I’ve seen debates between two schools on where things are heading. The first, exemplified by Thomas Piketty and his Capital, is that capital is increasingly locked into the hands of a few, leaving the many less well-off as they roll back the gains labor has achieved in the Industrial Revolution. The second, espoused by the Modern Monetary Theorists, suggests the exact opposite: that labor is going to become more powerful. Obviously, both can’t be right … or can they?
There’s a wild card here, and that is automation.
Consider this: automation will gobble up almost half of all available jobs. Obviously, a human who is unemployable has a labor worth that, while it may not exactly be zero, approaches it asymptotically, and is below the cost of employment. Why would Starbucks keep employing liberal-arts-degree’d baristas when BaristaBot2000™ can do it for pennies on the dollar? So, for this class, Piketty is ultimately right: they’re being locked out of the means to conduct their affairs.
Meanwhile, the other half, the ones who build and maintain the machines, who teach them how to think, who have fine skill and motor jobs that the bots cannot hope to replicate, who can fill custom niches … their labor value goes up. By a lot. Not only are they employable, they’re highly employable, and they’ll be able to command salaries that will shock and numb the unemployable class. For them, access to capital has never been higher, and upward mobility is, like, a real thing.
We can already see this today. Skilled coders in Silicon Valley are worth $150,000 minimum … I’d be lucky to land a salary that makes even a third that, before my landlord finally kicks me out. It isn’t just a rift between rote labor work and skill work: it’s what kind of skills you have, or more precisely, got while you were in college. Can you do higher math and code? Congrats, your labor is worth 2x that of someone who’s a gifted wordsmith. (After all, the bots are coming for low-skill journalist, PR, and media relations jobs anyway.) It’s a huge part of why the gap between the haves and have-nots is growing.
Extrapolate that, and we can see a three-classed society: the control class (the 1%), the valued-skill class (~50%) who have access to capital, and the devalued-skill and no-skill classes (~50%), who have no access to capital, the ones who are fundamentally unemployable.
But wait! you say. Aren’t you forgetting something? What about new work? After all, we don’t have very many horse shoers or sail makers anymore. Their jobs didn’t just go away. They were replaced.
And that brings us to the second problem, the reason why automation creates class stratification.
See, there are two major forces acting on labor worth. The first, innovation, creates new work and therefore increases net labor worth, as more work to do increases a worker’s value. The second, efficiency, slowly acts to reduce necessary inputs on known outputs, and in so doing, decrease a worker’s net value. Think about it this way: a company that used to have three workers doing a job they now have one worker doing is getting the work done more efficiently, and therefore is getting more bang for their labor buck … but the two workers that aren’t needed anymore are shit out of luck.
Let me here introduce the concept of replacement. The idea here is that the change in labor value (let’s call this Δv) is differentially dependent on two variables — that of labor efficiency (let’s call this ε), which squeezes work out of the system, and that of innovation (let’s call this I), which creates new work in the system. However, if the rate of change in labor efficiency (i.e. Δε) were to perfectly match that of innovation (ΔI), v<0, as natural increase adds fresh bodies to the system needing work. So that is also a value, P.
For replacement to be perfect, that is, for the amount of new work in the system to perfectly match the labor needs caused by old work leaving it as well as natural replacement’s increase in warm bodies, we can set up the simple system: where Δv=0, Δε=PΔI. But of course the system rarely works like that, and instead we’ll find inequalities: if Δε>PΔI, Δv<0; and if Δε<PΔI, Δv>0. What this means is, of course, if we want the value of labor to go up — that is, if we want the work people do to be able to afford “comfortable” lifestyles, the rate of change in innovation times the factor of natural increase must always exceed that of efficiency.
We kind of understand this relationship … it’s why we understand at a gut level that universities are important (they’re innovation factories!), but efficiency is easier to see and measure than innovation, and so historically economics has attached undue importance to it. There still isn’t really a good direct measure of innovation: Jane Jacobs suggests “breakaways” would have worked as an indirect measurement in Victorian Birmingham, in The Economy of Cities; that translates to small-business creation in the United States today. Our understanding about how to model innovation remains all but nonexistent (whereas efficiency is well-understood), and yet, when it comes to being able to afford stuff over the long term, grokking innovation is even more important than grokking efficiency.
This fundamental ignorance of the connecting variable, innovation, leads to Piketty and the MMTs both being right. Automation is a creation of efficiency — it’ll lead to Δε going up, and in a world where we already see that Δε>PΔI, which naturally means that Δv crashes. Over the broadest possible scale, Piketty is right. Meanwhile, however, in the sectors that automation does not affect, ΔI<0 — because much of the work needed to control the bots does not exist yet. Here, we see labor value rising quickly, as there is plenty of new work to be had. Here the MMTs have it right … but very, very fragilely so. And that sectoral ΔI is simply not enough to satisfy the work demand across the entire system … just among the small subset of those whose skills are automation-proof.
From there, it’s a hop, skip, and jump away from a rigidly unequal class structure. That has to happen, because without mass disenfranchisement of the have-nots, this system collapses in on itself. There are just too many people locked out, demanding work, otherwise.