Wednesday, April 18, 2012

Being a Little Less Gloomy

I have a little bit more to say on yesterday's gloomy robotics post.  The point I wanted to add is that this will inevitably become a huge political issue.  At the moment, the consensus that innovation is always good is not seriously questioned on either side of the political spectrum, except by a relatively small fringe (though Kevin Drum - definitely not a fringe character - did link to yesterday's post in broad agreement).  But as the processes of automation and technological unemployment continue, this will inevitably change.  Obviously, we will not create 100% unemployment amongst humans without provoking a massive political backlash along the way.

The nature of the kinds of exponentials involved is such that it will appear as though change is relatively slow until suddenly robots are popping up everywhere and job loss is threatening people who matter (or at least their kids).  So at some point - perhaps in the 2020s or 2030s - this will become a red hot political issue that everyone is talking about.  Then it will be possible to campaign for reform (reform being the preferable alternative to revolution - but one that doesn't get much traction until there is a credible fear of revolution).

So the task now is mainly one of thinking - of figuring out when the moment comes, what should we be trying to do?  What kinds of cultural changes will be needed, and what kinds of changes to legal institutions can produce a reasonably enduring human society in which people aren't like something out of WALL-E?  Clearly the main changes will have to come in the way we think about and regulate the innovation process.


Lars-Eric Bjerke said...


Already in 1969 there was a book published “The harried leisure class” by Staffan Burestam Linder, professor and foreign trade minister, describing how people will behave when production becomes more efficient but the productivity of services such as health care and repairs do not increase at the same rate. His predictions were just right- we will not reduce working hours and we will buy good which do not have the time to use or repair and most services will feel too expensive relative to goods. The book was republished lately.

Jared said...

Like Kevin Drum, I broadly agree with you. But here's something I think might help make things seem even less gloomy: a lot of the political rage against "looters and moochers" is only really popular because of the cost it imposes on the John Galts of the world. Absent personal costs, even callous unaccountable elites are willing to fork it over.

For example, the Saudi rulers responded to the Arab Spring by increasing benefits to their citizens. They have plenty of cash on hand, and the cost was negligible compared to a potential uprising. Gaddhafi also, I think, kept his people in line partially by profligate spending of his oil money to keep benefits (relatively) high. No one is going to accuse Gaddhafi of having been compassionate, but when you have mounds of cash lying around and dental insurance is cheap, it's an easy rational decision to make.

A future in which construction workers are being rendered unemployed en masse by robots is a future in which housing is very cheap. A future in which primary care doctors are being replaced by Watsons is a future in which basic health care is cheap(er). If providing food, shelter, health care, and the internet to the mass of unemployed really isn't that expensive, even Paul Ryan will be first in line to sign that bill.

This doesn't solve the problems of millions of idle workers and extreme wealth inequality. But at least the politics of avoiding societal collapse are pretty straightforward.

sunbeam said...

To me some obvious things to do are:

1) Start cutting the work week. Do it in stages. 36 hours now, 32 hours in 5 years. Do that for a while and see how it works. More technological unemployment? Cut the work week even more.

It's going to be necessary to do something like totally eliminate the salary system, for obvious reasons.

2) National Health Insurance. No if's, and's, or but's. No apologies, no mandatory insurance purchases. Totally free insurance based solely on citizenship (or maybe just being a human being). This insurance is a natural right, and has absolutely nothing to do with employment.

Like some other countries around the world.

3) Totally remove any control by employers over retirement funds. Want to have a retirement system to make employment at your firm more attractive? All you can do is give a donation to something like an IRA or 401k that is totally under the control of the recipient. Or maybe a pre-defined investment fund that is run by a trustworthy third party, solely for the benefit of the recipient, and that has some sort of tax breaks.

There is no way to dump someone a month before they qualify for a pension, you can't do anything with the retirement funds. They are solely and totally for the benefit of the recipient.

