Tessa Morris-Suzuki has opened up an interesting and necessary discussion of the consequences of automation in manufacturing and of the associated rapid growth of the software industries.footnote1 As she shows, using Japanese data, the use of robots, etc., is now developing at a speed which makes it important to think clearly about the theory of the automated economy. In her view, ‘such economies are not merely a theoretical possibility but are actually appearing before our eyes’ (p. 112). Technological forecasts are notoriously unreliable (and prone to fashion) but one does not need to accept any particular forecast concerning the date by which automation will be achieved—or, indeed, to accept even the forecast that it will be completely achieved—in order to agree with Morris-Suzuki’s statement that ‘the task of analysis and debate [has] very real importance and urgency’ (p. 112). The powerful trend towards automation is itself sufficient to make that task important, whenever—or even whether—the trend will come to complete fruition.

Correspondingly, it is important that one’s analysis be conducted within a coherent theoretical framework; that it should not be distorted by the attempt to use a set of concepts which can only confuse and mislead. Unfortunately, Morris-Suzuki’s discussion (which contains much fascinating material independent of any particular theoretical framework) is firmly cast within a completely inappropriate set of concepts. I refer to the following chain of ideas: profit is explained by the existence of surplus value; in a fully automated economy no labour is performed and hence no surplus value is generated; thus a fully automated economy cannot have positive profits. (Cf. the following from Morris-Suzuki’s argument: ‘It is only in the design of the new productive information and the initial bringing together of information and machinery that surplus value can be extracted. Unless this process is continually repeated, surplus value cannot be continuously created, and the total mass of profit must ultimately fall’ (p. 114).) It is precisely her (perfect explicit) adherence to this chain of ideas that drives Morris-Suzuki to suggest that full automation of manufacturing is possible only if labour continues to be employed—and to generate surplus value—in software production. But since software industries can and probably will be automated in turn, she argues, this gives only a provisional respite (p. 120).

Why is this conceptual framework completely inappropriate for a discussion of automation? Because a fully automated economy will exhibit zero surplus value and can be organized by private, individual owners of the means of production receiving a uniform, positive rate of profit. What is revealed by full automation, is not the ‘inner limit’ of capitalism but rather the ‘inner limit’ of the labour theory of value and of surplus-value theorizing. Consider Morris-Suzuki’s argument in the light of the following very simple example of an economy with three industries, producing software, hardware and a consumption commodity, respectively. In the interests of simplicity, both software and hardware are fully used up in one cycle of production; the consumption commodity is not used as an input to production; and any wages are paid at the end of the period. (The nature of our conclusions will not depend on these simplifications or on the particular numbers used in the example.)

The first row of Table A shows that one unit of software, one unit of hardware and L units of labour are used to produce 5 units of software. The second and third rows are to be interpreted similarly—but notice that hardware and consumption production are supposed to be fully automated. The final row shows that the net product of the entire system is just one unit of consumption commodity, since the outputs of software and hardware just equal the corresponding quantities used up in production. Let r be the rate of profit per period, ps and ph be the prices of one unit of software and of hardware, respectively, relative to the price of the consumption commodity, and w be the real wage rate, also measured in terms of the consumption commodity. Then, from Table A,

Equation (1), for example, states that the gross revenue in the software industry, 5ps, must cover the capital costs, ps + ph, plus profit at rate r on those capital costs, (ps + ph)r, plus the wages bill, wL. Equations (2) and (3) are exactly analogous, except that there is no wages bill to be covered.

Suppose first that L = 1; software production is not fully automated. Since a total of one unit of labour produces a net product of one unit of the consumption commodity, the labour value of that commodity is just unity. Thus if w = 0.5, half the net product goes to workers and half to capitalists. Consequently total surplus value = S = (0.5) x (1) = 0.5. It is readily calculated from (1), (2) and (3) that (approximately)

Now suppose, by contrast, that L = 0; there is complete automation and, inevitably, S = 0. From (1), (2) and (3) we find that (approximately)