It is time to stop quivering in our boots in pointless fear of the future and just roll up our sleeves and build it.
- Ray Pierrehumbert

Sunday, June 12, 2011

The Missing Automation Crisis

Among the things we are a half century behind in addressing properly is the economics of plenty. Let's start with one of the less controversial of the conundrums of the modern world: the disappearance of the Automation Crisis.

One of my great heroes, uber-geek Norbert Wiener, pretty much got himself into terrible trouble over this. He basically invented information technology (Though it was entirely analog and/or mathematical at the time, which is very interesting in itself. Most information theory taught these days is more based in Shannon's much more accessible ideas. But the whole network lives on top of an amazing hidden Wienerian world of transmission and detection that few people look at. While Shannon's ideas get more attention, we would not be where we are today without Wiener's work.)

Basically, he realized that a lot of mundane decision making, as well as mundane activity, would eventually be delegated to machines. Accordingly, he felt that low-skilled jobs everywhere were under a long-term threat, which would lead to a dangerous oligarchy of the skilled and educated. He construed this as a threat to the prosperity of the working class. As a mid-20th-century academic intellectual of obviously Jewish ethnicity, thus with Nazism being fresh in his mind, he in turn had little trouble extrapolating these stresses on the working class into a direct existential threat to pretty much everything he valued.

Wiener had already run afoul of the authorities by refusing to work with the Pentagon after WW II ended. As a mathematician first and an engineer second, Weiner's ideations had little to do with what was considered ideology in those days in some circles. (If it's obvious to anyone that closed loop systems like capitalism work better than open-loop ones like Stalinism it would be Norbert Wiener!!!) But this didn't stop J. Edgar Hoover from opening a dossier on the fellow. Fortunately, he ended up with the Douglas-Adamsish designation "mostly harmless" and was left alone.

Somewhere in those days, though, Wiener attempted to capture the attention of the labor movement, going so far as to have meetings with AFL-CIO chairman Walter Reuther.

Here is Wiener's astonishing letter to Reuther.

South Tamworth, August 13, 1949

Walter Reuther

Union of Automobile Workers

Detroit, Michigan

Dear Mr. Reuther,

First, I should like to explain who I am. I am Professor of Mathematics at the Massachusetts Institute of Technology, and Iam the author of the recently published book, Cybernetics. As you will see, if you know of this book, I have been interested for a long time in the problem of automatic machinery and its social consequences. These consequences seem to me so great that I have made repeated attempts to get in touch with the Labor Union movement, and to try to acquaint them with what may be expected of automatic machinery in the near future. This situation has been brought to a head by the fact that I have been approached recently by one of the leading industrial corporations with the view to advising them as to whether to go into the problem of making servo-mechanisms, that is, artificial control mechanisms, as part of their extended program.

Technically I have no doubt what direction my advice should take. My technical advice would be to construct an inexpensive small scale, high speed computing machine, together with adequate apparatus for putting the readings of photo-electric cells, thermometers, and other instruments into the machine as numerical data, and for putting numerical out-put data into the motion of shafts and other out-put apparatus. The position of these output shafts should be monitored by proper sense organs, and be put back into the machine as part of the information on which it is to work.

The detailed development of the machine for particular industrial purpose is a very skilled task, but not a mechanical task. It is done by what is called 'taping' the machine in the proper way, much as present computing machines are taped. This apparatus is extremely flexible, and susceptible to mass production, and will undoubtedly lead to the factory without employees; as for example, the automatic automobile assembly line. In the hands of the present industrial set-up, the unemployment produced by such plants can only be disastrous. I would give a guess that a critical situation is bound to arise under any condition in some ten to twenty years; but that if war should make the replacement of labor mobilized into the services an immediate necessity, we should probably have a concentrated effort put into this work which might well lead to large scale industrial unemployment within two years.

