The difficulty with supposing that automation is producing unemployment is that automation isn't new, so how can you use it to explain this new phenomenon of increasing long-term unemployment?

Clearly computers are exactly the same, and ought to be expected to have the same effects, as steam engines. Just look at horses, they're doing fine.

Now there's been a recession and the jobs aren't coming back (in the US and EU), even though NGDP has risen back to its previous level (at least in the US). If the problem is automation, and we didn't experience any sudden leap in automation in 2008, then why can't people get back at least the jobs they used to have, as they did in previous recessions? Something has gone wrong with the engine of reemployment...But this must mean something new and awful is happening to the processes of employment - it's not because the kind of automation that's happening today is different from automation in the 1990s, 1980s, 1920s, or 1870s; there were skilled jobs lost then, too. ...even I can see all sorts of changed circumstances which are much more plausible sources of novel employment dysfunction than the relatively steady progress of automation.

And there are also issues like labor hoarding and sticky wages/ratchets and tipping points and technologies reaching break-evens. Let me describe another plausible argument: "since computers and software have increased their usefulness smoothly albeit exponentially, we would see productivity gradually increase over time due to computers/software, and computers/software as so great that this would be obvious to the dimmest person using the most gross aggregate figures". This argument would be dead wrong, you would see essentially zero benefit from computers up to the '90s, and this massively counterintuitive and unexpected fact is dubbed the productivity paradox.

You don't even show that we didn't see this sort of abrupt jump in disemployment back then! For all you know, during the various panics and busts, there were huge disemployment effects as companies were forced or enabled to automate, but the people were able to switch sectors or find new jobs, which is the principle claim here.

Or to be less extreme, there are lots of businesses who'd take nearly-free employees at various occupations, if those employees could be hired literally at minimum wage and legal liability wasn't an issue.

Part of ZMP, as you should be aware, is that it's perfectly possible to have lots of humans who you would not hire at any wages at all, completely aside from the issue that the much-ballyhooed disemployment effects of minimum wage have been surprisingly hard to observe. For example, how many people would hire a black kid from the inner city to do their dishes for $0 an hour? Not many. How many would do so if they learned that like distressingly many such people, the kid in question has been convicted of some crime or other? I am guessing less than 100% of people would hire them. This is an obvious case where you would not hire someone at any price; ZMP simply extends this to say that there are many more such people.

We do not literally have nothing better for unemployed workers to do. Our civilization is not that advanced.

Sure we are. One video of an employee spitting in customer's food can go viral and do more damage to a chain's sales than that employee would earn for the chain in a hundred years. One person in an o-ring process can do an incredible amount of damage if they are only slightly subpar; to continue the NASA analogy, one loose bolt can cost $135 million, one young inexperienced technician can cost $200 million. Isolated examples? Well, just calculate the expected-value of reducing the number of such incidents by even 0.01%...

Many industries that would otherwise be accessible to relatively less skilled labor, have much higher barriers to entry now than in 1950. Taxi medallions, governments saving us from the terror of unlicensed haircuts, fees and regulatory burdens associated with new businesses - all things that could've plausibly changed between now and the previous four centuries.

What happened to your smoothness argument? It applies just as well to your libertarian examples here - better, actually, because many of your examples have origins in the Great Depression, for example, NYC taxi medallions in 1937.

Human beings, including employers, are very averse to downside risk, so this could plausibly be a major obstacle to reemployment.

What's sauce for the goose is sauce for the gander. Why doesn't this apply to firing people as well and fully explain how automation could be smoothly progressing while disemployment cyclical?

We need some new factor to explain why this wasn't true in 1950, and obvious candidates

No, the obvious candidate is the increasing skilledness and fragility of production as automation and precision and all-around technological sophistication increases. You want to know what manufacturing looks like in 2013, and not 1950? http://www.theatlantic.com/magazine/archive/2012/01/making-it-in-america/308844/?single_page=true is as good a place to start as any.

A. Then it's odd to see so many news articles talking about AI killing jobs, when plain old non-AI computer programming and the Internet have affected many more jobs than that. The buyer ordering books over the Internet, the spreadsheet replacing the accountant - these processes are not strongly relying on the sort of algorithms that we would usually call 'AI' or 'machine learning' or 'robotics'.

Those were AI. "AI is whatever we don't know how to do yet", remember? Look at the MIT AI Lab, and what it and other AI places were doing in the '70s and '80s due to and to support their work: intranets, Internet, hypertext, interpreted languages with garbage collection, GUIs, single-person workstations, parallel processing, online chat and email, networking algorithms and on and son.

And then there's all the robotic warehouses which help online retailers like Amazon compete. Hm. I bet in a past era those warehouses would've been run using humans.

Even then, the total number of people driving cars for money would just be a small part of the total global economy; most humans are not paid to drive cars most of the time.

The trucking industry alone employs ~3% of the entire American population. That's not trivial by any means. And how many of those employees do you think are skilled operations research PhDs who can easily find employment elsewhere in logistics?

If we imagine that in future decades machine intelligence is slowly going past the equivalent of IQ 70, 80, 90, eating up more and more jobs along the way...

Q. Could we already be in this substitution regime -

A. No, no, a dozen times no, for the dozen reasons already mentioned. That sentence in Hanson's paper has nothing to do with what is going on right now.

Oh yeah? Alright, here's a kid with IQ 70. He can lift things under 40 pounds and put them where you tell him to. I'm afraid he can't read past a third-grade level, or anything like that. It's probably not a good idea to let him near any moving machinery either. Fortunately for you, he doesn't throw any violent temper tantrums and he doesn't steal - he's a sweet kid, willing to work. Just dumber than a stack of bricks. Take him down Main Street and see if anyone will hire him. How many job offers did he get?

More generally, Eliezer, you seem to completely fail to grapple with the real proponents of these ideas like Autor or Brynjolfsson or heck, even Cowen. What is the point of this 'anti-FAQ' if you aren't dealing with the actual arguments (never mind steelmen)?