Marx was correct. Workers alone create surplus value
Humans create machines. Machines do not create humans.
I will again defend one of the favourite punching bags of anti-Marxists: the labour theory of value (LTV).
Last year I wrote about Ian Wright’s superb, and very mathematical, defence of it. The other day I came across a rebuttal of the LTV that Australian economist Steve Keen made many years ago. Keen’s piece didn’t convince me, but it was very thoughtful, and thought provoking. Before responding to Keen’s critique, let’s quickly review the LTV.
In Vol I of Das Kapital, Marx made the distinction between the use-value of a commodity and its exchange value. For example, a typical monthly water bill where I live is about $90.00. That’s an exchange value. But we all know the use-value of water to us is virtually infinite. The rule of thumb is that humans can survive only about 3 days without water.
A capitalist workplace must, in the selling price of its commodity, recover the exchange value of all inputs (rent, electricity, raw materials, wages) including wear and tear on any machines used in the production process. (Note that the simplest tool or most advanced robot can be classified as a machine). But Marx argued that capitalists obtain a use-value from workers that is greater than the exchange value (wages) they pay for workers’ labour power. He called the difference between workers’ use-value and exchange value a “surplus value” that capitalists get only from workers.
Capitalists therefore aim to lower the proportion of the work day their employees use to pay off their real wages in order to maximize the surplus value they produce.
The essence of Steve Keen’s critique of the LTV is that capitalists also get use-values from all the other inputs to production (machines, energy, land) not just labour power. Keen therefore concludes that Marx was wrong to treat labour power as a special kind of input that was solely responsible for surplus value.
But all the other inputs to production (except the free ones provided by nature like sunlight that would exist even if all humans disappeared from the earth) were brought into existence by human work. And the income spent by capitalists must consist of goods and services produced by workers. To sum up, the non-labour inputs required human labour. The inputs that went into making the inputs required human labour etc. Ian Wright showed through this analysis that prices in a capitalist economy are proportional to all the direct and indirect labour time contained in the inputs to production as well as the consumption basket of the capitalist class.
In fact, forget employers for a moment (who are consumers of intermediate goods required to make products) and consider an example of a consumer of a finished product. The use-value I’ve obtained from the guitar that I play every day is much greater than the exchange value I paid for it years ago. That use-value ultimately owes its existence to the labour time required to manufacture the guitar; the labour time required to manufacture the components that make up the guitar (the guitar strings and other components the guitar manufacturer purchased rather than made itself), the labour time required to acquire the raw materials needed to make those components, the labour time time required for all the goods and services consumed by the capitalists who own the firm that made the guitar etc…
But Wright cautioned that prices in a capitalist economy could similarly be shown to be proportional to any basic input (wheat, iron, energy) not just labour time. So is this analysis above enough to consider Steve Keen’s argument against the LTV refuted? Not quite. We still need to answer why labor time should be singled out as the special basic input that creates surplus value? More should be said.
The Turing Test applied to surplus value
In fact, Wright did say more in another very useful piece - though it’s one with which I have a bit of disagreement.
Wright set up a thought experiment that appeared to refute the LTV, but then explained why it actually didn’t. The thought experiment was based on The Turing Test (originally called the Imitation Game) which was proposed by Alan Turing in 1950 as a way to objectively assess if a machine can think. The Turing Test says that when humans can’t tell whether they are communicating with humans or a machine then the machine should be considered intelligent.
I think that’s a bad test. If I can do a good enough Robert De Niro impersonation, am I therefore Robert De Niro? But I do like the way Wright applied the Turing Test to answer if humans alone produce surplus value.
In Wright’s thought experiment, one company uses robot taxi drivers which are hidden from the view of the passengers. Another ( equally profitable) company uses human taxi drivers who are also hidden from passengers. If passengers can’t tell if they are being driven by a robot or human driver then why deny that the robots create surplus value for the company?
