Another Take on Robots, in Light of Marx and the TSSI

Apparently, while I’ve been messing around with fiction and farce, there has been a discussion going on in the economic blogosphere about the impact of robots. I owe this awareness to Magpie, who has put up a good post at his blog. Links to some of the prominent contributions in the debate can be found in Magpie’s post. Usually I would provide the links here also, and perhaps write an in-depth post of my own on the topic, but for once heteconomist, it seems, is ahead of the game, or, more likely, was behind the game last time it was played. So, instead of putting up a new post dealing with the topic in depth, I will simply draw attention to a previous post that relates to some aspects of the present discussion, and make a couple of cursory remarks:

Implications of a Purely Mechanized Economy

In relation to the post, I will note the following:

1. Krugman’s answer, in “Technology and Wages, the Analytics (Wonkish)”, to whether “rapid productivity growth is necessarily jobs- or wage-destroying” is that, “It all depends”. He arrives at this view by making reference to a neoclassical aggregate production function, which of course is a no-no around here but par for the course in the mainstream. Even so, the conclusion “It all depends” is the same as the one I draw on the basis of the temporal single-system interpretation of Marx.

2. Magpie, in relation to the present discussion, considers Marx’s contention that labor is the source of all value. In my earlier post, I found this a useful starting point for thinking about whether capitalism will be viable in a highly mechanized economy, and found, much like the answer to question 1 above, that “It all depends”.


22 thoughts on “Another Take on Robots, in Light of Marx and the TSSI

  1. AI can produce new value by producing an internal change. The capital value of an AI changes unexpectedly when it learns. We already have this concept for humans, i.e. human capital.

    Capital goods are increasingly virtual. There’s a major land grab going on with intellectual property rights. Corporations have won the right to patent existing life and have declared patent war. There needs to be more debate and not just unthinking protest. These enclosures cannot be fought without understanding.

  2. Thanks for your thoughts. In Marx’s terms, AI, other capital goods, animals, etc., are all productive of new real wealth, just not new value. Surplus value going to capitalists, rentiers, etc., is due to ownership of capital goods or land, not a productive contribution made by those individuals. The value they receive is the result of surplus labor (unpaid labor).

    Value, by definition for Marx, is socially necessary labor time or its monetary equivalent. Technical progress temporarily frees up some labor time that can then be re-employed elsewhere in the economy. Until this labor is re-employed elsewhere, which will be contingent on effective demand, value produced in aggregate actually falls (employment has fallen), even though real wealth (the real stuff produced), in all likelihood, has increased, thanks to the increased productivity made possible by the technical progress.

    These enclosures cannot be fought without understanding.

    I agree. But there is no hope of understanding through a neoclassical marginal productivity theory of distribution (e.g. Krugman’s appeal to an aggregate production function). That is a non-starter (capital debates).

    The starting point, in my opinion, should be that all means of production are to be considered part of the commons. These means of production were produced with past labor in combination with capital goods (not capitalists) and nature (not rentiers), both of which should be common to all.

  3. Thanks, Pete, for the kind words.

    But tell me with honesty what’s your opinion of the argument itself? Mandel meant that as a serious proof of the basics of Marx’s Law of Value.

    According to Mandel, there are three main proofs: Marx’s own, the reductio ad absurdum and the analytical one (essentially, to consider added values and that profits are just unpaid labour).

    I liked this one better. Marx’s is long and convoluted. The analytical one is simple enough and deals with notions most people already understand, but depends on a previous demonstration (that profits is surplus value expressed in money)

    To me, the reductio ad absurdum is convincing, if unorthodox. I also liked it because it is a simple and straightforward logical proof and neoclassical economists are supposed to suffer from Ricardo’s vice.

    Do you find it convincing? Do you identify any weakness, or fallacy in the argument?

  4. Incidentally, while having a quick look at Boehm-Bawerk’s critique, I noticed he references only Marx’s own proof. And Bortkiewicz seems to be replying to Boehm-Bawerk.

    That’s another reason why I prefer the reduction ad absurdum proof. Because Marx’s proof is long and convoluted and, as Mandel says, can be misinterpreted.

  5. To argue that the labour theory of value is essentially correct is to take a platonic view of essences rather than viewing the theory as a model of a more complex reality. The first problems encountered by neo-platonists are Godel’s incompleteness theorems. Then there’s the infinite regress problem which you’ve probably considered. There’s an obvious dissonance here because it seems likely that humans evolved from small burrowing mammals. Then there’s the problem that Marx had some concept of incompleteness and I don’t think he ever formed a decisive opinion. Even though the formal proof wasn’t developed until the 20th century, the concept of incompleteness existed informally in preceding centuries as a mathematical belief (I think Leibniz warned on this). This suggests that Marx described an economic game which approximates to production. Games are closed systems and are all the way up and down by definition. When discussing real production it’s therefore useful to induce new understanding as well as deduce predictions.