Retirement is totally and utterly portable. You get it paid from day one whether you work only a week (well it'd be miniscule), or 50 years. Your retirement is only for you, and you may enhance it by working for employers that pay more into it, but you get something from any job you have that has a retirement system.

What is so revolutionary about setting up something like a 401k for everyone in this country? I'm talking about a system like FERS, where the employer matches contributions, and they go into a fund where you can select several types of investments.

Only thing is I want even less control by the employer, whether the federal government or not, over retirement funds.

Basically I want to remove some of the control that employers have over people. It's not just the regular paycheck. It's usually more along the lines of health benefits, and later the pension (for a dwindling number of people admittedly).

I think what I am trying to say is I want to cut the workweek, so more people will be needed to do whatever actually needs to be done. And remove some of the controls employers have over employees, so it will be more workable to not have full time employment as has usually been conceived, or indeed two part time jobs or something.

Or maybe you go from job to job as a contractor. This would make it more in-line with traditional jobs as far as retirement and benefits (when I was a contractor I got neither).

Swany said...

If the robots are producing all the food and other goods we need, and they are distributed equitably, wouldn't 100% unemployment be just fine? We'd all be living like aristocrats or royalty. Now lets see if I can prove that I'm not a robot...

Allen Knutson said...

I expect the massive issue will come the day that turnkey robot drivers are cheaper than human drivers. There are something like 4M professional drivers in the US; mostly truck, some taxi and bus.

Once the statistics are weighty enough to show that robot truck drivers are (at that time) safer as well as cheaper, those jobs are gone.

Burk said...

Hi, Stuart-

I didn't take your post as gloomy at all. The more robots there are, the more interesting things the rest of us get to do. Most of agriculture has been robotized/mechanized. Who is complaining?

There is plenty of work being papparazi, playing guitars, and other more human activities. The question is whether we have the political will to distribute the ever-rising economic resources fairly, or whether we go down the bananna republic route, as we have been doing the last few decades.

In the new economy, food is cheap, clothes are cheap, machines are cheap, but humans remain expensive, like professors, manicurists, and psychoanalysts. The question is whether we let the concentrated power of the robot-handlers (i.e. corporations) turn into a perpetual feudal politico-economic system, or return to the kinds of post-war Keynesiansim and progressive taxation/moderate redistribution that constitutes a civilized jobs policy. Laissez-faire is not a jobs policy.

Brian said...

Or, one might argue, the main change would need to come in the way we think about people.

Currently, in many spheres of discussion, people are considered some kind of "cost", something to be removed, minimized or worked around. Perhaps if we change the things we value, we can change the ways we innovate.

Brett said...

If we're still wedded to the idea of employing people, I think 2030s US should lower work-weeks, try to lower some of the fixed costs of hiring people, and try to loosen occupational licensing for most fields. That would help to spread the work around, with most people working really low numbers of hours.

I think that would be more politically acceptable in the US than simply letting a large fraction of the population end up on welfare, although in practice that's what will happen anyways because of the aging population. Trying to get a small, wealthy, heavily mobile elite to pay for a larger population of non-workers is almost certainly going to be an unstable situation here.

Of course, I could just be like some late 19th century person predicting mass unemployment due to industrialization. Perhaps we'll just see an explosion in the diversity of services that humans can provide to each, many of which we can't even imagine now. With the increasing use of computer-generated secondary worlds, maybe using real human services will become a luxury product and sign of status.

Glenn said...

Unless those robots are all put to work building wind turbines, photovoltaic cells and batteries that exponential growth is going to run into quick energy limitations.

A Quaker in a Strange Land said...

Having only considered this for the past 15 minutes... I think that if this were to unfold in this manner that the impacts to the banking system might be problematic, at some point, to further exponential growth in unemployment (the mirror image of the exponential growth in robotics.

This requires further noodling on my part...

Still, an interesting dilemma.

Unknown said...

Kurt Vonnegut in his book "Player Piano" laid out one version of a futuristic society in the USA that included robotic factories which produced all our "needs".

Greg said...

