Logo IMG


Automation on the Job

Computers were supposed to be labor-saving devices. How come we're still working so hard?

Brian Hayes

On the Factory Floor

Perhaps the most thoughtful and knowledgeable of the early writers on automation was John Diebold, a consultant and author. It was Diebold who introduced the word automation in its broad, modern sense. He clearly understood that there was more to it than reducing labor costs in factories. He foresaw applications to many other kinds of work, including clerical tasks, warehousing and even retailing. Nevertheless, when he chose examples for detailed description, they almost always came from manufacturing.

Automatic control first took hold in continuous-process industries such as oil refining. A closed-loop control mechanism could regulate the temperature of a distilling tower, eliminating the need for a worker to monitor a gauge and adjust valve settings. As such instruments proliferated, a refinery became a depopulated industrial landscape. An entire plant could be run by a few technicians, huddled together in a glass-walled control room. This hands-off mode of operation became the model that other industries strove to emulate.

In the automation literature of the 1950s and ’60s, attention focuses mainly on manufacturing, and especially on the machining of metal. A celebrated example was the Ford Motor Company’s Cleveland Engine Plant No. 1, built in 1951, where a series of interconnected machines took in raw castings at one end and disgorged finished engine blocks at the other. The various tools within this complex performed several hundred boring and milling operations on each engine, with little manual intervention.

A drawback of the Ford approach to automation was inflexibility. Any change to the product would require an extensive overhaul of the machinery. But this problem was overcome with the introduction of programmable metalworking tools, which eventually became computer-controlled devices.

Other kinds of manufacturing also shifted to automated methods, although the result was not always exactly what had been expected. In the early years, it was easy to imagine a straightforward substitution of machines for labor: Shove aside a worker and install a machine in his or her place. The task to be performed would not change, only the agent performing it. The ultimate expression of this idea was the robot—a one-for-one replacement for the factory worker. But automation has seldom gone this way.

Consider the manufacture of electronic devices. At the outset, this was a labor-intensive process of placing components on a chassis, stringing wires between them and soldering the connections one by one. Attempts to build automatic equipment to perform the same operations proved impractical. Instead, the underlying technology was changed by introducing printed circuit boards, with all the connections laid out in advance. Eventually, machines were developed for automatically placing the parts on the boards and for soldering the connections all at once.

The further evolution of this process takes us to the integrated circuit, a technology that was automated from birth. The manufacture of microprocessor chips could not possibly be carried out as a handicraft business; no sharp-eyed artisan could draw the minuscule circuit patterns on silicon wafers. For many other businesses as well, manual methods are simply unthinkable. Google could not operate by hiring thousands of clerks to read Web pages and type out the answers to queries.

The automation of factories has gone very much according to the script written by Diebold and other early advocates. Computer control is all but universal. Whole sections of automobile assembly plants are now walled off to exclude all workers. A computer screen and a keyboard are the main interface to most factory equipment.

Meanwhile, though, manufacturing as a whole has become a smaller part of the U.S. economy—12 percent of gross domestic product in 2005, down from more than double that in the 1950s. And because of the very success of industrial automation, employment on production lines has fallen even faster than the share of GDP. Thus, for most Americans, the factory automation that was so much the focus of early commentary is all but invisible. Few of us ever get a chance to see it at work.

But automation and computer technology have infiltrated other areas of the economy and daily life—office work, logistics, commerce, finance, household tasks. When you look for the impact of computers on society, barcodes are probably more important than machine tools.

comments powered by Disqus


Subscribe to American Scientist