I do not wish personally to be responsible for any such state of affairs. I have, therefore, turned down unconditionally the request of the industrial company which has tried to consult me. However, it is manifestly not enough to take a negative attitude on this. If I do not put this information in the hands of the industrialists, it is merely a question of time when so obvious a method of procedure will be urged upon them by other people.

Therefore, the procedure which I shall follow depends finally upon whether I can get you and the labor interests you represent to pay serious attention to this serious situation. I have tried to do this in the past without success; and I do not blame you people for it, but since then there has been a turn-over in personnel among you and the present group of labor leaders seem to have transcended the point of view of the shop to a sufficient extent to make it worthwhile for me to make an appeal to you again.

What I am proposing is this. First, that you show a sufficient interest in the very pressing menace of the large-scale replacement of labor by machine on the level not of energy, but of judgment, to be willing to formulate a policy towards this problem. In particular, I do not think it would be at all foolish for you to steal a march upon the existing industrial corporations in this matter; and while taking a part in production of such machines to secure the profits in them to an organization dedicated to the benefit of labor. It may be on the other hand, that you think the complete suppresion (sic) of these ideas is in order. In either case, I am willing to back you loyally, and without any demand or request for personal returns in what I consider will be a matter of public policy. I wish to warn you, however, that my own passiveness in this matter will not, on the face of it, produce a passiveness in other people who may come by the same ideas, and that these ideas are very much in the air.

If you determine that the matter does not deserve your serious consideration, you will leave me in a very difficult position. I do not wish to contribute in any way to selling labor down the river, and I am quite aware that any labor, which is in competition with slave labor, whether the slaves are human or mechanical, must accept the conditions of work of slave labor. For me merely to remain aloof is to make sure that the development of these ideas will go into other hands which will probably be much less friendly to organized labor.

Under these circumstances, I should probably have to try to find some industrial group with as liberal and honest a labor policy as possible and put my ideas in their hands. I must confess, however, that I know of no group what has at the same time a sufficient honesty of purpose to be entrusted with these developments, and a sufficiently firm economic and social position to be able to hold these results substantially in their own hands.

I have a book ((The Human Use of Human Beings) which will be forthcoming with Houghton-Mifflin next spring which will bring these ideas to a head. If you so wish, I shall send you copies of the relevant chapters.

Naturally, I do not expect you to take these matters on my momentary say-so. If you show sufficient interest to be willing to push the matter further, I shall be glad to put my ideas both technical and social at your disposal, so that you will be able to judge them better.

Sincerely yours,

Norbert Wiener

Department of Mathematics

Massachusetts Institute of Technology

Cambridge 39, Massachusetts


Before long, to Wiener's satisfaction, Reuther was writing articles like this:


Inactivists will of course be delighted to note that neither was the recommended course of action followed nor did the dire consequences emerge, for three generations.

When we look at our current quandary, though, it may pay to ask how and why these things didn't happen. The simple theories of "growth" may be contrasted with the observational evidence that the average person in North America is not on the whole safer or more comfortable or happier than his or her predecessor. (Though the coffee and the razor blades are much improved!) So whatever the increased efficiency of labor is doing, it is not actually doing much for the benefit of the general population.

But the redistribution of wealth and allocation of responsibility as a matter of policy interest did not happen. Nor did massive underemployment. Indeed, in the intervening years, an expectation of employment of married female adults has actually greatly increased the proportion of the employed! So what happened?

Has the great chicken of the Automation Crisis finally come home to roost? And where has it been meanwhile?

20 comments:

Tom said...

Wiener in 1949: "I would give a guess that a critical situation is bound to arise under any condition in some ten to twenty years; but that if war should make the replacement of labor mobilized into the services an immediate necessity, we should probably have a concentrated effort put into this work which might well lead to large scale industrial unemployment within two years."

Civilian unemployment rate in

1949: 6.6%
1959: 5.5%
1969: 3.5%

http://webcache.googleusercontent.com/search?q=cache:EXmYLgzmV60J:www.gpoaccess.gov/eop/2006/B42.xls+unemployment+rate+1959,+1969&cd=6&hl=en&ct=clnk&gl=us&source=www.google.com

Dan Olner said...

