The politics of programming

control board

For the last post of the year, I give you a short excerpt from The Glass Cage.

It’s commonly assumed that any technology that comes to be broadly adopted in a field, and hence gains momentum, must be the best one for the job. Progress, in this view, is a quasi-Darwinian process. Many different technologies are invented, they compete for users and buyers, and after a period of rigorous testing and comparison the marketplace chooses the best of the bunch. Only the fittest tools survive. Society can thus be confident that the technologies it employs are the optimum ones — and that the alternatives discarded along the way were flawed in some fatal way.

It’s a reassuring view of progress, founded on, in the words of the late historian David Noble, “a simple faith in objective science, economic rationality, and the market.” But as Noble went on to explain in his 1984 book Forces of Production, it’s a distorted view: “It portrays technological development as an autonomous and neutral technical process, on the one hand, and a coldly rational and self-regulating process, on the other, neither of which accounts for people, power, institutions, competing values, or different dreams.” In place of the complexities, vagaries, and intrigues of history, the prevailing view of technological progress presents us with a simplistic, retrospective fantasy.

Noble illustrated the tangled way technologies actually gain acceptance and momentum through the story of the automation of the machine tool industry in the years after World War II. Inventors and engineers developed several different techniques for programming lathes, drill presses, and other factory tools, and each of the control methods had advantages and disadvantages. One of the simplest and most ingenious of the systems, called Specialmatic, was invented by a Princeton-trained engineer named Felix P. Caruthers and marketed by a small New York company called Automation Specialties. Using an array of keys and dials to encode and control the workings of a machine, Specialmatic put the power of programming into the hands of skilled machinists on the factory floor. A machine operator, explained Noble, “could set and adjust feeds and speeds, relying upon accumulated experience with the sights, sounds, and smells of metal cutting.”

In addition to bringing the tacit know-how of the experienced craftsman into the automated system, Specialmatic had an economic advantage: a manufacturer did not have to pay a squad of engineers and consultants to program its equipment. Caruthers’s technology earned accolades from American Machinist magazine, which noted that Specialmatic “is designed to permit complete set-up and programming at the machine.” It would allow the machinist to gain the efficiency benefits of automation while retaining “full control of his machine throughout its entire machining cycle.”

But Specialmatic never gained a foothold in the market. While Caruthers was working on his invention, the U.S. Air Force was plowing money into a research program, conducted by an MIT team with longstanding ties to the military, to develop “numerical control,” a digital coding technique that was a forerunner of modern software programming. Not only did numerical control enjoy the benefits of a generous government subsidy and a prestigious academic pedigree; it appealed to business owners and managers who, faced with unremitting labor tensions after the war, yearned to gain more control over the operation of machinery in order to undercut the power of workers and their unions. Numerical control also had the glow of a cutting-edge technology — it was carried along by the burgeoning postwar excitement over digital computers. The MIT system may have been, as the author of a Society of Manufacturing Engineers paper would later write, “a complicated, expensive monstrosity,” but industrial giants like GE and Westinghouse rushed to embrace the technology, never giving alternatives like Specialmatic a chance.

Far from winning a tough evolutionary battle for survival, numerical control was declared the victor before competition even began. Programming took precedence over people, establishing a misanthropic design philosophy that now dominates our thinking about automation. As for the general public, it never knew that a choice had been made.

3 thoughts on “The politics of programming

  1. Linux Guru

    “Any sufficiently advanced act of benevolence is indistinguishable from malevolence. ”
    — Charles Rubin (referring to artificial intelligence)

  2. Roger

    He is right, it is a naive view of technological evolution and progress. And yes, most naive views of complex processes are wrong.

    The more interesting observation is that despite the politics, path dependencies, conflicting interests and values that we still somehow sometimes see technological, scientific and economic progress. This is what requires explaining.

Comments are closed.