    Going back to your last post on growth. There’s been a debate on the lack of predictability of growth theories and there’s quite a few people who believe that the problem is the inability to model human learning. My argument is that workers always own part of the capital in the form of knowledge. Workers create this capital by learning. Learning is a mixture of passive learning (being told something) and active learning (figuring something out for yourself). Humans create new value when they learn actively by changing their internal state. As I see it, the counter argument to this is to say that learning and labour are separate processes. Making this argument dismisses the learning by doing approach. From the perspective of computer science, intelligence and evolution are very similar. We see them both as learning processes. We model intelligence with neural nets and evolution with evolutionary computation (genetic programming, genetic algorithms etc.). We also have mixed and novel techniques that incorporate these ideas.

    From this perspective, all new information comes from learning by doing, an approach that evolution has been following for 4 billion years. Therefore labour produces an internal change in workers because workers are not mindless machines. Production is therefore not just a transfer function because all production causes a change in state within workers. When life does things it learns things. This has led some people within computer science to argue that AI should be embodied. I don’t agree with this but that’s another debate.

  6. Magpie,

    Sorry I haven’t posted any comments directly on your blog. I don’t have a suitable ID at the moment. I thought Krugman’s post was a tautology but I didn’t think it was interesting enough to comment.

  7. Magpie, I agree with Mandel’s reasoning but am not sure whether, strictly, it constitutes a “proof”, or whether any argument could constitute a proof for a theory of value.

    Here is my take on the scenario you describe in your post.

    First and foremost, I agree that under 100 percent mechanized production, capitalism, at least in Marx’s sense, would cease to exist, and argued as much in my own post on purely mechanized production.

    Under pure mechanization, as you explain, there might still be a class of private owners (of means of production) and another class of non-owners. If so, the basis for pricing when it came to transactions within the owner class would be different than the basis for pricing in exchanges between the owner and non-owner classes.

    In this respect, I think that both yourself and Brendan (kapitalism101) make valid points in your current discussion.

    Specifically, I think that you are correct that, in the absence of government, prices would tend to zero when it came to transactions between the owner and non-owner classes, which implies that such exchanges would not occur except possibly as acts of benevolence on the part of owners.

    I think Brendan is correct that when it came to transactions within the owner class, prices would have a different basis, largely influenced by power imbalances among owners along with technical conditions of production. It is hard to say anything definitive about pricing within the owner sector unless there happened to be free competition, in which case only the technical conditions of production would matter.

    The absence of surplus value would not mean, of course, that the owners were having a hard time of it. With no effort — and solely due to the pattern of ownership — they would accumulate physical output, courtesy of the robots and private ownership of resources, which they could freely exchange among themselves as they saw fit. It would not be capitalism because there would not be any surplus value. But they would be living in abundance.

    The rest of us (non-owners) would be screwed unless the government stepped in and either: (i) taxed in physical form (confiscated) some of the output produced at the cost of no human effort and distributed it to the non-owners; or (ii) issued currency to the non-owners and imposed lump-sum tax obligations on the owners, to ensure the owners would be willing to sell some consumption items to non-owners for currency in order to meet their tax obligations.

    The effect of approach (ii) on transactions between the owner and non-owner classes would be to cause positive prices, since now it would be necessary for owners to cover the costs associated with the tax obligation.

    Needless to say, it would be better, from a socialist or communist perspective, for the means of production to be owned in common.

  8. I should have said predictive, not predictable. Hopefully my post can still be understood.

    On a more important issue, I would like to be able to convince people that knowledge is a form of real capital. I believe that this point is fundamental to a discussion of virtual goods. I don’t think it should be a difficult point to accept because knowledge is used in production and it isn’t destroyed by production (in fact the opposite, knowledge is increased by production). Is it hard to accept this point because knowledge is owned by workers and not capitalists?

  9. @Hacky,

    Thanks for the comments. I’ll keep that in mind.


    Thanks for the comments.

    “Specifically, I think that you are correct that, in the absence of government, prices would tend to zero when it came to transactions between the owner and non-owner classes, which implies that such exchanges would not occur except possibly as acts of benevolence on the part of owners.”

