One quick thought on incentives and social science. The idea that people only respond to incentives in familiar textbook ways (eg, buying less of something when the price increases) when they recognize the incentives as such strikes me as a mistake.
There's a fruitful tradition in economics of evolutionary game theory, where the basic insight is that economic actors do *not* have to decide what to do by reference to concepts from economics in order for those concepts to play a role in predicting and explaining market outcomes.
If firms that make profits tend to grow, and firms that experience losses tend to shrink and go out of business, then even if managers are making decisions randomly, without anything like a conscious attempt to maximize profits, you should expect that over time the firms that remain in the market will be pursuing (roughly) game theoretically optimal strategies. This is analogous to the idea that animals will tend to behave in "rational" ways (eg, not leaving calories or mating opportunities on the table) even though they're usually not consciously optimizing; conspecifics in earlier generations who pursued less effective strategies failed to reproduce.
Thanks Dan! I don't really know that Friedman disagrees with this, at least insofar as he seems on-board with the idea that market competition predictably produces good results, and his reasons why seem exactly like the sort of thing an economist would say. What he is criticising is a different idea: that people respond to what a theorist would say that the incentives are. He thinks this is wrong because the actor may view the incentives very differently. For my part, I think this point can bear the weight it needs to for his critique of technocracy, but it certainly can't justify his surprisingly sweeping dismissal of social science, at least a social science that sees itself as in the business of making predictions. I would assume that the critique of social science is secondary to the critique of technocracy, but I was quite surprised at this part of the book--it really did not seem supported by any of the arguments he gives.
I guess I was figuring that if the theorist is right about what the incentives are (big if, i know) then over time their predictions should come true (for broadly evolutionary reasons) even if the market participants conceive of their situation differently.
So the way I understand Friedman’s argument is broadly something like this: the problem with technocracy is that it relies on predictions that are often inaccurate and therefore it leads to all sorts of bad unintended consequences. This could be true even if over time the predictions get better for the reasons you suggest: a lot of damage happens while they are getting better, and even then they are still wrong in certain ways that continue to lead to problems. I find this a compelling critique of technocracy, especially when you view the failures of modern bureaucratic states through this lens. But yes I agree with you that it doesn’t seem like any of this gives us reasons to reject a predictive social science. I think Friedman is wrong if he thinks his argument against technocracy relies on his case against social science.
Maybe important to add: insofar as Friedman has a positive view it seems to be something like a small state libertarianism, with the odd quirk that he thinks that what would be far better than that is a world where wealth was redistributed on a massive scale. He also thinks that a bit of technocracy is required to maintain that so maybe you’re thinking we have the predictions we need to keep that going, in which case he maybe agrees with you more than it might seem.
Thank you for this very interesting summary and partial critique of Friedman's work. But I find it puzzling that Friedman (and McKenna) discuss these issues without referring to the unmissable work by Philip Tetlock and his collaborators on expert forecasters. (Which generally think in a totally different way from the overconfident 'experts' who serve as talking heads in the media) These expert forecasters are foxes rather than hedgehogs, sound something like Bayesian reasoners, constantly updating their predictions, and make probabilistic predictions. Laypeople who think this way can predict geopolitical events better than CIA analysts with access to classified information not available to the laypeople. But the predictions of even the best forecasters are not spectacularly better than chance - though they would do well in constantly repeated bets. And (if I remember correctly), if they have to predict events further in the future, at about six months even the best forecasters cease doing any better than chance. The other thing I miss in Friedman's account is the possibility of doing smaller scale experiments before introducing new policies, and in general the Popperian notion of piecemeal engineering - doing policy changes in small increments and seeing what happens after one smallish step before setting the next smallish step. I've just started reading Nate Silver's The Village and the River, and the River mindset (silicon Valley, poker players) may be related to that of Tetlock's Superforecasters. Successful venture capitalists must be able to predict the future well enough to have their hits outweigh their misses. The reason for their success is not that they are good at avoiding misses. I don't know whether rethinking in probabilistic terms all the issues Friedman discusses could lead to a less dire view than the one Friedman presents. But meanwhile kudos to him and McKenna for developing the view that we normally utterly overestimate what technocracy can achieve.
