Tuesday, July 31, 2012

Robocalypse Now!

A few weeks back in his ongoing series of failed end time prophecies, John Michael Greer included a section on The End of Work, and how predictions about this utterly failed. What he misses is that more thoughtful commenters on these topics were mostly borne out over time. This is one apocalypse that appears to have some merit, and grouping it with wide-eyed end-time prophecies does a disservice to some very real issues. Just as cornucopians believe technology will solve all our problems, sometimes there is a tendency for people who know better to go the other way and completely dismiss the effects of rapidly advancing technology on society. Neither approach is helpful.  Advancing technology and Peak Oil are not mutually exclusive. As I've done a fair deal of writing on this very topic, I feel a response is necessary.

In his debunking the end of work scenario, Greer makes a number of classical mistakes common to critics of this idea:

1. Seeing the robot revolution exclusively as humanoid automatons like Rosie the maid from The Jetsons or Robbie from Lost In Space as doing all our work for us like some sort of cartoon mechanical slaves. This view may have been a staple of techno-utopian science fiction novel covers, but unfortunately it tends to obscure reality. I define a robot as any machine that does human work, based on the original meaning of the word – robota, the Czech word for worker. Robots are workers, plain and simple, and any machine that performs human work is a robot by my definition. They look like whatever they need to to get the job done - in a car factory they may look like giant armatures, but the fact is, a robot may be nothing more than some silicon chips and look more like an oversized computer printer than the robot from Metropolis. Anthropomorphizing robots does nothing but obscure the real issue. A good example is the medical scanners that read and diagnose X-rays. These look nothing like people, but they do the jobs of people quite well.

2. The second mistake made by people who reject his idea is that it’s necessary for robots to replace all human workers. This is ridiculous and no one seriously proposes this. If nothing else we will need people to design and fix the robots! We will still need managers, scientists, engineers and creative thinkers, of course. But how many of these jobs are required? Just as deniers of Peak Oil insist that it’s about running completely out of oil, deniers of the effects of automation claim that it’s about completely running out of jobs. Both are straw men, of course – Peak Oil is as much about turning to lower quality and harder to get sources of carbon as it is about running out. Similarly, automation is about not creating enough jobs to supply a growing population. As we’ve said so often here, you need to create over a million jobs a year in the United States just to keep up with population growth, and there are similar numbers in most industrialized countries. Now that formerly agricultural nations are industrializing and women are expected to be in the workforce, this is an even more dire number. When those numbers are missed, as they have been consistently now for years and years all around the world, you have a crisis, even when the majority still have jobs.

Algorithms are the key to this whole thing. Essentially, any repeatable series of steps can be described by an algorithm. And in modern industrial society, that’s all the vast majority of jobs are – a series of repeatable steps. Anything that can be described by an algorithm can be automated, and that’s probably three-quarters of all jobs. Add to the fact that most jobs are totally unnecessary busywork. You need a certain amount of people to keep order, keep the utilities running, grow food, keep the shelves stocked, and fix what’s broken. That’s a very small number of the population. See this post: http://www.oftwominds.com/blogmay12/labor-paid-work5-12.html

The real issue is not a science-fiction future of robots taking over but technological unemployment, and it has been an area of real concern for economists from Karl Marx to J.M. Keynes.

3. Forgetting that part of the robot revolution is not just automatons but also increased productivity due to things like the computer. Mechanical devices like bulldozers, cranes and backhoes have reduced the need for human muscle, but computers have decreased the need for human brains. As an example consider a structural engineer in 1965 armed with nothing more than a slide rule and maybe a simple calculator. How many engineers would it take to plow thought the calculations required to build a Sears tower? How many draftsmen had to sit at tables cranking out hand-drawn sheets? Now with computers, small teams of engineers can plow through calculations in a fraction of the time using advanced computer models, and small teams of architects can model the entire building in 3D, changing things on the fly (this is how it works today). I could make a similar case with an accountant armed with an Excel spreadsheet who can now do the work of a dozen accountants from 1965. As is so often stated, the average worker has more computer power on their desks than did the original NASA team that put men on the moon. Do you think that might increase worker productivity just a bit?

