Talking about Machines
Julian E. Orr
· published 1996
This is technical anthropology, and it is thorough and much better than you'd expect from a potentially dry dissertation
from 1990, described as an “underground classic among ethnographers of work”. The central thesis of the book is that
(technical) knowledge is a socially distributed resource that is transported by way of oral culture and that
debugging works by constructing a narrative between technician, customer and machine. More details below, but this
is both a central insight, well described and documented, and applicable to most areas of (work) life.
Orr was shadowing Xerox technicians at their work of servicing and repairing photocopiers (helpfully, Orr had learnt
this job, on slightly different machines, too, so he wasn't just an annoying nosy interloper). He initially describes
their work as a “continuous highly skilled improvisation within a triangular relationship of technician, customer, and
machine”, which very clearly demonstrates the respect and depth Orr displays throughout the book. He was only
accompanying them during their work hours, there were no additional structured interviews.
Debugging as narrative
In Orr's experience, a diagnosis involves the creation of a coherent narrative explaining the problem, starting from
available pieces of unintegrated information that is transformed into a coherent narrative. This narrative is both used
to fix the problem, and to share the whole story of problem, diagnosis and fix with colleagues (see below).
A significant part of the job is social. Extracting information from non-technical users, discovering their
understanding, and then reflecting the technician's knowledge in a way that will shape the users' behaviour is tricky,
and also always fraught with office politics, history, colleague's opinions and so on. They are very aware themselves
that users are a large part of the equation – they even have a saying: “Don't fix the machine; fix the customer!”
(Side note: I'm shorthanding Orr's “diagnosis and repair” to “debugging” here.)
Once a story has been formed by means of diagnosis, and performed by means of repair, it can be shared with the
professional community. The tellers use it to build their identity (either of competence, when in the starring role, or
as knowledgeable part of the community, when repeating somebody else's story). Stories are told in a very concise
manner, as repeating obvious story elements would be superflous and mildly insulting to the listeners. It's assumed that
all competent technicians will tell stories, both to display status, and to instruct.
This comes with a certain civil competitiveness that manifests in one-upmanship in their stories about hard fixes, or
challenges when the narrator only tells the problem, and asks the audience to provide diagnosis. This kind of
one-upmanship, or challenge-response jokes, are only used in private, never in front of customers, where technicians
present in a carefully controlled manner, and only joke at their own expense.
This social cohesion also shows up in the interesting way that technicians are more worried about the social damage to
another technician, when they have to take a job in their territory, than about the machine they are servicing, because
they know full well that social relationships are harder to repair than machines.
Knowledge as distributed resource
Orr observes that the telling of debugging narratives both demonstrates and shares the technicians' mastery, creating
and celebrating their identities (in this case: as masters of the black arts of dealing with machines and customers,
both). They provide information about technical matters – past fixes, emerging problems, warnings of incorrect
documentation –, but also social and administrative topics, like changes among customers, area distributions, and
These stories, usually shared during meals, are necessary for the technicians to remain capable and perform their work.
The official documentation is not always correct, and rarely sufficient. Well-known workarounds int he fields (starting
with valid substitutes for individual parts and ending with MacGuyver style solutions to common problems) are only
On the nature of work
It's interesting to see notes from a time when service work was still coming into its own. The remark that “Leftists
routinely orient themselves to service work by assuming that service workers are a new proletariat even though most
service workers apparently don't see themselves this way.” has aged well.
This time also reflects that the technicians' status straddled both worlds: They were skilled workers dressed like
office workers, ties and all, their tool boxes made to look like briefcases, while also performing dirty work, setting
them apart from pure office workers.
Orr also pointedly observes that this shift has also shifted the meaning of “to work” to have an added, possibly even
primary, meaning of “to be employed” rather than “to do a thing”.
Defining the job
Another still very relevant comment is that the definition of work seems to be up to management, who focus on the
activities required to deliver a product, and disregard other equally necessary activities, that need then to be snuck
in by people on the ground. For example, there is no alotted time for services technicians to casually visit the
clients, inquire as to the state of everything, provide quick checks and education. But these encounters are vital, and
technicians take care to get them done despite their tenuous status as not-really-work.
Orr also mentions that all the common management definitions of work are focused on individual workers, while the fact
that work is commonly done by a group and an emergent professional community is rarely acknowledged. In this example,
all work would be much, much worse or even impossible without the frequent shared lunches and exchanges of experience.
However, no management would have classed those things as work.
But that's not the only failure of management! They also tend to see technicians not as skilled workers (as they see
themselves), but as easily replaced by semiskilled labour with a good set of instructions. This state is obviously
advantageous to the company, and they try to produce and maintain it by providing worse documentation, dumbed down to a
level that does not permit complex troubleshooting to emerge. So the second a problem diverges from the manual, it's
again down to skill and shared oral tradition. Because intentional de-skilling cannot deprive workers of skills they
have, it only reduces the amount of information they receive.
When management attempts to de-skill and break up a community like that, it will resist in several ways. It may switch
to "work-to-order" mode to show management what it would miss out on (in this case: only service machines according to
the manual, not developing further fixes, and not maintaining customer relations). They may also devalue promotions
heavily, preferring to remain a service technician rather than being promoted into in-house service or management.
"This resistance can surprise employers who think of labor as a commodity that can be arranged to suit their ends."
I liked the apt description of the relationship between customers and technicians as the “contract between the
customers' need and the technicians' ability”, and he goes on to describe the relationship: “The principal issues for
the technicians in this triangular interaction are control and understanding, and one reward for achieving the two is
their own identity as competent technicians.”
Equally, I liked the observation that technicians have trouble reasoning about details of the machine when they can't
see it in front of them. They can still figure out the larger scale problems, but they don't remember things like the
location of individual components in the machine. They also sometimes leave the technical field and get promoted to
be managers, but few succeed there, most return to being technicians: “Most technicians would rather remain a
technician-hero than become an organization manager”.
“The division between technician and senior technician is just one of seniority and salary; reputation and skill are not
thought to be related to this distinction.” – things never change. And, how Marxian: “The occupational community shares
few cultural values with the corporation; technicians from all over the country are much more alike than a technician
and a salesperson from the same district.”
“What really holds their interest is a situation they do not understand.” - haha, nerds.
Machines are very much not all the same: technicians talk about the ones in their territories as individuals, with their
specific histories, users, problems, anticipated problems, past fixes requiring special attention, and so on. And this
kind of tricky detail is what really stood out to me: there is infinite complexity in detail, and you don't know the
dimensions of complexity involved in any job until you have worked it or have had trusted conversations with people who
This book is part of the 2022 Backlog Incident.