I'll reiterate my comment from yesterday: in advanced economies, the dependency ratio is set to skyrocket, and demand for elder care will go up even quicker.

Given the choice between allowing hundreds of millions of migrants from sub-Saharan Africa to do this work, or robots, I expect Europe, Japan, Korea, and most other advanced economies to choose robots... up to a point.

Secondly, I'd be suspicious of the data. Lots of things get called robots that are not. They are telepresence devices such as military drone aircraft, machines for doing surgery, and the like. These machines still need a human operator, and it's not clear what their effect on employment will be. Yes, they can do more, but by lowering costs, they may increase demand and therefore employment.

True robots (machines that operate autonomously) are so far only employed in manufacturing, which is about one-eighth of total employment and falling. Before robots can be deployed in less controlled environments, a lot of problems in machine vision have to be solved.

Therefore the chart is probably overstating the rate of growth of true robots.

Third, I note that the last part of the chart has more than a billion robots being produced in a year. At an average cost of $20,000, that would be over $20 trillion. If global economic growth is 3.5% per annum, that would be more than 16% of the global economy. It can't happen.

Stephen B. said...

There was a book, Lights in the Tunnel, written by John Ford, that discussed this subject at length. (

Of course, automation isn't just strictly robots, but almost any kind of automation that reduces the need for people, from voice mail phone answering systems, to legal software that helps one lawyer write up a case...something that used to take a team of lawyers...and that allows a single lawyer to subsequently present it to court... the list is nearly endless.

I have been considering this question of increasing automation for some time and unlike Glen, I'm not so sure that higher energy prices and lower availability are going to derail the automation boom for many automation systems take relatively little energy. Imagine a desktop computer or even a notebook, assembling that legal case I was talking about, reducing the need for the law firm to drastically cut the size of the office space it leases, and hire far less people, provide HVAC for them, lighting for them, etc. Then too, look at my industrial potato farmer neighbor in Maine...His large potato harvester, self-powered but guided by a tractor he already has, replaces probably 2 to 4 dozen potato workers. Sure, it uses some diesel fuel....both machines in tandem use maybe 2 to 3 gallons per hour, but it cuts the need for a lot of people to drive to the farm, and even the old system of harvesting required the use of many smaller trucks and wagons in the field. There are countless other examples of how automation cuts energy use, rather than increasing it, except perhaps, for the embodied energy used to construct such systems, but even that, using the notebook computer example, we see a lot of office space being replaced with a single, small computer.

I think too that, with all due respect to Burk Brawn up thread, I don’t think we can all be “thinkers” with automation all allowing us to be creative. I say that because I just don’t think that everybody *can* be a scientist or advertising artist. I work with a segment of the population (at-risk kids) that has all kinds of educational issues and they won’t be easily re-employed in this brave, new world that’s being created. The world really can use a supply of simple jobs that give people of limited creative talent something to do with their day and yet, that supply jobs is shrinking every day.

And what, then has been the result? Well, we’ve seen the dependency class grow like never before. Large parts of our society sit home now with little to do, bought off by payments from government and corporations to stay home and not cause trouble. In short, the system doesn’t know what to do with these people anymore, so it shares a small part of the wealth, resulting from automation, to buy the unemployed’s cooperation.

I happen to think that there is a huge, huge, huge, untapped resource in those unemployed people for them to get out of the apartment, become self-employed, and do *something*, but nevertheless, it presently is what it is. Lots of people do not have the creativity to find new occupations for themselves in a world increasingly taken over by corporations, government, and the corpora-government’s automata.

I remain completely against the idea of socialism as it completely removes the work incentive from many people. But I likewise am very troubled by what I see in our society where machines, corporations, government laws (controlled by corporations), and the attendant automata seems to be engaged in a race to see how much wealth can be concentrated in the hands of a VERY SMALL super elite, all while throwing a few bones to the masses, placated into staying in their apartments.

There are a good many great minds that frequent this blog of yours Stuart, but they best not assume that they are so smart that they too cannot be obsoleted by this automation, centralization trend.