Thanks! Friedman does actually discuss Tetlock (I don't have the book to hand to say where). and he says exactly what you seem to think he should say: there may be people who are better at forecasting what will happen than others, but their forecasting powers are still relatively limited. In particular, their forecasts are not sufficiently reliable over the timespan required for successful policy-making. But he does seem to agree with Tetlock about the psychological profile that makes for good forecasting, and he contrasts this with what he sees as the psychological profile common among technocrats.
On smaller scale experiments: As I was reading the book I kept on wondering about this, as it really isn't something he discusses. There is a bit where he discusses a proposal from Dewey that would amount to controlled experimentation on the population: try something new, and keep all the other variables fixed so you can really see if it "worked". I agree with Friedman that actually trying this out woul be a form of madness. But he never discusses the prospects for much more limited forms of experimentation. I suspect the reason is that, while he says the book is an argument against technocracy as such, it is really (as I have explained in other comments) an argument against big and ambitious technocracy, of the sort that appeals to some contemporary liberals, and some democratic socialists of the past.
On venture capitalists: I don't know if Friedman says this, but it is exactly the sort of thing he would say. The problem with applying this sort of thinking to policy-making is that the consequences of the misses can be truly dire. If you, as a venture capitalist, make some bad investments then, so long as you also made some good ones, it can all balance out--or you can even end up ahead. If you do the same sort of thing as a goverment, you can really make a mess of things because the bad decisions you made have after-effects that you can't simply undo by making some other good decisions. This is one respect in which running a country is not like running a business or having an investment portfolio. You can't simply start a new country if you screw it up.
1. The better case for technocracy is that expertise is more likely to produce comparatively better solutions to social problems. Obviously, there are no guarantees. Unless Friedman really thinks all the models and concepts available to experts invariably fail to have advantageous effects on policy outcomes? (Even so, having experts make decisions creates feedback loops which test and refine expert knowledge over time—the optimist in me remains hopeful that progress is possible and desirable).
2. Here's an inversion of Friedman's concept of ideational heterogeneity: by instilling ideational homogeneity, systems can be made more predictable—indeed, it's arguably what makes prediction possible at all in the social sciences. Arendt made a similar point about how economics (at her time) could achieve a scientific character only with the rise of mass society and its levelling effects, the image of communities as societies of jobholders, the widespread conception of what one does as primarily geared towards sustenance, etc.
On 2: I agree with that; this was the point I was trying to make when I talk about predicting voting behavior based on demographics. But, being charitable to Friedman, I was trying to say that the failure of his case against predictive social science doesn’t mean the failure of his case against technocracy.
On 1: I didn’t put this in the piece because it was already quite long. But another criticism I have of his book is that what he is really criticizing—or rather where his criticisms are most effective— is not technocracy as such but what you might call bold and ambitious technocracy. This comes out in the fact that, when he identifies the historical roots of what he is attacking, he focuses on early 20th century Progressivism, which it seems really was highly technocratic. So the target of the book is really ambitious technocracy: the idea that, using the tools of social science, modern statistics or whatever, we can reshape society in the way that we want.
On the expertise thing: the book is not dismissing the idea that experts have more knowledge, relevant skills etc. it’s rather making the point that experts have blind spots too, and tone of the dangers of expertise is that you can become over confident, and lose sight of what you don’t know. Friedman thinks that there’s a kind of selection bias here too: experts who are more confident and dogmatic will tend to have more sway with policy makers, so the basic problem is exacerbated. Fwiw I agree with the broad outlines of this, though as I say in the piece I think it’s a bit too simplistic.
I find it surprising that you are as favourable as you seem to be when Friedman is critiquing such an incredibly broad range of systems. A “polity that aims to solve, mitigate, or prevent social and economic problems among its people” is, to a first approximation, all polities. What polity is not technocratic under this understanding? Isn’t, for example, instituting laws and punishing law breakers an attempt to prevent various social problems? Does Friedman’s thesis not collapse into the vulgarest sort of anarchism?
