Saturday, 7 December 2013

One Day The Luddites Will Be Right

Whenever a person proposes that technological advances might reduce human job opportunities in the long term, someone responds with the Luddite Argument: "the Luddites thought the Industrial Revolution would destroy their jobs, and they were wrong, so you're wrong too" [1]. Some go further and explain that the Luddites were wrong because technological productivity improvements are balanced by finding new uses for human labour. Wikipedia has a good summary. However, it seems obvious to me that at some point technological advance will --- or at least could --- be a net destroyer of jobs. All you have to do is imagine a world where robots can do everything a human can do, at lower cost than maintaining a human life. Clearly, there are no economically rational job opportunities for humans in that state [2], so at some point of technological advance short of that state, there's net job destruction.

The only question is whether and when we will reach that point. It seems inevitable we'll reach it unless something halts technological progress or some very strong flavor of Cartesian dualism holds. Economic arguments that human labour will still be worth something at that point are just wrong.

[1] Actually the Luddites were right; the Industrial Revolution did destroy their jobs, and drove them into misery. But they were wrong in that they did not forsee the net benefits to future generations.

[2] There could be sinecures to keep humans occupied, but they would not be economically motivated.


  1. Brave New World? Soma?

    Or Star Trek?

  2. I agree with you. The slightly more sophisticated argument is using comparative advantage ( Comparative Advantage ). An early example of this is in "The Shape of Automation", 1960 by H.O. Simon where he claims that even when "men are less productive than machines in all processes" "By the operation of the marketplace, manpower will flow to those processes in which its productivity is comparatively high relative to the productivity of machines".

    Comparative advantage does mean that humans and 'technology' will still have mutual benefits to trading even after 'technology' can do any job more efficiently than a human can. However, at some point, the transaction costs (costs that exist because two groups are trading such as negotiation) exceed the cost of the 'technology' trading with humans. At this point either the 'technology' gives humans things freely, or humans are on their own.

    (I put a longer post on this up at The age of smart machines )

    1. I don't see how comparative advantage applies here even if we ignore transaction costs. The formulations of comparative advantage I'm familiar with assume that the production capacities of the producers are independent, and that isn't true here.

      Basically, whenever a machine can do the same job as a human, the human production side has all the same constraints as the machine side (access to the same pool of capital, raw materials, etc), plus the very expensive human component. I don't see how it makes economic sense to introduce the human component.

    2. This is a partial reply. Basically, if there is some difference between the ratios that humans and machines can produce, so say humans can produce 3 apples for every 1 orange, and machines can produce 3 oranges for every 1 apple, then humans would be happy to trade 1 apple for 1 orange, and the machines would be happy to trade 1 orange for 1 apple (assuming both wanted the other fruit).

      If the intelligent being input is the most expensive part (as you say 'the very expensive human component') than there probably are gains to trade since computers and humans will have different aspects of intelligence that are most efficient.

      It is when the intelligent being input is a small fraction of the cost and the cost is dominated by the raw materials and energy cost, that if the machine is say 5% more efficient in use of energy and raw materials that trade stops making sense. This is because the intelligent part is now the cheapest part.

  3. When robots can do every job cheaper than humans, do humans need jobs anymore? Seems like that would be an end to scarcity and a net benefit to future generations.

    1. There will still be scarcity due to limited resources, if not labour. But you're right that there could be a net benefit. However:

      1) We'd have to radically alter our economies --- the way we manage resources. Right now most people obtain resources by trading their labour. If that goes away, we need something completely different, e.g. a guaranteed basic income scheme. It's unclear what should be done, what will actually happen, and whether it will work.

      2) Work has value beyond just acquiring resources. It gives people something to do and a sense of worth. Destroying that has severe negative consequences, as we can see where social welfare schemes have gone wrong. This has to be solved.

    2. A guaranteed basic income and working for the work itself (a.k.a. "hobbies") sounds just about perfect to me. I am an optimist, though.

    3. I quite like the idea of a guaranteed basic income. However:

      "Right wing" people won't like the idea because it's statist and redistributive.

      "Left wing" people won't like the idea because you'll have two classes of people, those with only the basic income and a much smaller class who own all the robots and have all the power, i.e. much worse inequality than we have now. Even if you don't subscribe to the politics of envy, that's a recipe for instability.

      I think both of those objections have some merit.

    4. Also, hobbies as a substitute for work works well for some people but clearly doesn't work well for other people, at least, not the way we've managed things to date.