I wish dearly that the end of cheap energy spelled doom for this whole process, but I’m just not sure that is the case.

Stephen B. said...

My mistake, The Lights in the Tunnel was written by Martin Ford, not John Ford.

Stuart Staniford said...

Jared: I strongly disagree - the Saudi rulers forked over the dough not because they were indifferent to the loss (it was a great deal of money) but because they felt credibly threatened by the examples of Tunisian and Egyptian regimes being toppled just down the street. Thus they felt it necessary to make large concessions in order to maintain the stability of the regime.

Stuart Staniford said...

jim/Burk - I see these kinds of arguments a lot but I think they are societally naive. If you want a prototype for the future, just look at young men today who can't find a job. We already have almost 20% of working age men not working - higher than that for the uneducated - so is this a happy situation with adequate provision for a comfortable life, lots of leisure time for creative expression, and decent social status? Far from it - it's some mix of young men in bad neighborhoods hanging out on street corners, people moving back in with their parents, disability fraud etc. And a completely unprecedented fraction of the population in prison.

Stuart Staniford said...

Glenn: yes, I would posit that as both necessary and feasible.

Stuart Staniford said...

Greg: I plan to address your argument further in a future post, but Stephen B has already outlined the basics of the answer - human worker insist on only working 8-12 hours and then driving a big vehicle home to a big house. Robots (and I will let them stand in for automation in general) just stay on the job and work all night too and don't insist on the house/car/vacations/etc. That frees up an enormous amount of resources that can be invested in a higher growth rate.

Greg said...

Stuart - on the contrary, increasing the utilisation of capital (from a third of a day per day, to 100% of it) would decrease the investment required for a given increment in production.

Together with the decrease in demand resulting from "releasing" all those workers, the decreased investment requirement* would cause a dramatic drop-off in demand. Robots won't be making anything if no-one can/will buy it.

This is a recipe for prolonged economic depression, not a boom. That's actually worse. Not only do we get hundreds of millions with no prospect of a job, the crying needs of the old and infirm (and the young and the in-between) go unmet.

The first step of the way out of this is for robots to do work for which there is latent demand, but which people can't or won't do - it's too expensive to use people for the work. Home help and geriatric nursing support (in the great numbers expected to be necessary) qualify, as do undersea mining and a few other exotic occupations.

As soon as robots move into existing work to the point that net employment starts decreasing, though, the crash in people's net wealth and their expectations for the future delays the further deployment of robots.

This comment is already too long, but I want to point out that concentration of wealth is an intrinsic feature of capitalism under uncertainty, as anyone who has played Monopoly by the rules ought to realise. You could also consult Fernholz and Fernholz (2012), "Wealth distribution without redistribution".

The choice is "socialism" (redistribution) or apocalypse. How crazy are we?

*I'm using investment in the economic sense: an increase in the quantity of productive machinery and structures; or an improvement in processes. If robots can work three times as long as human workers per day, they need only one-third the quantity of building for the same level of production. Buildings are the greater part of investment.

Greg said...

I should have provided a link for my assertions about the old-age dependency ratio. Here it is:-

United Nations Economic and Social Affairs, Population Division: 2010 estimates for 2010-2050, medium scenario.

You can display the curves for up to 5 countries at a time by Control-clicking in the "Country" selection box.

sunbeam said...

"That frees up an enormous amount of resources that can be invested in a higher growth rate."

At that point, growth for what?

Who exactly is going to consume what all this new productive capacity can produce?

I think there is another issue you've touched on very lightly so far, except as you've discussed issues with the dole.

That is what are most people going to be doing to earn income so things will be made for them?

Kobayashi said...

I think a lot of people here are underestimating people’s needs to be in a human society/community and their need to have interaction with real people when receiving a service (or just leading a life and spending time). People are very social (and here I don’t refer to social media and applications): they want to be with a spouse, family, friends, among people. [Having said that, I see realise that statistics show a different trend, especially in the Western-world, that is an increasing number of people live alone etc.]