Relevant Krugman: "remember, the Luddites weren’t the poorest of the poor, they were skilled artisans whose skills had suddenly been devalued by new technology."

There are also parallels with arguments about trade barriers generally: the tech effects are actually similar. Dean Baker wrote a short punchy piece (PDF) on this, noting that economists have been - until recently - part of the groups likely benefit from those changes.

I have some nice quotes lying around, I'll dig em out...

Arthur said...

The remarkable thing about employment is that it is as close to 100% of those who want to work as it is. After all, we can feed ourselves these days using, what, maybe 2% of the adult population, so why do the rest of us do anything at all? Why isn't unemployment over 90%?

If you think about that for a while you realize that, to a large degree, labor makes its own employment. That is, people who having nothing to do go around trying new types of work until they settle on something that other people find useful, and are willing to pay for. Sometimes that's hard to find and unemployment rises. Sometimes it's easier and unemployment falls. But automation has never and will never have much to do with it.

Keynes "General Theory" goes into this in the context of the Great Depression, obviously another time when useful work was hard to find, but I think somewhat misses the point on waste and usefulness of what people do. I don't know of any economist who has treated this in a way that makes complete sense, I'd love a reference or two.

Rich Puchalsky said...

I'm inclined to think that one reason the Automation Crisis didn't happen for the same reason that AI in general didn't happen. We're really good at building robot arms, and really bad at building robot brains.

Which is not to say that an assembly line can't be automated, because assembly-line robots make essentially no decisions other than whether to stop if something goes wrong. But all of those robot arms are expensive machines that require expensive maintenance. In an expensive-labor society, the drop in the assembly line part of industrial employment is to some extent camouflaged by needing to hire more people to maintain the machines. In a cheap-labor society, labor is still cheaper. The reason why labor can be cheaper is because the general increase in productivity means that it's really pretty cheap to pay someone just enough to keep them going.

Michael Tobis said...

I am closer to Rich than to Arthur on this one. Arthur's view is of a world where work is optional. This is the world we need to build, and the one Reuther was trying to describe.

Rich's view is of a world where resources are controlled by an oligarchy in the interests of the oligarchy. Work delivers necessities, not a surplus. This has led to the absurd situation in which we find ourselves, as I will try to describe in follow-on articles.

I think Rich's view does succeed in getting at how we avoided the unemployment crisis by instead sliding down a scale of well-being. It's true, of course, that there's far more wealth to go around, but it's not true that being honest, decent, and willing to work hard is enough to get you an adequate share of that wealth.

It once was true, in the advanced countries. And the avoidance of the unemployment crisis is not the same thing as the continued thriving of the segment of the population that used to be union members.

Michael Tobis said...

In defense of Dr Wiener, though, I should add that an immense number of micro-decisions are embedded in any serious automation process today.

Wiener's word "cybernetics" is today associated with AI. The story of Wiener's association with the emergent AI culture and his backing out from it has mostly been told from the latter's point of view, and has been hopelessly bungled, I suspect. I have seen an account where his wife was made the villain of the episode.

While personal tensions may have been involved, I think Wiener's abandonment of AI was mostly based in what to him would appear an insurmountable level of mathematical intractability. But "cybernetics" qua "decision-making in animals and machines" as Wiener understood it actually focuses at a much more elementary level: is my aim at this goal in need of mid-course correction?

Every time you are in an elevator which stops at exactly the floor level (and you are replacing an elevator operator who was a union member) you should pause and thank Norbert Wiener and his cybernetics. Meanwhile, the person who might have had a job running the elevator could blame him as well.

Rich Puchalsky said...

I should perhaps qualify my previous comment, then, by saying that we can build fairly good really lower-level parts of brains -- the parts that guide arms in course-correction midway to doing a very constrained and repetitive task. (Although somewhat higher level tasks for humans, like catching a ball or recognizing a face, are still difficult.) The machine's response to anything going wrong, though, is still very limited.

