The Boy Who Cried Robot: A World Without Work

[crazy canuck]

Again - if low productivity workers are being systematically replaced by high performing technology, we shouldn't see declining productivity.

A machine doesn't have to be more productive than a person to replace them, just a reasonable facimile thereof and cheaper in the long run.

If a machine is not more productive than a person how would it be cheaper in the long run?

[jimmy olsen]

There's one significant expense, the purchase,  and then periodic maintenance. As long as that's less then paying someone a salary, it's cheaper. No need to be more productive,  it can be the same, or even a little worse if the daving is big enough.

[The Minsky Moment]

Productivity is output per labor unit.
So if you replace a worker with a machine that is just as capable, productivity must go up. 

[Iormlund]

It's not always about productivity, actually.

Automation allows the designer of a process to overcome many challenges. Machines don't care about alcohol blood levels, toxicity or dangerous radiation exposure. They are remarkably accurate, consistent and fast. I've seen guys with over 25 years experience trying to weld the same parts our cells manufacture for Audi. It takes them hours to do so manually, while the robots can do it in only a few minutes and always get the same results.

[Valmy]

Machines don't care about alcohol blood levels, toxicity or dangerous radiation exposure. They are remarkably accurate, consistent and fast. I've seen guys with over 25 years experience trying to weld the same parts our cells manufacture for Audi. It takes them hours to do so manually, while the robots can do it in only a few minutes and always get the same results.

How is that not about productivity?

[Admiral Yi]

It seems some of you guys are misunderstanding Joan's point about productivity.  He's not talking about a productivity competition between humans and robots.  He's saying if robots were replacing humans then we would be seeing increases in human productivity, which we're not.

[The Minsky Moment]

Right. It's just a mathematical identity

If productivity is defined as Output/labor hours or output/worker, then any replacement of a human worker through automation should increase productivity even if the machine isn't so great.

[frunk]

Right. It's just a mathematical identity

If productivity is defined as Output/labor hours or output/worker, then any replacement of a human worker through automation should increase productivity even if the machine isn't so great.

Assuming that any work done to design/build/maintain the automation is also considered output.

[Admiral Yi]

Assuming that any work done to design/build/maintain the automation is also considered output.

All output means all.

[frunk]

Assuming that any work done to design/build/maintain the automation is also considered output.

All output means all.

Which means it isn't a particularly useful metric to measure whether automation is increasing or not.

[Admiral Yi]

Which means it isn't a particularly useful metric to measure whether automation is increasing or not.

Yes it is.  Automation means the numerator (output) is increasing and the denominator (man hours) is decreasing.

[Iormlund]

It seems some of you guys are misunderstanding Joan's point about productivity.  He's not talking about a productivity competition between humans and robots.  He's saying if robots were replacing humans then we would be seeing increases in human productivity, which we're not.

Not necessarily, if you are transferring workers from already productive industries to others that are less so.

[frunk]

Which means it isn't a particularly useful metric to measure whether automation is increasing or not.

Yes it is.  Automation means the numerator (output) is increasing and the denominator (man hours) is decreasing.

Let's say a robot replaces a highly skilled worker, does it slightly less well than the human worker but it's cheaper so it works out.  The human worker, in order to survive, is pushed to work a fast food job at minimum wage at the same number of hours.  The human worker's productivity has dropped through the floor, the robot isn't quite as good as the human before, so total productivity is about the same as before.

Automation has increased but it won't be reflected in the productivity numbers.

[The Minsky Moment]

Let's say a robot replaces a highly skilled worker, does it slightly less well than the human worker but it's cheaper so it works out.  The human worker, in order to survive, is pushed to work a fast food job at minimum wage at the same number of hours.  The human worker's productivity has dropped through the floor, the robot isn't quite as good as the human before, so total productivity is about the same as before.

Automation has increased but it won't be reflected in the productivity numbers.

In your example, productivity would still increase uncles the decline in production from the robot is greater than the value of the Big Macs generated by the worker.

One can theoretically conceive of such a scenario and then extrapolate it economy wide but it doesn't seem like a very persuasive description of reality.

The phenomenon that the OP is talking about is not the replacement of workers by crappy, mediocre machines - it is the obsolescence of human labor by highly capable and sophisticated machines.  That is the phenomenon that isn't showing up in the data.

[Malthus]

When crying robots take over - what will happen to the productivity of the Japanese porn industry?