I would have a lot more sympathy if there was some sort of attempt to distinguish between the sorts of social and economic problems that are amenable to solution and those which are for various reasons likely to be both difficult to understand and likely to involve unintended consequences. But I gather that is not the approach.
Maybe this should have been clearer in the review. It (the more limited sort of technocracy you envision) implicitly is the approach, though a lot of the time he sounds more radical than that. But (this is clearest in the last chapter) he seems to want a very limited technocracy, of the sort that would be amenable to a small government libertarian: you need a government (I think he thinks all government is to a degree technocratic) for things like defence, law and order, but not much else besides. Maybe he wants some kinds of regulations on business: he’s not worried about regulation because it stifles innovation as such, but more because it runs the usual risk of unintended consequences. It’s probably best to read the book as a critique of the ideals of social democracy, and of the role that technocracy had played in left wing politics. For myself, I buy that critique, but I don’t think this was the right way to defend social democracy in the first place.
Isn’t this just warmed over Burke, then? Trying to fix problems is very hard, these silly reformers get everything wrong because they think they are so smart? Admittedly, he prefers to trust individual initiative over and against the weight of tradition, but… the response is just the same as the response to Burke, which is that we can use our judgment to identify those areas where we have justified confidence in the prospects for positive change.
I feel like any successful critique of technocracy is going to need to be a lot more specific with the definition. And I very much doubt that there is a good general case to be made even with a more specific definition, because the social sciences do in fact gradually improve at practical prediction over time, contra what seems to be Friedman’s contention. So if we cannot expect to solve a problem very well yet, perhaps if we research it more we will eventually learn what to do.
Yes it is different from Burke in precisely that way: there is absolutely nothing in there about the weight of tradition. It's more like the opposite, in that his arguments would give us grounds for thinking that tradition is often wrong, and can breed certain kinds of closed-mindedness and dogmatism. I wouldn't say he trusts individual initiative, because it is pessimistic about positive change in the way you are worried about, whether that change might come from individual initiative or something else. The most optimistic he gets is when he talks about markets, where he sounds a bit like Hayek. But even there the value of markets is limited: they are good at providing options, and more options are good.
The reason I'm enthusiastic about it (this is what I was getting at with my previous reply) is that it has real value as a critique of our currently existing technocratic institutions, but also as a critique of common attitudes towards politics and policy making. Consider the common refrain that "we" (politicians, the people) know how to solve problems like generational poverty, increased housing, expensive or hard to access healthcare (not where I live, but in other parts of the world), decreasing fertility, or what have you, but what is stopping us is the fact that politicians are evil, corrupt, incompetent, etc. Now, I have a lot of sympathy for those who think that politicians and political institutions are corrupte(ed), and some of them are no doubt incompetent. But we're engaging in the pretence that solving these problems is easy. It isn't. It's hard, for the sorts of reasons that Friedman gives.
I totally agree with you that there is no good argument in here that social science can't predict things, or that it can't get better at prediction over time. I do agree with Friedman (many others also think this) that, in a certain sense, social science is more difficult than natural science, because predicting/explaining human action is just more difficult than predicting how inanimate things will "behave", or explaining why they "did" what they did. But that isn't enough to ground his radical scepticism.
I’m sorry you had to struggle through what sounds like an extraordinarily weak book.
Among all the weaknesses pointed out by other commenters with Friedman’s argument (as you’ve presented it): technocracy actually has a pretty great track record! It’s why we live in healthier, more prosperous, more free and capable societies than we ever used to!
At the end, you say that we might have to abandon consequentialist arguments for political systems. But the book and article aren’t about systems themselves, but prospective solutions within those systems. I think we can justify, say, liberal democracy in a consequentialist manner not by saying it will definitely solve future problems, but by pointing at its overall good past results and saying that we cannot reliably know that switching will produce good results. Technocratic skepticism points to Burkean Conservatism.
Fascinating stuff!
