gentiles friends, blood brothers, partners in crime, united by hatred against German Jews. Call it National Socialism. German Gentiles will feel bound to each other, a nation in peril, under siege from their victims who they imagine are not victims but victimisers, oppressors, conspirators in a plot, to do to Us what is done to Them.”In my view, the principal problem that faces the socialist ideal is that we do not know how to design the machinery that would make it run.” Not so. The machine was built. It ran. German Gentiles stole from German Jews (then Europe's Jews, then everyone) and made gifts to each other, like friends, in Hitler's People's State.
I suppose to some extent it depends on what you mean by “socialist”. Take for example the maxim “to each according to his needs, from each according to his ability”. Open source doesn't achieve that goal, someone with a lot of ability is perfectly free to chose to not write software.Another definition of socialism is “worker ownership of the means of production”, which rather conflicts with the open source attitude of software being free, although I suppose worker ownership of the computing hardware is common. A third definition is “public ownership of the means of production” and I don't see how that fits in with open source at all. Turning away from software to markets with a more traditional positive marginal cost of production, the advantage of markets is that it provides a guide as to what goods to be produced. Lets for example consider producing food. Food is different to software in that if I eat an apple you can't also eat the same apple, so it's a rivalous good. Now agriculture is a complex activity with a myraid of different production possibilites. There are numerous crops that can be grown, mostly with differing requirements for inputs (eg fertilisers, varying amounts of labour using different sorts of skills, for example building a drip-feed watering system is quite different to harvesting apples). What sort of crops a piece of land can support is massively varied, for example the same piece of land could grow one crop superbly and another adequately with extra labour assistance, land varies in details as small as a paddock, it can be changed by various investments, for example building a warm wall to shelter tender plants from bad weather. Land also varies in its distance from consumers and thus in the transport costs, and in the form of transport costs, for example land near to the sea opens up the possibility of water transport. People vary in their preferences for foods, from the extreme of some people risking death if they eat something like peanuts, to minor cases of feeling that, “well, rice is nice, but I wouldn't want to eat it every night of the week.” There are some specific subgroups, for example people who can't consume gluten, so even if wheat is the best crop to grow absolutely everywhere in terms of calories per unit of input it makes sense to grow some crops other than wheat (I know of no reason to believe that wheat is the best crop to grow absolutely everywhere, this is perfectly hypothetical). The advantage of allocating agricultural decisions by markets is that it allows all this to be taken into account without requiring ridiculously large amounts of knowledge for individual decision-makers. Farmers have a reasonable idea of the production possibilites they face and can consider these against the price they would likely receive for growing a particular crop. Consumers don't need to consider all the ins and outs of various cropping possibilities, instead they merely consider if a particular food is worth more to them than the money it would cost, and concentrate their attention on things like their dayjobs, raising kids, writing open source software, arguing on Internet forums, etc. And as the world changes, prices change along with them. If a sudden storm wipes out half the apple supply prices rise until consumers eat half as many apples as before (or more apples are produced from somewhere, eg somewhere further away with higher transport costs). This is why many economists have specific views about how we should intervene in the market, if we want to. For example, if our concern is that some people can't afford food then it makes sense to give them money to buy food (or food vouchers) rather than redirecting farming, as the poor can then take into account all their varied tastes in their purchases and this will be communicated to farmers. A problem with some transport methods is that they impose externalities from air pollution, but a tax (or a cap-and-trade system) on the source of the method signals to farmers and consumers how to best adjust their production or consumption to reduce those costs, for example if it's very cheap for farmers to switch to a different transport method they can do so and keep supplying the same crops, but if it's very expensive it might make sense to let land a long distance away revert to nature and adopt more labour-intensive farming methods that use less transport. Or it might make sense for consumers to switch to crops that have less transport costs. Or plausibly a combination of them all. It seems unlikely that anyone will invent another method of disemminating the information about customer preferences and production possibilites that prices can allow.
I suppose to some extent it depends on what you mean by “socialist”. Take for example the maxim “to each according to his needs, from each according to his ability”. Open source doesn't achieve that goal, someone with a lot of ability is perfectly free to chose to not write software.Another definition of socialism is “worker ownership of the means of production”, which rather conflicts with the open source attitude of software being free, although I suppose worker ownership of the computing hardware is common. A third definition is “public ownership of the means of production” and I don't see how that fits in with open source at all. Turning away from software to markets with a more traditional positive marginal cost of production, the advantage of markets is that it provides a guide as to what goods to be produced. Lets for example consider producing food. Food is different to software in that if I eat an apple you can't also eat the same apple, so it's a rivalous good. Now agriculture is a complex activity with a myraid of different production possibilites. There are numerous crops that can be grown, mostly with differing requirements for inputs (eg fertilisers, varying amounts of labour using different sorts of skills, for example building a drip-feed watering system is quite different to harvesting apples). What sort of crops a piece of land can support is massively varied, for example the same piece of land could grow one crop superbly and another adequately with extra labour assistance, land varies in details as small as a paddock, it can be changed by various investments, for example building a warm wall to shelter tender plants from bad weather. Land also varies in its distance from consumers and thus in the transport costs, and in the form of transport costs, for example land near to the sea opens up the possibility of water transport. People vary in their preferences for foods, from the extreme of some people risking death if they eat something like peanuts, to minor cases of feeling that, “well, rice is nice, but I wouldn't want to eat it every night of the week.” There are some specific subgroups, for example people who can't consume gluten, so even if wheat is the best crop to grow absolutely everywhere in terms of calories per unit of input it makes sense to grow some crops other than wheat (I know of no reason to believe that wheat is the best crop to grow absolutely everywhere, this is perfectly hypothetical). The advantage of allocating agricultural decisions by markets is that it allows all this to be taken into account without requiring ridiculously large amounts of knowledge for individual decision-makers. Farmers have a reasonable idea of the production possibilites they face and can consider these against the price they would likely receive for growing a particular crop. Consumers don't need to consider all the ins and outs of various cropping possibilities, instead they merely consider if a particular food is worth more to them than the money it would cost, and concentrate their attention on things like their dayjobs, raising kids, writing open source software, arguing on Internet forums, etc. And as the world changes, prices change along with them. If a sudden storm wipes out half the apple supply prices rise until consumers eat half as many apples as before (or more apples are produced from somewhere, eg somewhere further away with higher transport costs). This is why many economists have specific views about how we should intervene in the market, if we want to. For example, if our concern is that some people can't afford food then it makes sense to give them money to buy food (or food vouchers) rather than redirecting farming, as the poor can then take into account all their varied tastes in their purchases and this will be communicated to farmers. A problem with some transport methods is that they impose externalities from air pollution, but a tax (or a cap-and-trade system) on the source of the method signals to farmers and consumers how to best adjust their production or consumption to reduce those costs, for example if it's very cheap for farmers to switch to a different transport method they can do so and keep supplying the same crops, but if it's very expensive it might make sense to let land a long distance away revert to nature and adopt more labour-intensive farming methods that use less transport. Or it might make sense for consumers to switch to crops that have less transport costs. Or plausibly a combination of them all. It seems unlikely that anyone will invent another method of disemminating the information about customer preferences and production possibilites that prices can allow.