    The idea of depending on the goodness of strangers doesn’t offer me much comfort. Call me pessimistic if you will, but my hunch, and I admit it it is only a hunch, is extinction at least of our side.

    And I’m talking of extinction in the same sense that we speak of the extinction of wildlife or American, African and Asian historical cultures.

    Before proceeding, I am not saying that it is inevitable that we reach that stage. Perhaps it is not even likely that we reach it. But if we do reach it, that is my bet.

    “I think Brendan is correct that when it came to transactions within the owner class, prices would have a different basis, largely influenced by power imbalances among owners along with technical conditions of production. It is hard to say anything definitive about pricing within the owner sector unless there happened to be free competition, in which case only the technical conditions of production would matter.”

    Although I have my doubts (former capitalists would have no reason to keep producing and don’t really need to do it), I’ll admit Brendan’s outcome is not impossible. And perhaps new classes, with their own class conflicts, somehow will evolve and history will follow its course.

    If I had to guess, however, I would bet on stagnation, decadence and fragmentation. Whether it would be permanent or temporary, or how short or long (in case it were temporary), I cannot possibly say. But I wouldn’t rule out it was permanent and terminal.

    If you ask me, if that is how our species meets its fate (and that seems to be the fate of all species), then the universe is not losing much.

  10. One of the current fears of software developers is that we won’t be able to build software without receiving threats by patent trolls. This is primarily a problem with American patent law because software cannot be patented in the UK. The debate on intellectual property up to now has been farcical with media companies accusing everyone of theft and piracy; and libertarians claiming that everything should be free. The arguments are heated and inflammatory. Both sides use arguments that are misleading. Media companies don’t acknowledge that copyright violation is not legally considered theft because it doesn’t deprive the owner of property. The libertarian argument doesn’t acknowledge that open source communities use intellectual property law to protect their rights (e.g. GPL or Creative Commons).

    I know how to build robots and this knowledge is mine by possession (it can’t be removed from my head by force). The economic value of virtual goods has been growing relative to physical goods and ownership of virtual real capital is absolutely crucial for capitalists. This is a problem for capitalists because the way to own knowledge is to learn, not to buy. This has been one of the factors causing investment substitution and debt because people cannot legally be owned but they can be indentured. Debt is only a partial tactic which is subject to demand for debt. Direct ownership of virtual capital would be a much more successful tactic if it were possible. If corporations are allowed legal property rights on all virtual capital then I would not be allowed to release a software program without paying rent for the ideas that I’ve used. Paying rent to use knowledge may seem fair if someone else created the knowledge but people frequently invent similar concepts independently.

    What are the legitimate rights of intellectual property? I can maintain ownership of my knowledge by keeping secrets. This is ownership by physical possession. How would enacting the same right in law change the situation? It would take away the right of someone else to have the same ideas independently. Therefore, intellectual property is a bargain between the right to protect my knowledge and the right of someone else to have the same knowledge independently (people with dependent knowledge are unaffected because they are also deprived of knowledge by keeping secrets). Intellectual property rights have traditionally been highly weighted towards the commons but this is because they have been seen as having little economic value. That is no longer the case. One way or another, intellectual property will become the main form of ownership and the debate is being led by corporate interests. If intellectual property rights are mangled by corporate ownership then people will resort to secrets and this is highly inefficient and damaging to society. We need rights that will encourage everyone to share information. People will share more freely when they feel more secure. As with many institutional issues, sharing is a coordination game.

    A patent lawsuit is an implicit threat of violence because of the state monopoly of violence. Capitalism couldn’t exist without the state monopoly of violence. Without this threat, violence becomes the means of possession by appropriation. With regard to property law and state violence, I’ve tended to subscribe to Karl Popper’s tolerance of intolerance argument and Proudhon’s idea of legitimate property being created by labour and illegitimate property being taken by force. I take the view that human rights are preferable to ad hoc utilitarianism because ad hoc arguments tend to be subjective and self interested. There’s also the problem of understanding which rights create the best incentives (with reference to game theory problems). Rights should also be for people not legal personalities (though this battle has been lost for the time being).

    I agree with the preference for substantially collective ownership but believe that this should be based on a system of rights that are intelligible to most people. Contrary to popular belief, commoners were not the lowest of the low. They were people with legal rights to the commons. At a higher level of abstraction, the law is just a rule based system of trust and other methods may work better. My main concern is the creation of systems that work.

  11. Hey Pete,

    Elsewhere Ramanan told me about a Kaleckian “paradox of costs”. The only info I managed to get comes from an encyclopedia of economics or something.