In a sense, I think that what Wiener feared has already happened. I just think that he overestimated the ability of machines that do low-level tasks to really take over industrial work. There is more high-level human thought involved than you'd imagine, just looking at the most mechanized part of the assembly line.

And of course the part about the oligarchy is correct, but I'm classing that as "what Wiener feared". Basically he was right, and the only reason he appears to have been wrong is that he overestimated technology, not that he underestimated society.

Arthur said...

My point, which perhaps I conveyed poorly, was intended to be independent of whether a society encourages high-wage or low-wage work, whether it has strong or weak systems of welfare for the poor, whether ownership is highly concentrated or well diffused among the population, and largely independent of the system of government.

Those things all make a difference in the distribution of wealth and happiness and all that. Technological change goes along with personal market and charitable choices and government choices on taxation and expenditures and central bank monetary policies in determining how the wealth circulating around actually ends up being distributed.

But my question was more - why does almost every society (the oil-rich Middle Eastern nations perhaps an exception) arrange itself so the vast majority of work-capable people actually do work?

And I think the answer is simply that any non-working person, even if their basic needs are met, would be more happy with slightly more money to spend, and so (if the society is free enough to let them) will take at least some time to pursue opportunities that provide utility to other people. The existence of that person's time and wealth in the hands of others creates a real incentive to work. Those opportunities may seem to have little direct utility - live music performances, personal trainers, financial advisers and the like - but they perform the economic act of division of labor in new ways that allow the wealth of all to increase. And they exist, I believe, given a free enough society, independent of any level of automation or cybernetics, to a degree that can always come close to full employment of all those who wish to work for these reasons.

But, as I said, I haven't seen an economic text that adequately addresses this in a way I believe - not that I've read many yet...

manuel "moe" g said...

> Arthur's view is of a world where work is optional.

I think what Arthur is saying is that the *output* of vast majority work is optional (the output of work carries a value, but the vast majority of work is for higher-level needs, not primary needs).

But the desire to feel one is creating and contributing value is very strong. So work is created. Instead of "find a need and fill it", you have "needs and work are created in near perfect tandem, up to nearly full employment".

This is an interesting idea. That primary needs and secondary needs are so badly labeled and thus thoughtlessly commingled, explains why systems are simultaneously resilient and brittle, every disruption has stories of resilience and stories of shameful breakdown.

No textbook delivers the necessarily really-existing hairy-scary economics and cog-sci to talk about issues such as this.

King of the Road said...

That the number of people working is within 10% of the number of people who want to work even during relatively bad times and within 25% of the number who want to work during the worst of times is something that has always left me wide-eyed with amazement.

That this was (seemingly) true when an adult's 140 watts was all there was and is still true when (in the U.S.) we're using 11 kilowatts per capita is even more amazing. I will say that seven year old children selling Chiclets at the Mexican border is a stretch in this paradigm (thanks John Michael Greer) but, in the OECD nations, it holds up pretty well.

I'm not sure that Arthur's explanation captures the whole of it but it seems clear that it's a closed loop system of some sort. And, like Arthur, I've not read a satisfying explanation of the mechanism in any Econ. text. I don't think that it's the Government (of whatever stripe) wisely pulling the levers though.

Rich Puchalsky said...

I think that the repeated "no economics text covers this" statement is a little bit silly. I wouldn't tell anyone to read Marx, any more than I would tell anyone to read just about anyone from the 19th century. Still, it's not exactly a mystery about why the unemployment rate never gets too high. If people are unemployed, they starve. Not literally, perhaps, but their quality of life is vastly reduced. This is a feature of the system, not a bug.

The point is to keep them working yet pay them as little as possible so that you can make a profit on their labor. If unemployment goes too high, social stability is threatened because people start starving en masse (again perhaps not literally, etc). If it goes too low, wages have to be raised because employers have to compete for workers. (See all of the economic B.S. about the necessary level of unemployment). So governments do not magically pull the levers and automatically succeed, but they do devote a lot of effort to making sure that unemployment stays within a certain band, neither too high nor too low.

