Machines Replacing Humans: They Shoot Horses, Don’t They?

Eric Brynjolfsson and Andrew McAffee have a new Kindle instant book out, Race Against the Machine, that very nicely describes the issues related to technological unemployment. It’s well-written, content-packed, cogently argued, usefully hyperlinked, and well worth the $3.99 they’re asking.

But I think there’s one crucial topic they don’t address, highlighted by the following.

They deliver a quotation from Gregory Clark’s 2007 Farewell to Alms, speaking of the decline in employment of horses in England in the 19th and 20th centuries. The quotation concludes:

There was always a wage at which all of those horses could have remained employed. But that wage was so low that it did not pay for their feed.

The question that B&M fail to ask or answer: Why couldn’t those horses continue working for a living wage?

Answer: because they couldn’t upgrade their skills (to drive trains and tractors, or whatever). They were biologically incapable of “improving” themselves in such a way that they and the new machines were “complements.” So the machines replaced them instead of complementing them.

Imagine that those horses had been free to mate and have offspring at will (as opposed to the PRC-style, top-down population control imposed by breeders). They couldn’t work (nobody would hire them because their marginal productivity relative to machines was so low), so they couldn’t make any economic claim to a share of the (massively increasing) pie of production and prosperity.

They would have faced an unequivocally Malthusian situation: large numbers of horses would have starved.

Now suppose that the population-control option was … not an option. And that laissez faire — letting them just live (and die) with their subsistence or sub-subsistence incomes — was also not an option. The only solution would be to subsidize their lives, transferring money from those who own and profit from the machines to those who can’t benefit economically from using those machines.

You know where I’m going. (I’ve taken you there before.) Human capacity has limits. By definition, 50% of people have an IQ below 100. The prescription we so often hear — that those folks should train to operate computer-driven lathes, and to perform in the sophisticated, information-driven organizational structures associated with that kind of equipment — is … less than realistic.

Rapidly accelerating machine capabilities can result an economy in which those people’s marginal productivity drops to near zero, or even negative — so no matter how hard they work or train, their contributions aren’t great enough for them to “deserve” a decent share of the pie. (Define “decent” as you will; for me in our wildly prosperous society it extends to a good chunk of leisure/recreation time with friends and family, taking family vacations now and then, such like that. Let’s hear it for “family values.”)

If you’re reading this, then like me you probably can’t even imagine what it would be like to have really low intelligence (or other below-average capacities) — how desperately hard it would be to make a go of things, to raise a family and give them a good life, in the country we live in today. Imagine just trying to get through high school.

Some people may get unsavory feelings from this thinking, so (to quote BHO), let me be perfectly clear: I’m not saying that less-capable people are horses. Quite the contrary. They’re people. So yes, I’ll say it: at least if they’re willing to work (or are unable to do so), they deserve a decent share of the pie. They shouldn’t be driven to the bottom while the machine owners wallow in luxury, just because the technological economy they happen to have been born into doesn’t value them any more.

In wonk-speak: income and purchasing power must be decoupled from human participation in production. Not completely, of course — requiring participation in the work force is necessary to solve the free-rider problem. But in an economy that’s bursting with surplus, we can be better off overall if individual incomes (at least at the low end) are not rigidly associated with individual capabilities.

Which brings me back, yet again (getting practical here), to a greatly expanded Earned Income Tax Credit — encouraging both work and hiring — with benefit levels indexed to some measure of unemployment. This would require, of course, that we implement a tax system that actually is progressive, to pay for the increased EITC.

Imagine a post-scarcity future (or … present?) in which high productivity (think: Star-Trek-style replicators) means there’s plenty for all, with little work required. My utopian image is of something like an Athenian Agora — people of all types gathering in public spaces, meeting their friends, and exchanging their wares and their news — instead of constantly scrambling to make a buck and collapsing in their homes when they’re done.