One quick thought on incentives and social science. The idea that people only respond to incentives in familiar textbook ways (eg, buying less of something when the price increases) when they recognize the incentives as such strikes me as a mistake.
There's a fruitful tradition in economics of evolutionary game theory, where the basic insight is that economic actors do *not* have to decide what to do by reference to concepts from economics in order for those concepts to play a role in predicting and explaining market outcomes.
If firms that make profits tend to grow, and firms that experience losses tend to shrink and go out of business, then even if managers are making decisions randomly, without anything like a conscious attempt to maximize profits, you should expect that over time the firms that remain in the market will be pursuing (roughly) game theoretically optimal strategies. This is analogous to the idea that animals will tend to behave in "rational" ways (eg, not leaving calories or mating opportunities on the table) even though they're usually not consciously optimizing; conspecifics in earlier generations who pursued less effective strategies failed to reproduce.
Thanks Dan! I don't really know that Friedman disagrees with this, at least insofar as he seems on-board with the idea that market competition predictably produces good results, and his reasons why seem exactly like the sort of thing an economist would say. What he is criticising is a different idea: that people respond to what a theorist would say that the incentives are. He thinks this is wrong because the actor may view the incentives very differently. For my part, I think this point can bear the weight it needs to for his critique of technocracy, but it certainly can't justify his surprisingly sweeping dismissal of social science, at least a social science that sees itself as in the business of making predictions. I would assume that the critique of social science is secondary to the critique of technocracy, but I was quite surprised at this part of the book--it really did not seem supported by any of the arguments he gives.
I guess I was figuring that if the theorist is right about what the incentives are (big if, i know) then over time their predictions should come true (for broadly evolutionary reasons) even if the market participants conceive of their situation differently.
So the way I understand Friedman’s argument is broadly something like this: the problem with technocracy is that it relies on predictions that are often inaccurate and therefore it leads to all sorts of bad unintended consequences. This could be true even if over time the predictions get better for the reasons you suggest: a lot of damage happens while they are getting better, and even then they are still wrong in certain ways that continue to lead to problems. I find this a compelling critique of technocracy, especially when you view the failures of modern bureaucratic states through this lens. But yes I agree with you that it doesn’t seem like any of this gives us reasons to reject a predictive social science. I think Friedman is wrong if he thinks his argument against technocracy relies on his case against social science.
Maybe important to add: insofar as Friedman has a positive view it seems to be something like a small state libertarianism, with the odd quirk that he thinks that what would be far better than that is a world where wealth was redistributed on a massive scale. He also thinks that a bit of technocracy is required to maintain that so maybe you’re thinking we have the predictions we need to keep that going, in which case he maybe agrees with you more than it might seem.
Thank you for this very interesting summary and partial critique of Friedman's work. But I find it puzzling that Friedman (and McKenna) discuss these issues without referring to the unmissable work by Philip Tetlock and his collaborators on expert forecasters. (Which generally think in a totally different way from the overconfident 'experts' who serve as talking heads in the media) These expert forecasters are foxes rather than hedgehogs, sound something like Bayesian reasoners, constantly updating their predictions, and make probabilistic predictions. Laypeople who think this way can predict geopolitical events better than CIA analysts with access to classified information not available to the laypeople. But the predictions of even the best forecasters are not spectacularly better than chance - though they would do well in constantly repeated bets. And (if I remember correctly), if they have to predict events further in the future, at about six months even the best forecasters cease doing any better than chance. The other thing I miss in Friedman's account is the possibility of doing smaller scale experiments before introducing new policies, and in general the Popperian notion of piecemeal engineering - doing policy changes in small increments and seeing what happens after one smallish step before setting the next smallish step. I've just started reading Nate Silver's The Village and the River, and the River mindset (silicon Valley, poker players) may be related to that of Tetlock's Superforecasters. Successful venture capitalists must be able to predict the future well enough to have their hits outweigh their misses. The reason for their success is not that they are good at avoiding misses. I don't know whether rethinking in probabilistic terms all the issues Friedman discusses could lead to a less dire view than the one Friedman presents. But meanwhile kudos to him and McKenna for developing the view that we normally utterly overestimate what technocracy can achieve.