    Do you know more about it? Like where did Kalecki mentioned it?

  12. Magpie, sorry for the delay in responding. You caught me napping in the sense that I’m not sure the term was actually used by Kalecki himself, and I have no internet access to his works, so investigating the matter would involve a trek to the university library. My guess, from a quick google search, is that the term came later with a 1981 paper by Rowthorn, but I could be wrong.

    The notion itself, if I’m not mistaken, refers to the possibility that a reduction in the aggregate markup, corresponding to an increase in the real wage, will result in higher consumption demand, higher capacity utilization, higher realized aggregate profit, a higher realized rate of profit, and therefore greater impetus for accumulation.

    Below is a link to a paper by Lavoie (always worth reading) that appears to include a consideration of similarities and differences between Kaleckian and Sraffian (and also earlier Post Keynesian) approaches to accumulation. I have not yet read it myself, but am in the process of doing so.

    Kaleckian Effective Demand and Sraffian Normal Prices : Towards a Reconciliation

  13. Hey Pete

    Nick Rowe has caught the “robot fever”! From his post on “Production of Robots by means of Robots” [*]:


    Assume that all capital is robots, and robots are perfect substitutes for human workers.”


    “Let’s measure wages in terms of consumption goods. Because consumption is what people care about. Robots and humans earn the same wages.”


    “In this simple model, improving technology for producing robots has no effect whatsoever on wages.

    “Not at all nightmarish, is it?”

    So, if one assumes that robots are perfect substitutes for workers and they earn the same wages (!!!!!)… what is it that makes robots, well, different from humans? I mean, other than Prof. Rowe’s word.


  14. Pete,

    While some speak of robotic and human workers as being perfect substitutes and earning equal wages (!!!) the BBC went to see how an automated warehouse for e-commerce actually runs:

    Logistics: Rise of the warehouse robots

    Disclaimer: The academic account is entirely fictional. Any similarity with real automation is mere coincidence.

  15. Hey Peter

    I haven’t checked this paper yet, but the abstract seems interesting, RE: our discussion with K101

    Marx’s Critique of (Ricardian) Political Economy, the Quantity Theory of Money and Credit Money.
    John Milios
    Department of Humanities, Social Sciences and Law,
    National Technical University of Athens
    The Marxist concept of value is very frequently equated, whether explicitly or merely tacitly, with the corresponding Ricardian concept of “labour expended”. This paper argues that unlike the Ricardian theory of value, the Marxist theory of value is a monetary theory. In the Marxist system, the value of a commodity is expressed not through itself but through its distorted forms of appearance, in prices. Moreover, it cannot be defined in isolation, but exclusively in relation to all other commodities, in a process of exchange. In this relation of exchange value is materialised in money. The essential feature of the “market economy” (of capitalism) is thus not simply commodity exchange but monetary circulation and money. Commodity exchange presupposes thus the (positive) prices of all commodities involved. In other words, prices are not determined after the establishment of a non-monetary equilibrium system of barter between “production sectors”, like the Sraffian “linear production systems”. On the contrary, barter is for Marx non-existing, as all exchange transactions are made up of separate acts of exchange of commodities with money.
    Prices are determined in the process of capitalist commodity production, i.e. in a historically unique process of (capitalist) production-for-the-exchange, a process which unites immediate production with circulation. Money is thus conceived as the adequate form of appearance of capital, that is a material embodiment of abstract and therefore equal human labour, which the capitalist appropriates, and which in the framework of capitalist relations of exploitation is accumulated and functions as a “self-valorising value”.
    Only these Marxian concepts of value and money enable, on the one hand, a radical critique to the Quantity Theory of money, and on the other an insight into the process of credit-money formation, in the framework of the reproduction and circulation of the total social capital.

  16. In the “rationing scarcity by price” model of distribution, use value is assumed to be identical with exchange value as determined by price. The question then becomes how to either prove it or disprove it.

    This is significant in that the rationale of capitalism is dependent on price discovery in free markets. Of course, there are other issues with this assumption, too, but Marx noticed that there was something quirky going on with “value,” and he was correct.

  17. Tom,

    “In the “rationing scarcity by price” model of distribution, use value is assumed to be identical with exchange value as determined by price. The question then becomes how to either prove it or disprove it.”

    I am not entirely sure what the rationing scarcity by price model of distribution is, but I’ll suppose you mean the marginalist model of supply and demand (dependent on the notions of scarcity and utility). So, the question would be how to either prove or disprove it.