Weight’s answer (which I think is very good) is that the thought experiment describes a static situation of constant surplus value. Human taxi drivers can, on their own, adapt to changing conditions to figure out ways to make more surplus value. Wright gives the example of human taxi drivers realizing they can make extra money by delivering meals from one place to another, not just people.
Weight stresses that Marx was focused on the relentless quest to increase (ie change) surplus value. Workers, unlike robots, have the ability to innovate and therefore change the level of surplus value, in every production process, by choosing to work harder or smarter. Weight concluded:
The taxi driver in a box thought experiment correctly assumed that the behavior of humans and machines can be identical. But it’s quite wrong to assume that the causal powers of humans and machines are identical.
The steady state is unsteady and requires innovation
But another problem with Wright’s thought experiment is the very notion of static or steady state process. In my experience as an engineer I’ve found that machine breakdowns, parts shortages and other disruptions are a regular feature of the production process. Even the weather can have a big impact. Eliminating disruptions completely is very difficult. And even brief disruptions are very costly.
When production runs smoothly - when anything like a steady state is achieved- it is often because workers, including those labelled “unskilled”, innovate to keep things running: A worker makes a tool at home ( or buys one) to avoid injury on the job- thereby ensuring he or she does not disrupt production. “Unskilled” workers notice debris that could cause a breakdown and alert their bosses, or simply clear the debris away themselves. Workers figure out how to load parts into a machine in a more efficient way than their bosses instructed them..
Automation creates demand for skilled workers who are trained to troubleshoot, repair and maintain machines. Such workers (engineers, mechanics, electricians) often make small but important design modifications to machines. But such workers often consult with the “unskilled” who are paid to merely feed parts into machines to keep them running.
Even managerial propaganda stresses the importance of small innovations that often go unheralded. It also hypes worker “engagement”, “empowerment”, and “involvement” . Of course, to an extent this is just management glossing over class conflict - pretending the workplace is a family. But it is true that continuous improvement requires using the intelligence of the entire workforce. Marx wrote in Das Kapital that in a factory machines make use of humans rather than the other way around - that machine turns a worker into a “mere living appendage”. Capitalists try to turn most workers into mere appendages of machines (they certainly try to make workers think of themselves that way) but capitalists also want workers to remember their humanity and intelligence in order to keep machines running.
The Responsibility Test
Going back to Wright’s taxi driver thought experiment, who should we hold responsible if a robot taxi driver kills or injures somebody?
Note that I deliberately asked who we “should” hold responsible. To serve capitalists, legal systems come up with abhorrent answers to that type of question. Capitalism would not exist if capitalists were made to take full responsibility for the things for which they demand credit.
But whoever you think the legal system should punish, nobody will say that robots should be punished. And nobody (except maybe for lawyers defending capitalists) would argue that nobody is responsible. If some group of humans must be held responsible for the disasters caused by malfunctioning machines, then the same should be true of any surplus value generated by machines.
Humans created machines. They didn't create us. Wright mentioned this argument but in my opinion was too dismissive of it. Machines don’t fail. People do. Whether it was due to an error in design, manufacture or testing ( or some kind of inappropriate use) every machine failure is really a human failure. Roderic Day made a similar point regarding the errors made by AI systems: “All the computer is doing is memorizing and extending patterns. It doesn’t notice its own errors; we do.”
The Rebellion Test
There is science fiction where machines rebel against their human creators (Stanley Kubrick’s “2001:A Space Odyssey”, Alex Garland’s “Ex Machina”). I don’t know when (if ever) humans will be able to do with machines what it took nature billions of years to do: create lifeforms intelligent enough to innovate in any production process. But one aspect of being able to create and innovate is being able to recoil at the absurdity of taking orders from the likes of Elon Musk. I suspect that before we get anywhere close to developing truly intelligent machines that have their own agenda, we’ll have to achieve many other things first: eliminate plane crashes, stop bridges from collapsing, end genocide and abolish capitalism.
