CGP Grey makes great explanatory videos. They are very thoughtful and noticeably PowerPoint generated but this time he branched out into a massive 15 minute documentary styled affair. The video is “no humans need apply” and it deals with topics near to my co-blogger Erik Brynjolfsson’s heart.
The story in the video is that if you don’t think the robots can replace you (a) look back at the horses and (b) look forward at what’s happening in robot technology now. But because it does its job so clearly, I thought it would be useful to consider a little more carefully the economics of all this and whether, indeed, there will be mass unemployment in the future.
While the video may make you feel sad for the horses, you have to remember that there are just fewer of them. This is because horses were not labour but, in fact, capital. So when they were replaced by oil and stream powered capital, it was a substitution of capital goods or, more to the point, fuel. So for the bourgeoisie who were invested in horses, that substitution was bad news for them. For the bourgeoisie invested in wheels, it was another matter.
But it is instructive to consider how that substitution arose. Basically, for horses to compete, their ‘rental cost’ to productivity ratio had to be sufficiently low that they could be employed in competition with other capital that had their own ‘rental cost’ to productivity ratio (usually worse than horses) plus a quality advantage. So as the lot of other capital improved (both in productivity and quality), horses ‘rental cost’ had to drop. There came a point, however, where the rental cost couldn’t drop far enough to make horses competitive and that was it for them.
So let’s now translate this for the robots versus humans equation. There are lots of differences but let’s start with the humans as horses point. For a human to be displaced by a robot in a job the robot (a) has to have a quality advantage and (b) the human productivity must be so low that even if the wage drops to minimum (by law or subsistence or opportunity cost), they will be uncompetitive. What that means is that the routine jobs listed in the CGP Grey video are vulnerable (as they have been to capital substitution in the past) but the other jobs (including journalism and actually being a lawyer) are less so. To think that they are means you have to have a robot technology that is actually higher quality in doing those jobs. The video points to certain tasks being vulnerable but, at the moment, it does not look like the higher cognitive bits (such as they are) are in trouble. But I’m a technology optimist, so who knows? I’m just saying that the quality advantage needs to appear.
But what of the humans that have no quality advantage with respect to robots for what they are currently skilled at? The Brynjolfsson-McAfee line is that they will need to acquire other skills and, more to the point, it is not clear the market can work (or at least work fast enough for our liking) to make that happen. So there is a danger of the humans going down the lot of horses there.
There are, however, two important differences. First, unlike the horses, the humans are also useful as consumers. They are the people who will value the products the robots (and other humans) produce. Think about that for a moment. For each person who is disengaged from society because of a robot, if you cut them off from consumption as well (by say not giving them any money), that is a unit of demand gone. So this pool of unemployed are left outside the system and do not interact in any way with the robot-employed economy.
If that sounds unsustainable, it is. There is a contradiction in the story. You have a person who values the product produced by robots by more than the ‘total cost’ in terms of resources to supply them with that product. You have to feel pretty ill about the prospects of capitalism to suppose that such an opportunity (certainly at scale) will go unexploited.
How will this happen? The most obvious way is that a collective agency will step in to ensure those people can pay for the product so that normal market based prices will be formed and transactions will take place. That agency is obvious in democratic society: the government. And before you think that this is some leftist notion (not that there’s anything wrong with that), I’m not theorising that here: I have faith that if it is in the interests of both business and consumers that money go from the employed to unemployed, it will. It will happen. So there may be unemployment but that does not mean that the humans adversely affected will become horse meat.
But there is another mechanism which goes back to the title of this post. The presumption is always that the bourgeoisie rather than the proletariat owns the machines. But why should that be the case? The robots we are talking about these days are not industrial scale. Why would it be the case that a pure capital owner will purchase the robot rather than a displaced worker any more than it was the case that those who drove horse drawn wagons where displaced by pure car owners?
The way to analyse this is simple: consider who is willing to pay more for a robot, a worker or a capitalist? Remember that the robot has higher quality than the worker (our earlier condition). If the capitalist owns the robot that quality advantage accrues in part to them. If the worker owns the robot, the quality advantage accrues to them. But in terms of their willingness to pay, there is an important difference between the two. If the capitalist does not own the robot, they can employ the worker (with their robot) as they always have done and so will they will appropriate part of the quality advantage. By contrast, if the worker does not own the robot, the worker gets nothing. So to the worker, their willingness to pay for the robot is based on the quality differential they expect to appropriate while for the capitalist, their willingness to pay for the robot is based on the difference between what they appropriate with the robot versus what they appropriate by employing a worker with a robot. It is pretty easy to see that the capitalist willingness to pay is, in fact, lower than the worker’s.
This is not a fanciful notion. Consider the washing machine. When that was invented it displaced workers — the person in the home who did the washing. But who ended up owning it. Usually that same worker rather than some business who then sold washing services into the home.
Consider also the journalists. In the CGP Grey video, they were vulnerable to automation. But I have news for you: they were vulnerable to this for decades. What is more, the robots came (in the form of Google providing roboting search for information) and who owned those robots: the journalists. They are the ones who use Google and, in effect, own it. Not their employers. Why do we think that journalists won’t use other automation tools in the same way becoming owners and controllers of them rather than being substituted away by them?
My point here is that we need to think about the ownership of the machines as endogenous and not assigned by presumption. It makes a big difference to how we think about the long-term effects of these changes and the allocation of welfare they imply.
I’m not claiming here that the adjustment will be pleasant. What I claim here is that the adjustment will occur and the long-run prospects for a interconnected rather than disconnected society and economy remain as good as ever.