Thanks! Friedman does actually discuss Tetlock (I don't have the book to hand to say where). and he says exactly what you seem to think he should say: there may be people who are better at forecasting what will happen than others, but their forecasting powers are still relatively limited. In particular, their forecasts are not sufficiently reliable over the timespan required for successful policy-making. But he does seem to agree with Tetlock about the psychological profile that makes for good forecasting, and he contrasts this with what he sees as the psychological profile common among technocrats.
On smaller scale experiments: As I was reading the book I kept on wondering about this, as it really isn't something he discusses. There is a bit where he discusses a proposal from Dewey that would amount to controlled experimentation on the population: try something new, and keep all the other variables fixed so you can really see if it "worked". I agree with Friedman that actually trying this out woul be a form of madness. But he never discusses the prospects for much more limited forms of experimentation. I suspect the reason is that, while he says the book is an argument against technocracy as such, it is really (as I have explained in other comments) an argument against big and ambitious technocracy, of the sort that appeals to some contemporary liberals, and some democratic socialists of the past.
On venture capitalists: I don't know if Friedman says this, but it is exactly the sort of thing he would say. The problem with applying this sort of thinking to policy-making is that the consequences of the misses can be truly dire. If you, as a venture capitalist, make some bad investments then, so long as you also made some good ones, it can all balance out--or you can even end up ahead. If you do the same sort of thing as a goverment, you can really make a mess of things because the bad decisions you made have after-effects that you can't simply undo by making some other good decisions. This is one respect in which running a country is not like running a business or having an investment portfolio. You can't simply start a new country if you screw it up.
1. The better case for technocracy is that expertise is more likely to produce comparatively better solutions to social problems. Obviously, there are no guarantees. Unless Friedman really thinks all the models and concepts available to experts invariably fail to have advantageous effects on policy outcomes? (Even so, having experts make decisions creates feedback loops which test and refine expert knowledge over time—the optimist in me remains hopeful that progress is possible and desirable).
2. Here's an inversion of Friedman's concept of ideational heterogeneity: by instilling ideational homogeneity, systems can be made more predictable—indeed, it's arguably what makes prediction possible at all in the social sciences. Arendt made a similar point about how economics (at her time) could achieve a scientific character only with the rise of mass society and its levelling effects, the image of communities as societies of jobholders, the widespread conception of what one does as primarily geared towards sustenance, etc.
On 2: I agree with that; this was the point I was trying to make when I talk about predicting voting behavior based on demographics. But, being charitable to Friedman, I was trying to say that the failure of his case against predictive social science doesn’t mean the failure of his case against technocracy.
On 1: I didn’t put this in the piece because it was already quite long. But another criticism I have of his book is that what he is really criticizing—or rather where his criticisms are most effective— is not technocracy as such but what you might call bold and ambitious technocracy. This comes out in the fact that, when he identifies the historical roots of what he is attacking, he focuses on early 20th century Progressivism, which it seems really was highly technocratic. So the target of the book is really ambitious technocracy: the idea that, using the tools of social science, modern statistics or whatever, we can reshape society in the way that we want.
On the expertise thing: the book is not dismissing the idea that experts have more knowledge, relevant skills etc. it’s rather making the point that experts have blind spots too, and tone of the dangers of expertise is that you can become over confident, and lose sight of what you don’t know. Friedman thinks that there’s a kind of selection bias here too: experts who are more confident and dogmatic will tend to have more sway with policy makers, so the basic problem is exacerbated. Fwiw I agree with the broad outlines of this, though as I say in the piece I think it’s a bit too simplistic.
I find it surprising that you are as favourable as you seem to be when Friedman is critiquing such an incredibly broad range of systems. A “polity that aims to solve, mitigate, or prevent social and economic problems among its people” is, to a first approximation, all polities. What polity is not technocratic under this understanding? Isn’t, for example, instituting laws and punishing law breakers an attempt to prevent various social problems? Does Friedman’s thesis not collapse into the vulgarest sort of anarchism?
