Humans need not apply
CGP Grey is a dude who publishes awesome videos on YouTube. At the bottom of this post I've embedded the video that prompted this screed: about the coming of automation, and the problems it's going to present.
Short form: people are going to be displaced from their jobs by machines. Count on it. It's going to happen, and we need to think about how to deal with it before it happens.
There are two theories about the coming increased automation. One is the "this has been predicted before, it's never happened, and it's never gonna happen" theory. The other is the "this time it's different" theory. I think that this time it's going to be different.
Proponents of the "this has been predicted before" theory point to the Luddites, the poster-children for the wrongness of the belief that this will happen. The Luddite movement arose in the 1800's when the automation of textile production--for example the invention of lace-making machines--began to put artisans--for example lace-makers--out of work. The Luddites were right--lace making machines did obsolete artisanal lace-making--but they did not account for the overall benefit of less expensive end-products--like lace--and the creation of new, and better jobs for those displaced.
The Luddites were so wrong that there's a fallacy named after them and this description says what I just said, only more economistically:
Economists apply the term "Luddite fallacy" to the notion that technological unemployment leads to structural unemployment (and is consequently macroeconomically injurious). If a technological innovation results in a reduction of necessary labour inputs in a given sector, then the industry-wide cost of production falls, which lowers the competitive price and increases the equilibrium supply point which, theoretically, will require an increase in aggregate labour inputs.
If you've got a fallacy named after you then you must be wrong. Right?
Maybe not. I go with the "this time it's different," crowd, because--well, because this time it's different.
Today's human beings, the ones who can read, write, drive cars and send texts, and even the vastly smarter ones who can do nuclear physics are not all that different from the humans whose peak economic activity was gathering nuts and berries. Those folks developed brains. Nut and berry gathering is not that challenging, but living in a human society--even a primitive human society--is demanding. So those brains had to be good ones--or no babies.
Over time the economic environment changed to match the social environment and people had to develop and exercise a wider range of skills. The basic talents that people apply to accounting, woodworking, and computer programming were always there. The opportunity to use the talents was not.
Historically human brains were under-utilized. Working in a factory takes physical skills, but it also takes mental ability, which is why humans work in factories and why monkeys, who might work for less money, do not. Over time we've created jobs that use more of our natural human talent. But we're limited. We can create new jobs because of cultural evolution, but we are constrained because the hardware on which the software of culture operates changes very, very, very slowly. And we are limited.
Not every human who can forage can learn to hunt or to farm--though most can. Not every human who can learn to use a hoe can learn to run a tractor--though most can. The fall-off rate may not be large but it is steady, and for every set of new skills that must be acquired to do a job there are some who cannot. Why? Because their hardware is not sufficient to let them do it. Thus, not every "manual laborer" can become an effective "knowledge worker." And not every knowledge worker can become a computer programmer. It's not a matter of education, though education can help. But all evidence says that you can send someone to school for years to learn a skill for which they don't have the basic smarts and they'll never be very good. Not for want of trying, but for want of wiring.
The relentless march of technological improvement is a march toward better and better jobs that fewer and fewer people can do well. Working in a salt mine is a job that almost any healthy person can do well enough to not be beaten to death by the salt-mine-overseer for non-production. Working in an office is a job that some healthy people cannot do well enough to get fired. The jobs are better. The consequences of non-performance are less dramatic. But there are fewer people who can do the newer jobs.
In the limiting case, which we are not close to approaching, but which we will approach soon enough, there are relatively few people who can do jobs that computers cannot do better.
Today there are still enough jobs that most people can do so that most people can find work. But to me the trend is clear: the jobs that will remain will not be jobs for everyone--even with the best possible education, even with the best computer assistance to help them.
And the trend is accelerating, and that's the second problem.
Back in the old days, when change was slow people lost their jobs slowly and new opportunities appeared slowly. But the rates were well matched. The old generation might end up out of work, but there were new, and better jobs for the new generation--except for the small, incremental number who could not do the new jobs. Never mind! There are not that many of them. We can afford to support them with some sort of "social safety net," or with jobs that are not strictly necessary but not clearly worthless, or by spreading the work among more people. In 1860, according to this report, the average work week in manufacturing in the United states was nearly 70 hours. And working in manufacturing was probably one of the better jobs you could get. Now it's something north of 40 hours a week.
Of course we do create new opportunities faster than before, but new opportunities are not the same as new jobs. In 2014 Facebook, a company with $2.5B in revenue had fewer than 10,000 employees. Yes, I know a billion dollars isn't what it used to be, and there are lots of people who do work that Facebook pays for who are not employees, but was there a company in 1914 with comparable revenues with that few employees? I haven't done the research, but I don't believe so.
I believe that as change accelerates the number being displaced will grow faster than our willingness to spread the work around or to share the benefit. It's a matter of willingness, not ability.
Some work will always have to be done by humans. If most the world's population has to work forty hours a week to provide what the world's population needs in order to have a satisfactory life, and if productivity increases so that that only one fourth as much human effort is required, then what happens?
We can do some job-sharing, but we can't let everyone work 10 hours a week because some of the work will be of a kind that most people cannot do, no matter how many hours they work.
Do some people work 40 hours and others let benefit from their productivity without working? Or do they let them starve? Starvation is not an option. If a substantial number of people find themselves at the point of starvation they will exercise their right to do whatever the fuck they need to do to survive, which includes taking up arms and overturning the established order.
The worst realistic case is that a balance will be struck between dissatisfaction, "welfare" programs, and suppression.