In fact, it looks a whole lot like the zillions of public squares you find throughout Europe — surrounded by residences, and cafés and such where people mingle and spend time with their friends — the kind of environment that pretty much doesn’t exist in America.

And I haven’t even touched here on the idea that this redistribution may be economically efficient, even necessary, to maintain aggregate demand and support productive enterprises, keeping the log that is our economy, rolling.








4 responses to “Machines Replacing Humans: They Shoot Horses, Don’t They?”

  1. Olav Martin Kvern Avatar
    Olav Martin Kvern

    re: “By definition, 50% of people have an IQ below 100.”

    …but you’re assuming that people can’t be made smarter. I’m not talking about acquiring skills and knowledge (and I agree that that’s a mostly useless approach*, relative to increases in machine intelligence**), I’m talking about increasing intelligence. How do we make people smarter? I’m thinking that an engineered IQ-enhancing virus–liberally applied to the channel buttons of TV remote controls–would do the trick.

    No, but, seriously, I think we’ve got to find a way to increase human intelligence. I am horrified, every day, by our stupidity, especially my own.

    Clearly, it’s going to take us awhile to figure this out. Two things stand in the way: the assumption that intelligence is a fixed quality of a person, and that it’s hard to experiment on people. (The latter has been explored in fiction–“Flowers for Algernon,” of course, or the vastly better and more messed up “Camp Concentration” by Thomas M. Disch, to name a couple.)

    In addition to finding a way to improve basic human intelligence–whether its done by biological or nanotechnological means–we’ve got to get rid of all of the stuff that keeps people from thinking effectively: religion, tribal mentalities (racism, sexism, etc.), psychological problems, and so on.

    I really like your Utopian vision, and, essentially, share it. But I want that agora filled with intelligent, interesting people, going wherever their talents (enhanced or otherwise) take them.

    On people not being horses:

    The problem horses had was that even the very smartest horses were not smart enough to come up with a plan that would provide for most or all of their species to carry on after they lost their jobs in the human world.

    Can humans do better than that? I’m hoping so–especially since, when horsepower became obsolete, the smartest horses, the working breeds, were the ones that died out/were killed off the quickest. Humans kept some of the stupid ones as pets.

    * Note that I’m not saying that acquiring knowledge and skill aren’t important–I think a fair amount of practical intelligence comes from having a pool of knowledge in which to make connections. But learning to do things that no human should have to do (i.e., things that haven’t been automated yet) is a dead end.

    **Note, as well, that advances in artificial intelligence might have benefits for increasing human intelligence. Whenever people talk about integrating circuitry with the brain, they tend to focus on increasing memory, adding new facts, etc. Why not statistical co-processors or enhanced visual processing? Or, as we used to say, a “personality module.”

  2. Asymptosis Avatar

    @Olav Martin Kvern

    Yeah all sounds great. I wouldn’t be so dismissive of memory modules, though. That’s floating up the wish list for me, lately…

    I really like that I came across today…

    “not that we don’t have enough stuff. don’t have enough ways for people to work and prove that they deserve this stuff”

  3. SS Avatar

    Meet the FarmBot:

    You should not starve the horse you are riding…

  4. James Oswald Avatar

    I responded to your comment in a post on my blog.
    “In wonk-speak: income and purchasing power must be decoupled from human participation in production. ”
    I agree. Without human involvement in the production process, most people will have nothing to sell to the market to get what they need to survive. Economic distribution will become more and more politicized. There will be no “Atlas Shrugged” moment because no one will be needed to make things.

    “…to maintain aggregate demand and support productive enterprises, keeping the log that is our economy, rolling”

    Without labor income, there is no reason to stabilize aggregate demand. The point of stabilizing aggregate demand is to stabilize income and employment. When there is no employment, demand can be left to fluctuate based on actual human needs rather than trying to artificially prop it up.

    When wages are 0, an EITC doesn’t get the job done. You’d need a guaranteed minimum income or something like that.