Auto-Industrialism

DIY Capitalism and the Rise of the Auto-Industrial Society

Peter Murphy

• Publisher: SAGE Publications Ltd
• Year: 2016
• ISBN:
  9781473998810
• Edition: 1
• Topic: Sozialwissenschaften
• Subtopic: Soziologie

DIY check-outs, drones, self-driving cars, and e-government all are signs of the coming auto-industrial age. Will this end in mass unemployment or will new kinds of work emerge? Will 3D print production, desktop workshops and mass customization make upforlost blue-collar jobs? What will happen to health and education in the auto-industrial age? Will machines replace teachers and doctors? What might the economic and social future dominated by self-employment and a large DIY industry look like? Peter Murphy?s lively, provocative book addresses these questions head-on.

1 Work

THE ROBOTS ARE COMING

A lot of the technology that the auto-industrial society uses is familiar to us. This is typical of technological change. It begins slowly. It tends to have a long gestation period before being widely applied. The idea of automation is an old one. In classical Greek antiquity, Aristotle speculated about machines that moved by themselves. He imagined self-moving looms and lyres. He thought that if such machines were to exist then neither servants nor slaves would be necessary. Little under the sun is new. There were designs for automata in the ancient Hellenistic period. Leonardo da Vinci sketched an idea for a robot knight. Nicolas-Joseph Cugnot’s first approximation of the automobile — a steam-powered tricycle — appeared in 1769. Karl Benz’s similarly tricyclical Motorwagen was patented in 1886. Ransom Olds began factory production of automobiles in 1902 in Lansing Michigan. The same kind of slow-burning development is true for robotics as well. The Czech playwright Karel Čapek coined the term ‘robot’ in 1920.¹ Norbert Wiener formulated the theoretical principles of robotics in 1948. The first factory robot, the Unimate, was installed in a Trenton New Jersey plant in 1961.

Since the rise of the automobile, and then the personal computer, human–machine interactions have become pervasive. This reflects the fact that human beings on the whole are comfortable with machines — not just at work but also at home. People get endeared to their cars. They populate their houses with washing machines and dish washers. Computing devices become extensions of themselves — indispensable, convenient, personalized, necessary and time-saving. After two decades of personal computing, things that once would have been unthinkable are now very conceivable. In the future much aged care assistance will gradually be turned over to robots of one kind or another. The novelty of this will quickly disappear and care-bots will seem no more unusual in a person’s life than an automobile. In many cases they will provide auto-mobility. They will also reduce the incidence of abuse and cruelty in nursing homes where repetitive daily feeding and cleaning tasks breed anger and hostility in support staff. That robotic machines are unfeeling, like personal computers, is a plus not a minus. Not all human feelings are good. Nor are all human interactions beneficial. Take the case of autistic children. Two of the impediments to autistic children learning are their short attention spans and the fact that human facial expressions overstimulate them. This makes teaching them in a regular way difficult. An instructor or parent has to have endless patience in repeating the same exercise. Dispensing with facial expressions is even more difficult. But this is not so for robot instructors. They run the same routines endlessly; and they do not have to have faces, just voices.

The history of technology is one of long slow development punctuated by dramatic upswings. Successful technologies evolve slowly and then take off. Currently we seem to be in a take-off period. The vacuum cleaner robot, Roomba, illustrates the point. It was introduced in 2002. Ten million of these units have been sold worldwide. We are seeing the application of robotics and related automated processes accelerating. Examining data from seventeen countries, Graetz and Michaels in 2015 observed that during the period 1993 to 2007 the use of robots had raised the growth rates of the countries surveyed by about 0.37 percentage points overall.² Robot density (robots per million hours worked) increased 150 percent, the nominal price of robots fell by half and the quality-adjusted price fell to one-fifth. The function of machine automation is to replace labour. The researchers found that unlike more general computer automation robots did not polarize the job market by shrinking mid-tier employment and expanding the low- and high-tiers. Rather factory robots reduced the hours of low-skilled employees and (to a lesser extent) mid-skilled employees and did not affect high-skilled workers. In 2013 industrial robot sales worldwide increased by 12 percent.³ In countries like the United States, manufacturing moved offshore after 1970. It is now returning in the shape of high-tech peopleless factories. High-tech factory automation is not just happening in post-industrial nations. It is also occurring in China. Industrial China can no longer convince the generation born after 1990 to take factory jobs. They are not interested. Even in the world’s factory, the machines are rising.

More striking still algorithmic and automated processes are replacing white-collar employment.⁴ As with blue-collar technological unemployment this is not unprecedented. Office jobs have been automated before. Automatic teller machines, pioneered in the late 1960s, replaced bank teller jobs. What is new, though, is the scale of the looming change. Take the case of tax agents. The tax agent was a classic post-industrial mid-tier office occupation. After 1970, young people streamed into universities, took business degrees, exited with qualifications and became tax agents. At a conference in Sydney in 2014 the Australian Tax Office (ATO) warned tax agents that within 2 years Australian businesses would be able to report directly to the ATO every time that a sale or purchase was made, a payroll transaction occurred, or an employee was hired or fired — thus in principle eliminating most of the functions of the tax agent in preparing business activity statements and tax returns.⁵ In other words algorithmic digitization makes possible large-scale disintermediation. In this case the service agent is replaced by machine–machine interaction: an industry business machine communicating with a tax office business machine, with data entered directly at the industry end.