Lastly, the official unemployment stats in many nations are scewed because they only count people as unemployed if they are looking for work and not finding it. The people who are happily not working, for whatever reason, often aren't counted as unemployed.

Michael Tobis said...

Again, I am siding with Rich for the present (we may part company as the position I am trying to develop here emerges). It is pretty much a key to the argument that I am working on making here that "full employment" is an artifice, not a natural condition. So at the least the argument to the contrary is inconvenient for my case.

But the fact remains that, rather than getting credit for rescuing the world from the brink of the deregulated chaos that ended the GW Bush, the present situation is that Obama is being blamed for the modest but stubborn (and rather unfixable) decline in total employment in the US, and that similar arguments are raging elsewhere.

I will contend that this focus is lunacy. It's not a point you can make effectively on Twitter. You can say "full employment is not a good goal" but you can't really describe what an alternative would look like.

guthrie said...

Rich, re unemployment stats - it is often worse than that - if you only count people actually recieving unemployment benefits, you miss out the people who perhaps aught to be on benefits but have broken some law/ been on them too long and had them cut. Anyone over 30 here in the UK knows about the 1980's, when the gvt fiddled the figures to hide 3 million or more unempoyed, and it hasn't really got much better - they now report "Benefits claimants" as if that covers everything.

Rich Puchalsky said...

"the present situation is that Obama is being blamed for the modest but stubborn (and rather unfixable) decline in total employment in the US, and that similar arguments are raging elsewhere"

Well, there we do disagree -- about the "unfixable" part. Official unemployment in the U.S. is somewhere over 9%, which is well over what economists regard as the natural level of employment. What Obama was advised to do about this was fairly simple Keynesianism -- having the government spend money, employ people, create demand etc. Obama did do this, but to an insufficient degree, and when it became clear that more spending was needed, he'd already trapped himself by having talked about the nonsensical need to reduce the deficit for the last year. In Europe, it's even worse: countries are being told to adopt grinding austerity policies that can only hurt.

I'm not speaking from a leftist perspective here; I'm summarizing a very standard Keynesian one. If economics has any science to it at all, that's what it says. The arguments that heightened U.S. unemployment is unfixable generally have to do with the unemployment being "structural" (a term of art), and the data do not support this conclusion in any way.

I suspect that your argument is going to be at a level of generality at which the particular current situation doesn't really matter, though.

Arthur said...

I'm not sure why anybody thinks I would disagree with Rich - I haven't read Marx, perhaps I should. Is there a more modern update that addresses the same issues for an information-age society?

I completely agree that the present unemployment in the US is very fixable and the government, for reasons largely related to the Senate's super-majority veto power granted to the Republicans in 2009-2010, and now thanks to the Republican control of the House as well, has failed to do what is necessary to fix it. I quoted Keynes in my first comment. The solution is government spending as a temporary measure to get things moving. The most logical thing for that spending to be focused on is the one that would solve other problems as well: investment in renewable energy, rail transport and efficiency measures. The required level of government spending is not terribly large, and the investment will be self-sustaining and lead to greater employment in the long term as our nation's wasteful expenditures on fossil fuels and their associated health and environmental costs are reduced.

There are powerful interests opposed to this solution. I think we're all familiar with them. They have control of one of our major political parties and much of the media landscape in one form or another. At least a handful of NY Times columnists and other luminaries have pretty clearly communicated on these points, but as with climate science there's a lot of noise out there pretending that nothing can be done, or that government spending is the problem, rather than a solution.

But all that is a parochial temporary US problem, I have some faith it will be resolved, if only by the harsh reality of continued failure of current policies.

Shining Raven said...

There is more relevant Krugman on this. I think you have to take a view of the whole, complete economic activity of a nation, not just one aspect where large efficiency improvements occur.