But just as there is an assumption that continuous gdp growth is the natural state of any economy and people want that a political preference, there seems to be an assumption that technological innovation is necessarily good and should (or will) spread to every segment of human society.

But a lot of people actually do not prefer GMO seeds, like to sip coffee in Italy and talk to friends, want to by Burberry coat sewn and put together by real people etc. Sure, these people may be the affluent, but I think a lot of people would always prefer to have a subjectively better quality/happier life than to work more, amass and consume more industrial products, be more ‘rational’ or technocratic (and increase the GDP in the meantime) even in emerging markets (plus happiness figures tend to be higher in many poor developing countries than in heavily industrialised ones).

Seth said...

To limit the free-rider problem, we expect each other to "earn" a share of our total economic output. To "earn", we require people to have a "job". So the income distribution problem is effectively a 'work distribution' problem.

We don't have an adequate means of sharing out the work. Culturally, we aren't ready to think of idleness -- even a shortened work week! -- as something anyone is eligible for unless they already have a vast amount of money. For those who *do* have great wealth, cultural norms applied to the rest of us are simply suspended. We manage this trick for very high status people, but it only seems to reinforce our desire to punish idleness in the powerless. The resentment is turned away from those out of our reach and towards those we *can* kick.

This is the problem of human nature (or culture) which blocks the charming utopian visions described in Keynes "Economic Possibilities for our Grandchildren" ( ).

Jared said...

Thanks for your comment, Stuart. I don't think our disagreement really amounts to much: the absolute amount of money involved isn't important, as long as the government can afford it without painful contractions in the elite's living standards. In fact, in my example, the higher the amount of money that repressive autocrats are willing to fork over at the first sign of unrest, the brighter the outlook for future cases.

If we had persistent 30% unemployment right now for non-robot reasons, providing a basic standard of living for the unemployed would be possible (greatest nation in history and all that) but very painful. If we have 30% unemployment in 2040, the cost will be more manageable. Under those circumstances, I expect the Paul Ryans of the world to be slightly more generous than, say, Gaddhafi, as soon as signs of serious unrest appear.

TiradeFaction said...

I'm curious, what do you mean by "people who matter"? As in strictly a political/voter sense, or something else?

Stuart Staniford said...

Tirade - I meant "people whose views the political system takes seriously" (ie especially the upper middle class and the rich, and especially not the poor)

TiradeFaction said...

Hmm, maybe we need to get the poor to matter more in our political system. Organize! (May day ;) )

Michael Cain said...

Where will the electricity to power all of these robots come from? I'm asking that as a serious question. Consider the US Eastern Interconnect. In 2010, the states of the Eastern Interconnect accounted for 90% of all the electricity generated in the US. 47% of that from coal, 22% from natural gas, 21% from nuclear, 5% from conventional hydro, and 5% from all other sources combined. In 2035, less than 25 years from now, almost all of the nukes will have reached the end of their license extensions and be retired. Any attempt to be serious about CO2 requires replacing some large fraction of those coal plants (or adding carbon capture and sequestration, a very expensive undertaking). Despite the current glut, there are limits to how much NG can be extracted annually.

It seems to me that in 25 years, the eastern US is going to be less concerned with millions of robots taking away jobs than simply with the problems of keeping the lights on reliably.

Nick G said...

As long as scarcity exists, robots won't be primary cause of a structural unemployment problem.

We certainly are far away from eliminating scarcity: people in the US need a lot more services (health care, education, eldercare, child care). People in developing countries need a lot more hard goods. We have a lot of work to do fixing environmental damage (fisheries, ocean acidification, climate change, species extinctions, etc, etc).

Robots will cause the same transitional problems we have always had: what to do with farm workers, horse carriage manufacturers, railway workers, auto workers, etc, etc, etc.

Nick G said...


We have plenty of wind and solar power. Plus, I expect those nuclear plants will be re-extended.