Expect the tax agent example to be multiplied many times over in the coming decades. In an influential 2013 report, Carl Benedikt Frey and Michael A. Osborne estimated that 47 percent of US occupations had a significant probability of being reduced or eliminated by computerization.⁶ The researchers looked in detail at 702 occupations. They compared occupational profile data from the US Bureau of Labor with studies of recent advances in machine learning and robotics as well as the prospect of offshoring (via the Internet) routine information-based tasks to lower-wage countries. In 2014, they applied their model to the United Kingdom. They concluded that 35 percent of contemporary jobs in the United Kingdom (and 30 percent of jobs in London) were at high risk of disappearing over the next decade as a result of computerization.⁷ Jobs in sales, office and administrative support, services and management were notably vulnerable. The ‘education, legal, community service, arts and media’ cluster also showed vulnerabilities in spite of the common assumption that creative and social jobs defy automation. As it is the law and creative professions are already heavily computerized. The routine aspect of any computer-mediated role is liable eventually to be automated. One can also easily foresee the transfer of automated learning technologies into the regular classroom for teaching repetitive tasks like the times table or the periodic table.

Computerization is not a new process. As Frey and Osborne note, over past decades computers have already done away with the jobs of bookkeepers, cashiers and telephone operators. The economists Henry Sui and Nir Jaimovich observe that routine office and administrative support jobs have been declining since the 1980s.⁸ The numbers of secretaries, bookkeepers, filing clerks, mail sorters and bank tellers have visibly shrunk. Automation and computing reduced demand for these white-collar occupations. From 1982 to 2012, routine occupations as a share of total US employment fell from 56 percent of employment to 44 percent. This structural shift became visible after the US recession of 1991 and again after the 2001 and 2009 recessions. The series of recessions during the second half of the post-industrial era repeatedly triggered reductions in routine occupations. As economic recovery occurred after each recession, the routine labour force did not return to its previous levels. In effect, computer capital began to replace repetitive labour.

As computers grow more sophisticated and (crucially) cheaper, computer capital increasingly supplants routine labour cohorts. To a significant extent already, personal assistants and secretaries have been replaced by mobile phones, digital assistants and personal computers. The decline in the real cost of large-scale computing services and robotic devices enables this substitution. It makes economic sense for employers to swap relatively expensive labour for cheaper computer or computer-controlled capital. The advances in computing mean that computers increasingly are able to replace labour where the tasks performed are repetitive and well-defined.⁹ A well-defined task is one that is broken down into the most elementary parts, usually a series of steps. Repetition means repetition of the same set of steps. These are tasks that are in effect defined by procedures. In many cases these procedures branch. Such branching can be defined by programmable decision trees. Computer algorithms provide if-then rules for computing what to do. As automated decision tree systems grow in sophistication so does the range of tasks that become computable — enabling an ever-larger range of undertakings to be automated.

This means that automation is now extending into areas of employment such as driving a car through city traffic that previously were routine for human beings (most people can learn to drive a car) but seemingly impossible for robotic machinery. Some things that are routine for human beings such as recognizing handwritten letters still remain difficult for machines. But the scope of computerization in recent times has significantly expanded. This is due to advances in data mining, digital sensing, machine vision, computational statistics and AI. The result is that an increasing range of human tasks is being computerized and the pace of computerization is accelerating. A distinct intense wave or phase of automation is occurring. This accelerated phase extends roughly from 2006 when, after a lengthy period of disinterest, investors once again became interested in the potential applications of robotics in industry.¹⁰

In 2007, Bill Gates published an article in Scientific American predicting the coming of a new industry (aka robotics) that was comparable in nature to the personal computing industry that Gates pioneered in the 1980s. He envisaged a future world populated by robots doing laundry-folding, lawn-mowing, surveillance, floor-cleaning and food-and-medicine dispensing. We are some way away from a proficient laundry-folding robot but the rest are in use or nearly so. As Gates noted, it is no easy thing giving robots the kinds of capacities that human beings take for granted such as recognizing, grasping and navigating around objects. But he added that ‘researchers are starting to find the answers’ due to ‘the increasing availability of tremendous amounts of computer power’. This power is still far from the capacity of the human brain but it is sufficient now to enable robots to do an expanding number of routine tasks.¹¹ The intersection of computing power with today’s low-cost electronic sensors and the Global Position System has created a new technology spike. Like all such spikes, it is occurring on the back of decades of incremental development.

In its 2014 London Futures report, the consultancy firm Deloitte identified the kinds of jobs in London that have been significantly reduced already by automation in the 2001–13 period: library assistants and clerks, sales-related occupations, filing and record assistants and clerks, travel agents, counter clerks, PAs and secretaries, collectors and credit agents, pension and insurance clerks, account clerks and bookkeepers.¹² Worldwide across 2008–13, rising job types were iPhone developers, social media interns, data scientists, big data architects, cloud services specialists and digital marketers. Demand for high-end work continued to expand while demand for mid-tier office-and-sales jobs declined. The post-industrial era was an age of intermediation. It involved a big expansion of intermediate job layers and functions in medium-sized and large organizations. The auto-industrial era is removing these mid-tier jobs and roles. This is not just about technology. It also represents a cultural shift.

New technology creates jobs and eliminates jobs. The net result is what matters. Do the numbers of jobs created exceed the numbers of jobs eliminated — or not? The predictions of the 1970s were that computerization would cause a net reduction in the number of jobs. Mass technological unemployment would follow. That proved not to be true but it is just as interesting to understand why this was so. Job levels were sustained because government, education and health jobs expanded — or rather more precisely government, education and health bureaucracies expanded. Compliance became the post-industrial industry par excellence. Australia in 2014 had a GDP of $1,500 billion. It spent $250 billion on compliance — 15 percent of its GDP.¹³ In other words, eight weeks of Australia’s working year in 2014 was devoted to compliance activities. The effect of the post-industrial mania for compliance can be seen in American hospitals where today there is a minimum of 30 minutes paperwork for every hour of patient care.¹⁴ A further 25 perc...