I would have a lot more sympathy if there was some sort of attempt to distinguish between the sorts of social and economic problems that are amenable to solution and those which are for various reasons likely to be both difficult to understand and likely to involve unintended consequences. But I gather that is not the approach.
Maybe this should have been clearer in the review. It (the more limited sort of technocracy you envision) implicitly is the approach, though a lot of the time he sounds more radical than that. But (this is clearest in the last chapter) he seems to want a very limited technocracy, of the sort that would be amenable to a small government libertarian: you need a government (I think he thinks all government is to a degree technocratic) for things like defence, law and order, but not much else besides. Maybe he wants some kinds of regulations on business: he’s not worried about regulation because it stifles innovation as such, but more because it runs the usual risk of unintended consequences. It’s probably best to read the book as a critique of the ideals of social democracy, and of the role that technocracy had played in left wing politics. For myself, I buy that critique, but I don’t think this was the right way to defend social democracy in the first place.
Isn’t this just warmed over Burke, then? Trying to fix problems is very hard, these silly reformers get everything wrong because they think they are so smart? Admittedly, he prefers to trust individual initiative over and against the weight of tradition, but… the response is just the same as the response to Burke, which is that we can use our judgment to identify those areas where we have justified confidence in the prospects for positive change.
I feel like any successful critique of technocracy is going to need to be a lot more specific with the definition. And I very much doubt that there is a good general case to be made even with a more specific definition, because the social sciences do in fact gradually improve at practical prediction over time, contra what seems to be Friedman’s contention. So if we cannot expect to solve a problem very well yet, perhaps if we research it more we will eventually learn what to do.
Yes it is different from Burke in precisely that way: there is absolutely nothing in there about the weight of tradition. It's more like the opposite, in that his arguments would give us grounds for thinking that tradition is often wrong, and can breed certain kinds of closed-mindedness and dogmatism. I wouldn't say he trusts individual initiative, because it is pessimistic about positive change in the way you are worried about, whether that change might come from individual initiative or something else. The most optimistic he gets is when he talks about markets, where he sounds a bit like Hayek. But even there the value of markets is limited: they are good at providing options, and more options are good.
The reason I'm enthusiastic about it (this is what I was getting at with my previous reply) is that it has real value as a critique of our currently existing technocratic institutions, but also as a critique of common attitudes towards politics and policy making. Consider the common refrain that "we" (politicians, the people) know how to solve problems like generational poverty, increased housing, expensive or hard to access healthcare (not where I live, but in other parts of the world), decreasing fertility, or what have you, but what is stopping us is the fact that politicians are evil, corrupt, incompetent, etc. Now, I have a lot of sympathy for those who think that politicians and political institutions are corrupte(ed), and some of them are no doubt incompetent. But we're engaging in the pretence that solving these problems is easy. It isn't. It's hard, for the sorts of reasons that Friedman gives.
I totally agree with you that there is no good argument in here that social science can't predict things, or that it can't get better at prediction over time. I do agree with Friedman (many others also think this) that, in a certain sense, social science is more difficult than natural science, because predicting/explaining human action is just more difficult than predicting how inanimate things will "behave", or explaining why they "did" what they did. But that isn't enough to ground his radical scepticism.
But, as you say,
I’m sorry you had to struggle through what sounds like an extraordinarily weak book.
Among all the weaknesses pointed out by other commenters with Friedman’s argument (as you’ve presented it): technocracy actually has a pretty great track record! It’s why we live in healthier, more prosperous, more free and capable societies than we ever used to!
At the end, you say that we might have to abandon consequentialist arguments for political systems. But the book and article aren’t about systems themselves, but prospective solutions within those systems. I think we can justify, say, liberal democracy in a consequentialist manner not by saying it will definitely solve future problems, but by pointing at its overall good past results and saying that we cannot reliably know that switching will produce good results. Technocratic skepticism points to Burkean Conservatism.