To date that excess productivity has been channeled into more complexity, more volume, and shorter timeframes. Bigger, faster, cheaper, more. So those architects and engineers worked on ever-more buildings, larger, and with much more complex designs and framing. And they turned them out on shorter and shorter timeframes. But that can’t go on forever, and it’s finally coming to a halt. And without the extra volume of activity, there’s just no need for those extra workers anymore.

4. Forgetting that people providing services for themselves via the internet is as much a part of the robot revolution as actual robots, and may be an even bigger driver of job loss! Have you ever used and ATM, or a bar-code based scanning station at a store? Guess, what, those are robots in my book. Have you ever downloaded a song instead of going to a CD store? Have you ordered a book or any other goods off of Amazon.com? Have you booked a trip on Kayak or Orbitz? This is a huge part of the robot revolution that his almost always ignored. Guess what, all of these contribute to job loss!

When America deindustrialized, economists touted the “service economy” as the absorber for all those displaced workers. Now the service economy is looking pretty shaky (leaving aside the crappy wages and benefits, and the fact that these jobs are nontradebale and not value-added), because anyone with smart phone can do these things for themselves – book a trip, buy a movie, buy a stock, transfer money, learn a language, etc. Bill Gates has termed this frictionless capitalism. Some call it simply the elimination of middlemen. Good for consumers, but middlemen need jobs too. Already we see so-called "brick-and-mortar" stores closing down. Now Amazon plans same-day delivery all over the U.S. and has purchased the largest factory automation company in the world. A sign of things to come? See this article: http://www.slate.com/blogs/moneybox/2012/04/06/the_end_of_retail.html. Being able to do things for yourself is great, but it has consequences for the overall economy.

And let’s not forget that you can now hire workers to do cognitive/professional/clerical tasks literally anywhere in the world. I can now have an accountant in New Delhi armed with a laptop and Excel (see #3 above) do all my accounting for cheap. While not strictly robotics, this is another outgrowth of computer technology that has a tremendous impact on jobs. And while enthusiasts tout the benefits of online learning in educating the world, for practical purposes this dilutes the value of education. When everyone can become a “knowledge” worker thanks to the internet, such workers will be bought for pennies one the dollar by corporations who enjoy monopolies over nearly every industry (and already are).

Pretending that there are all these unmet needs that are out there and that people will just all start new companies and become self-employed is living in a state of denial and delusion. But denial and delusion is our specialty.

6. Forgetting we’ve only just begun to scratch the surface. We’ve only begun to exploit the "long tail" of automation, as Kevin Drum points out. Many of these innovations are only a decade old, some barely a few years old. It’s a wide open field for entrepreneurs, but today’s online entrepreneurs are eliminating as many or more jobs than they create, despite our culture’s fetishization of entrepreneurship as the solution to all our problems (e.g. Thomas Friedman’s claptrap). This is called the back-half of the chessboard problem.

With one in four young people unemployed in the U.S. and one in two in parts of Europe, and with unemployment causing instability in places like the Middle East, I find it hard to dismiss this problem as casually as Greer has. If we factor in underemployment, temporary jobs, people working below their abilities, seasonal employment, part-time work, etc., the situation becomes even worse. If this is not a jobs apocalypse, then what is?

For his part, Greer argues that the rising costs of energy and falling wages will make job scarcity a thing of the past as it suddenly becomes cheaper to employ people than machines. Eventually, maybe, but there is currently no evidence – none, zero, zilch - that wages are falling low enough to shelve automation in manufacturing, agriculture or professional services. The recession has only increased the drive for automation in tandem with the increasing abilities of this technology. Is he unaware of people buying his books over the internet? We're already several years past the global oil peak, and the higher energy costs are being absorbed without slowing down the drive for automation. Again, where is the evidence?

