The Soul of Maintaining a New Machine - Third Draft | Books in Progress

TTHEY ATE TOGETHER every chance they could. They had to. The enormous photocopiers they were responsible for maintaining were so complex, temperamental, and variable between models and upgrades that it was difficult to keep the machines functioning without frequent conversations with their peers about the ever-shifting nuances of repair and care. The core of their operational knowledge was social. That’s the subject of this chapter.It was the mid-1980s. They were the technician teams charged with servicing the Xerox machines that suddenly were providing all of America’s offices with vast quantities of photocopies and frustration. The machines were so large, noisy, and busy that most offices kept them in a separate room. An inquisitive anthropologist discovered that what the technicians did all day with those machines was grotesquely different from what Xerox corporation thought they did, and the divergence was hampering the company unnecessarily. The saga that followed his revelation is worth recounting in detail because of what it shows about the ingenuity of professional maintainers at work in a high-ambiguity environment, the harm caused by an institutionalized wrong theory of their work, and the invincible power of an institutionalized wrong theory to resist change.The cover of Julian Orr’s influential Talking About Machines shows technicians working on the Xerox 5090 photocopier, introduced in 1990. Though they were doing messy blue-collar work, Xerox required the technicians to act and dress white-collar. They carried their tools in a briefcase.SourceThe anthropologist was Julian Orr. In 1979, he was hired by Xerox PARC (Palo Alto Research Center) in northern California to provide technical support for two machines being developed there—the Alto computer and a color laser printer. By 1984 he had migrated to studying Xerox service technicians, encouraged by John Seely Brown, then a lab manager, later director of PARC. Orr’s research culminated in a remarkable book titled Talking About Machines: An Ethnography of a Modern Job, published in 1996.

His book shows that the most baffling problems the technicians faced in their machines were solved by discussion, and the most instructive element in their conversation was what Orr calls “war stories”—narratives the technicians told each other about how they worked through a bewildering problem in a machine to arrive at a satisfying solution.The stories also establish the teller’s contribution to the local community of technicians. Orr writes:Given that the only status within the community is that of competent practitioner, fame can only be based on a reputation for extraordinarily competent practice, the ability to solve newer and harder problems. Since technicians normally work alone, achievements will only be known if the person responsible tells them. Moreover, technicians want the information to circulate, so that others can address similar problems. A team shares responsibility for its calls, so there is incentive to have all members competent for as many problems as possible.1Often, the issue was not with the copier but with unintentionally destructive behavior by the users. That, too, was considered fixable. Orr declares that the technicians’practice is a continuous, highly skilled improvisation within a triangular relationship of technician, customer, and machine…. Narrative forms a primary element of this practice. The actual process of diagnosis involves the creation of a coherent account of the troubled state of the machine from available pieces of unintegrated information…. A coherent diagnostic narrative constitutes a technician’s mastery of the problematic situation.Narrative preserves such diagnoses as they are told to colleagues…. The circulation of stories among the community of technicians is the principal means by which the technicians stay informed of the developing subtleties of machine behavior in the field.2By the mid-1980s, Xerox copiers had reached such a degree of complexity that “individual machines,” Orr writes, “are quite idiosyncratic, new failure modes appear continuously, and rote procedure cannot address unknown problems.”3A Xerox 9400 with its panel open for technicians to get at the guts of the machine.SourceSpecifying the exact features for a copying job on the 9400’s control panel was not for the faint of heart. Users varied in their sophistication. When the machine had a problem, some customers could communicate helpfully with the repair technicians. Many could not. Some were, in fact, the cause of the problem.

SourceJane Fonda’s first-day-on-the-job character is overwhelmed by a Xerox 9400 belching copies at her in the feminist comedy 9 to 5, made in 1980, co-starring Dolly Parton and Lily Tomlin. Her humiliating defeat by the copy machine was a common experience for office workers in those years.SourceFor example, the famous Xerox 9400 that was introduced in 1977 weighed one and a half tons and took up floor space measuring nine by fifteen feet. It cost $85,000 ($430,000 in 2024). Automatically feeding up to 3,000 sheets of paper, it made copies at a rate of two per second and collated them into 50 separate bins. It could read and write two-sided sheets, and the image size was adjustable. Every stage of the process required extreme precision—from imaging to paper handling to managing the sequence of electric fields that transferred the image-bearing toner onto the paper, then pressing and baking the toner into the paper, and cleaning everything to be ready for the next image a half-second later. A fault anywhere in that sequence or in the control system could cause degraded copies or take down the whole machine.A technician who serviced 9400s in the 1980s recalls:This machine was the main method of information distribution for the entire Federal Government for quite a while… I took service calls on these machines coming from the Pentagon, the CIA, the State Department, the Defense Mapping Agency, the Nuclear Regulatory Agency, Pax River Naval Research Center, the National Institutes of Health, the Uniformed Services University of the Health Sciences, and the National Bureau of Standards…. Xerox must have made an unbelievably enormous amount of money from that product—and spent an almost equally unbelievable amount maintaining them, because I also remember the nearly endless field retrofits we had to perform to keep them running…. Oh, those were the days!4The technicians were organized into regional teams, servicing all the machines within a geographic area. Each team member had sole responsibility for a group of customer offices and machines. Orr emphasizes that all their attention was focused on the work, with little to spare for the corporation that paid them.