Krugman gives a nice model for this in his "Accidental theorist". The article is still online at slate.com

http://www.slate.com/id/1916/

If I understand the argument correctly, then you are in danger of falling into the same trap.

Automation in the auto industry will indeed lead to layoffs in the auto industry in the short term, but that does not say anything at all about what happens in the economy overall.

Seriously: In the 19th century, most people worked in agriculture. These jobs are all gone, does this mean that everybody who should have worked on a farm (50 % of the population or thereabouts?) in the 19th century is now unemployed? Of course not...

And I agree with Rich in that I believe there are economics texts that get into all of this...

King of the Road said...

"if they don't work, they starve." Really? Then many of my in-laws must be zombies. Believe me when I say that there is a whole segment of society who has chosen not to work, but has the latest athletic shoes, flat screen televisions, broadband internet, plenty of children, cars to drive, food to eat, etc. Do they wish they had more money? Yes. Does this distinguish them from 99% of society? No. This is not a Republican theory of mine (though I'm not a Republican), this is observational fact.

A long essay on how these families and individuals provide for themselves could be (and, no doubt, has been) written, but mostly it's a combination of family ties, governmental programs, and black market activities of one sort or another.

But, when these people decide that, for one reason or another, it's time to work for a while, they wind up working. Obviously, there are feedback mechanisms at work here. Perhaps there are plenty of texts out there that address this balance, but none that I've read (not a lot, admittedly) satisfyingly covers these mechanisms.

Rich Puchalsky said...

I think that Shining Raven is right. Where Krugman goes wrong, in the link provided, is that he assumes that there is no upper bound to consumption. That's clearly not true, for all sorts of reasons that I'm sure that MT and everyone else reading this is familiar with. But within the model of unbounded consumption, Krugman's point holds, pretty much.

Next King of the Road:
"mostly it's a combination of family ties, governmental programs, and black market activities of one sort or another"

I'll pass over the racist implications of "the latest athletic shoes" being number one on your list of things that this segment of society is supposed to have. Well, I won't exactly pass over them -- I'll just allude to them, sigh, and go on. I'll even accept your anecdata as proving, for the sake of argument, that this segment of society exists.

Who supports them? If it's family connections, then their families must be working, or must be independently rich off of assets. The number of people supported by the latter is not large. If it's black market activities, then those are work.

The only thing that calls the "work to live" statement into question is governmental programs. And this is precisely the area of conflict in many societies, between people who want governmental support to spread the benefits of high productivity around, and people who want the threat of poverty to always be there to force people to work.

Dan Olner said...

MT: " `full employment' is an artifice, not a natural condition."

It occurs to me that, if everybody was happy with the level of wage labour they were putting in, that would be full employment - regardless of what the actual hours worked were. Unemployment isn't measured by some absolute number of things-not-done, only by the number of those who want to work but can't.

So how would you know when you've arrived at the kind of work-leisure-automation world you'd consider a utopia? What sort of measurements would you need, given how important self-reporting is to the issue?

King of the Road said...

Rich,

The only point I'll concede is that black market activities can be considered to constitute work. The families to which I specifically refer are not rich, far from it. They utilize a small number of jobs, "communal" resources, section 8 housing, food stamps, payday loan companies, etc. For medical services, they use the emergency room. They put a small amount down at "we carry our own credit" auto dealers and, when the vehicles are repossessed, they repeat somewhere else. And, in fact, they borrow from me and don't pay it back - something I anticipate when I "loan" the money. As I said, when they decide that work would supplement their income or provide for some temporary need, they look for and find some paying work. They look at this as normal.

As to racist, please understand I'm talking about a broad spectrum of my own in-laws and their extended family and friends and very specific activities on their part. They are asian pacifica islander, caucasian, latino, african american - really an ultra diverse group.

In any case, it's still a mystery to me what mechanisms are in place to make the peanut butter and the bread come out more or less even when the difference between very good and very bad is 5 percentage points.