    It depends on what one considers to prove or disprove a model. One may consider proving/disproving a model on logical, a priori grounds, or on empirical ones. Some (like some strains of Austrianism) consider the logical way the only way of disproving a model; other strains (like the more mainstream neoclassicals) although accept other ways of disproving their model, also admit this possibility.

    To disprove the marginalist model from a logical point of view means to show it either contains false premises or that the conclusions derived from it do not follow logically from the premises.

    However, to logically disprove the marginalist model becomes more complicated than that, for at least two reasons.

    Firstly, Austrians, at least those closer to Mises, define the premises in a ambiguous and vague fashion: what they call the Action Axiom. This makes it very difficult to show it false: essentially anything could be justified with the Action Axiom.

    Secondly, the rest (including mainstream neoclassicals) defend the marginalist model insisting now (a la Milton Friedman) that the realism of the premises (i.e. their correspondence with reality) is not important, predictive capacity (i.e. the correctness of the conclusions derived from the premises) is. (As I see things, this is nonsense, from a purely logical perspective, but there you have it)

    Is my understanding that, in principle, this is the position of the more positivist strain of mainstream neoclassical economics.

    This leads us to the empirical interpretation of “proving”/”disproving” a model. From an empirical perspective to disprove the marginalist model seems much easier: it only needs to be shown its predictions, derived from its premises, are false (i.e. they do not correspond to reality).

    The problem now lies on a “free of jail” card they’ve pulled out of their sleeves: it’s not that the marginalist model fails, but that government intervention interferes with it. That’s the “law of unintended consequences”.

    To sum things up: the whole marginalist edifice is very difficult to assault. In part, because a whole fauna has evolved from it, over about a century and a half, and it is nearly impossible to find a magic bullet for all the species in this fauna.

    Moreover, the sign posts keep changing: when the realism of the assumptions is challenged, then it doesn’t matter, what matters are the predictions/conclusions. When the predictions fail to materialize, then it’s because the government’s intervention has unintended consequences.

    At most some admit it failed, but with the modifications they propose it won’t fail again.

    In a libertarian utopia, with no government, some excuse would have to be found and although I can’t conceive of it now, I am sure they would come up with something. By that time, however, chances are there will be very little concern for theoretical niceties: if you raise questions, they could just burn the heretic and be done with it.

    In this the marginalist model tends to resemble religion: it’s impossible to prove that gods, afterlife, heaven and hell, miracles do not exist. Hopefully, with the marginalist model we haven’t yet reached the point where it is absolutely impossible.

    But, as with religion, the lack of a non-existence proof is evidence enough for the faithfuls.


    But I’m intrigued by the rest of your comment (“Marx noticed that there was something quirky going on with “value,” and he was correct.”). What do you have in mind?

  18. Magpie, since Marx predated marginalism, he was thinking in other terms, which I have called the rationing scarcity by price model of distribution that is fundamental to market capitalism. The presumption is that market price always reflects the actual value of a good and that price discovery in markets enables efficiency in allocating scarcity.

    Marx objected that the exchange value of a good was quite different from the use value of a good. In a barter system where individuals exchange their surplus goods based on use value there is no profit motive, no profit extraction, and no ROI, which involves rent. With capitalism, everything changes. The market depersonalizes transactions, commodifies labor, introduces rent, and extracts surplus value as owners’ share, i.e., profit. Use value, which is the actual (real) value, drops out in assuming that there is only one value, the exchange value expressed as nominal price. It’s the sleight of hand that enables exploitation characteristics of capitalism as new form of privilege.

    In subsequent Marxian POV, the purpose of a genuine economics is to deconstruct mainstream orthodox economics and expose the sleight of hand and “ledgerdemain” going on there. It’s essentially a con game that allows parasitical extraction of actual value based on presuppositions about price discovery in markets being “efficient” in allocating scarcity. The purpose of the con game is to create a privileged class based on wealth and power, similar to the feudal system. In fact under capitalism, land is folded into capital as the dominant factor, with labor servile to capital.

  19. Robert Vienneau has an interesting post up today, Functions For Fashion, which is he asserts that mainstream economics is basically diversionary. It’s models don’t address actual issues and to enter the debate under that framing is a trap.

    “If you want to argue against mainstream economics, a mainstream economist can dismiss you as ignorant of some model variation and as attacking a strawperson. Furthermore, this dismissal could be “justified” by just checking whether you have a degree from a small number of schools, and, if you do, just mocking you as not having fully learned what they are teaching. Thus, your time can be taken up with argument about whether you know what you are talking about. The mainstream economist never need get to the point of engaging a critique.”

Comments are closed.