Indeed, they “shared few cultural values” with the rest of Xerox and did not seek to rise within it.5 They relished working within the technician-customer-machine triangle, where their competence was tested and rewarded daily.Since half of the problems they had to fix were caused by misuse of the machines, they had an adage: “Don’t fix the machine, fix the customer.” The copiers were so sensitive that users could screw things up by using toner from a different machine or a cheap knock-off supplier. Or they could mistakenly put paper in the feeder tray curl-side-up instead of curl-side-down (it was vice-versa in other copiers).In 1983, PARC intern Lucy Suchman conducted a famously diabolical experiment with the supposedly user-friendly Xerox 8200 copier. She made videos of PARC people using it. Tech historian John Tinnell recounts what happened when PARC director John Seely Brown (who was a long-time motorcycle buddy of Xerox CEO Paul Allaire) took one of the videos to corporate headquarters: “[Brown] played the video for a room of Xerox managers. It featured two men struggling and failing to produce a copy of an article [Ronald] Kaplan had written. Everyone back at PARC knew [that] Kaplan was a world-renowned computational linguist, and [Allan] Newell was revered as a founding father of AI. The video showed Newell peering over Kaplan’s shoulder, oscillating between curiosity and confusion as the pair exhausted the full arsenal of their joint expertise, to no avail, for over an hour and a half. One of the Rochester men slammed his fist on the table and hollered, ‘Goddammit, Brown! What did you do, get those guys off the goddamn loading dock?’ Brown looked at him and smiled, ‘Well, actually, let me introduce these two stooges to you,’ He revealed Kaplan’s and Newell’s pedigrees, and the Xerox execs sat speechless.”9Lucy Suchman later became the manager of the “Work Practice and Technology” research group at PARC, which was composed of all the anthropologists at PARC plus a few computer scientists. Julian Orr was a member from the beginning.

Suchman’s 1987 book, Plans and Situated Actions: The Problem of Human-Machine Communication, became a classic text.SourceThe machines were so complex that even sophisticated customers could lose their way—for example by failing to replace a baffle after clearing a paper jam (which would affect the paper’s temperature and cause further jams) or rashly re-using paper that had been through the machine once and was, therefore, oily (which would make rollers so oily that they could no longer feed properly).Even when the problem was purely mechanical, the customer was a primary source of important diagnostic information about when and how the breakdown had occurred. That meant, says Orr, that “the customer must be initiated into the technicians’ community of discourse,”6 complete with an understanding of how the machine worked, how to recognize the noises it made at the various stages of copying, and the correct language to describe its many failure modes. Some customers resisted learning any such thing, and the technicians had to find a way to jolly them into learning it anyway. Orr writes that users weretaught by the technicians how to talk about the machine. They know what to observe—the state of the originals, where the machine leaves paper when it stops, and where in the cycle trouble occurs—and they know most of the terms to describe these phenomena.7As a consequence, the technicians became protective of the nuanced social relationship they built with each customer. Orr notes: “Technicians worry more about the social damage another technician can do in their territory than about what might happen with the machine, perhaps because the machine would be easier to repair than the delicate social equilibrium.8”Ethnographer Orr had a sharp eye for detail. He noticed when a technician on a call began by examining copies that had been thrown in the trash and deduced from them that the problem with the machine was different from what the customer had reported. “The trashcan is a filter between good copies and bad,” one technician explained “Just go to the trashcan to find the bad copies and then… interpret what connects them all.”10Another time, Orr observed a technician joking with a customer about the need to keep any engineers in the building away from a machine that needed fixing, because “engineers believe they have a right to fiddle with any machine they encounter, whether they know anything about it or not.