Now Greer is right in that there are some jobs that are a combination of both difficult to automate and low-paying that will probably continue to employ human workers, even if they can be theoretically automated. Two examples that come to mind are home health care aides and agricultural workers. Even if these could be automated - and there is some evidence that they can to some extent – the abysmally low wages in these fields will probably make it easier to use human workers, especially since there are so many. But that merely confirms the hypothesis – we cannot run an economy where the only jobs available are low wage scut jobs. Yes, factories are leaving China for lower-wage countries like Vietnam. But that's only for very simple manufacturing - high-end manufacturing is done in places like Germany, Japan, and yes, the United States in highly automated facilities that use less people every year. And as I’ve said earlier, oil costs will cause wages to rise, not fall, as the costs of living go up. Machines require no vacation, healthcare, benefits, etc.

So here’s one “apocalypse” that appears to be unfolding before our eyes, if we forget our silly preconceptions and recognize it as such. And unlike gray goo, this one is eminently solvable. It's not a problem of technology, it's a problem of how we've organized our society. What should be a boon has become a curse.

If you're still not convinced, have a gander at this article: The End of Chinese Manufacturing and Rebirth of U.S. Industry
The robots of today aren't the Androids or Cylons that we used to see in science-fiction movies, but specialized electro-mechanical devices that are controlled by software and remote controls. As computers become more powerful, so do the abilities of these devices. Robots are now capable of performing surgery, milking cows, doing military reconnaissance and combat, and flying fighter jets. And DIY'ers are lending a helping hand. There are dozens of startups, such as Willow Garage, iRobot, and 9th Sense, selling robot-development kits for university students and open-source communities. They are creating ever more-sophisticated robots and new applications for these. Watch this video of the autonomous flying robots that University of Pennsylvania professor Vijay Kumar created with his students, for example.

The factory assembly that the Chinese are performing is child's play for the next generation of robots--which will soon become cheaper than human labor. Indeed, one of China's largest manufacturers, Taiwan-based Foxconn Technology Group, announced last August that it plans to install one million robots within three years to do the work that its workers in China presently do. It found Chinese labor to be too expensive and demanding. The world's most advanced car, the Tesla Roadster, is also being manufactured in Silicon Valley, which is one of the most expensive places in the country. Tesla can afford this because it is using robots to do the assembly.

Then there is artificial intelligence (AI)--software that makes computers do things that, if humans did them, we would call intelligent. We left AI for dead after the hype it created in the '80s, but it is alive and kicking--and advancing rapidly. It is powering all sorts of technologies. This is the technology that IBM's Deep Blue computer used in beating chess grandmaster Garry Kasparov in 1997and that enabled IBM's Watson to beat TV-show Jeopardy champions in 2011. AI is making it possible to develop self-driving cars, voice-recognition systems such as Apple's Siri, and the face-recognition software Facebook recently acquired. AI technologies are also finding their way into manufacturing and will allow us to design our own products at home with the aid of AI-powered design assistants.

Even if the Chinese automate their factories with AI-powered robots and manufacture 3D printers, it will no longer make sense to ship raw materials all the way to China to have them assembled into finished products and shipped back to the U.S. Manufacturing will once again become a local industry with products being manufactured near raw materials or markets.

So China has many reasons to worry, and manufacturing will undoubtedly return to the U.S.--if not in this decade then early in the next. But the same jobs that left the U.S. won't come back: they won't exist.  What will the new jobs be? We can only guess. Autodesk CEO Carl Bass says that just as we have created new, higher-paying jobs in every other industrial transition, we will create a new set of industries and professions in this one. Look at the new types of jobs and multi-billion dollar businesses that the Internet and mobile industries created--these came out of nowhere and changed our lives, Bass says.

Carl Bass is one of the leading authorities on 3D printing and digital manufacturing, and I share his optimism that we will create an era of abundance.  But I worry if we will create the new jobs fast enough and distribute the prosperity.

1 comment:

  1. To be precise, "robota" means "corvée", the duty of labor a serf is obliged to his lord. Given the increasingly evident push towards neo-feudalism nowadays, the irony is rather bitter.


Note: Only a member of this blog may post a comment.