T H E  N E T W O R K  O B S E R V E R

  VOLUME 1, NUMBER 9                                SEPTEMBER 1994


  This month: The epidemiology of Unix
              Imagining the future of labor markets
              The infrastructure of public relations
              Books about writing
              E-mail addresses for gay activists


  Welcome to TNO 1(9).

  This month's issue includes a couple of articles by the editor.
  First is a commentary upon a savagely humorous new book based
  on the UNIX-HATERS mailing list, leading to a discussion of how
  unfortunate technical standards can become entrenched.  Unix is
  merely a nuisance mess, but many other standards are downright
  dangerous, and an understanding of the social dynamics of
  standards might avoid further dangers in the future.

  Then comes a deliberately frightening scenario about the future
  of information-intensive labor markets.  As computers decrease
  the transaction costs associated with hiring, and as networks
  increase the geographic scope over which firms can hunt for
  people to do specific units of work, human labor may become
  radically commodified through standardization.  What's your
  cheerfulness rating?  The point of this exercise is not concrete
  prognostication, but rather restretching our imaginations.  Big
  Brother provides one dystopian vision of institutional abuses of
  computing, but many other scenarios are possible.


  The epidemiology of Unix.

  According to The UNIX-HATERS Handbook (a funny but serious new
  book from IDG Books in San Mateo, California, edited by Simson
  Garfinkel, Daniel Weise, and Steven Strassman):

    Unix is a computer virus with a user interface.

  Maybe you have to be a computer person to get the humor in this,
  but I personally find it all too funny.  UNIX-HATERS is a mailing
  list begun in 1987 by an MIT Media Lab graduate student named
  Mike Travers, so that people worldwide who spend far too much
  time struggling with the Unix operating system could compare
  notes and engage in rude, exasperated flaming.  Over time the
  UNIX-HATERS list distributed a remarkably large number of very
  nasty, very well-informed messages, which three members of the
  list have now included in a book.  Although they are listed as
  editors, in fact they wrote a great deal of additional text,
  within which the UNIX-HATERS messages are for the most part
  simply illustrations.  (They even included one small message of
  mine, about one version of the Unix "rm" command.)

  Unix is not literally a computer virus, of course, but the
  analogy is striking.  Unix was written by a couple of guys who
  needed a simple operating system to enable them to do something
  else.  From there it grew and spread exponentially, gathering
  a bewildering variety of additional hacks and variant versions
  which have successfully resisted numerous attempts to impose
  order and standards upon them.  Why?  One reason is that anyone
  could get it basically for free -- including its source code.
  But the more fundamental reason is that Unix filled a niche --
  lacking a standard operating system for "workstation" computers,
  everyone found it easier to incrementally adapt Unix to their
  purposes than to buy something else or write it from scratch.
  The history of Unix, then, is not a matter of conscious design
  and planning but of ceaseless mutation and multiplication.
  This is the sense in which Unix is like a virus.

  I was utterly absorbed the UNIX-HATERS book.  When one of the
  editors, Simson Garfinkel, sent me a draft, I sat right down
  and read it pretty much straight through, marking typos and
  accumulating a batch of written comments.  I had better things
  to be doing, to be honest, but there was something compelling
  about the ceaseless barrage, page after page, of outrageous
  stories about Unix functionality, including dozens of instances
  of bad design that had bitten me over the years.  The book is
  written in an exaggeratedly nasty style which actually somehow
  works, perhaps because the undead Unix is the product of no
  single villain.

  The scope of The UNIX-HATERS Handbook is almost exclusively
  technical.  These are computer people whose sense of good design
  has been offended, and they are exacting revenge in the name
  of much better operating systems that have been relegated to
  the software museum by the haphazard spread of viral standards.
  And indeed, for sheer harmless moral outrage, the book is hard
  to beat.

  My own questions about Unix, though, are more institutional.
  If Unix is a virus, surely we need an epidemiology.  Just as
  the spread of human viruses is abetted through the mechanics
  of air ducts, the civil engineering of sanitation, the culture
  of sexuality, and the economics of pharmaceuticals, likewise
  the spread of technical standards is influenced by a variety
  of non-technical factors relating to the technology's place in
  the larger world.  Why did Unix spread?  One important reason,
  at least in the early days, was the many separate choices of
  individual programmers who just needed to get individual systems
  running -- Unix was quick and easy if you knew how everything
  worked and just needed to make forward progress.

  In the long run, though, a more important factor was the decision
  of companies like Sun Microsystems to aggressively promote Unix
  as a standard.  When a new market niche opens, like the emerging
  niche for workstation computers that Sun nearly dominated for
  a long time, quality doesn't matter nearly as much as speed --
  how fast you can get products to market.  Sun could take Unix
  off the shelf, promote it as a standard, cut prices like mad, and
  promise to fix the problems later.  In economic terms this was an
  entirely rational decision.  Part of the reason for this, though,
  was that very many of the people who bought Suns were not aware
  of the high costs of Unix system maintenance which ate up much of
  the savings on the hardware.  This includes the hidden costs of
  the legions of frustrated system users and programmers trying to
  overcome the numerous manifestations of Unix's bad design -- or
  more precisely, its lack of design.  Many of these hidden costs
  are hidden because their victims simply weren't aware that the
  world could be organized in any other way.

  Unix, of course, is simply an extreme example of something that
  happens a lot with technical standards.  Take the familiar QWERTY
  keyboard, which is well known to be ergonomically inferior to
  other designs but which persists because it's next to impossible
  to get millions of people sufficiently coordinated to make the
  shift.  Given that the world is full of QWERTY keyboards, someone
  deciding to learn to type is best-advised to learn QWERTY typing.
  Once they've learned QWERTY typing, it's hard to convince anyone
  that it's worth learning Dvorak typing, or any other optimized
  typing scheme.  Keyboard manufacturers, for their parts, are
  facing a world in which most typists know the QWERTY system and
  not any other.  Thus the QWERTY standard remains entrenched just
  because it got established and keeps reproducing itself.  With
  Unix, of course, the situation is much worse because of the
  ceaseless mutations it undergoes.  But the principle is similar.

  So who do we arrest for causing the Unix epidemic?  How about
  the executives at Sun and other similar companies?  Did they
  have reasonable alternatives?  Probably not.  The lesson, I
  think, is to be extremely aware of the rise of new standards.
  If we want things done right, we need to make noise quickly,
  at the beginning of the cycle, before something random becomes
  entrenched.  The consequences can be severe.  With Unix it's
  just a matter of cost and inconvenience.  But with other types
  of systems it's also a matter of civil liberties.  Consider the
  vast number of systems that collect personal information indexed
  by a personal identifier such as a Social Security Number.  Did
  those systems really have to work that way?  Could they have
  used cryptography to protect individuals from abuse of their
  information, and to give individuals more control over the
  ways in which their information was used?  Yes, most likely
  they could.  So why didn't they?  One answer is profit, since
  secondary uses of personal information so often bring in good

  But another answer is that the systems were designed that way
  by default -- since that kind of plain, explicit representation
  is what programmers are taught in school.  We'll probably always
  need legislation to make sure that information is not abused.
  But perhaps a better, longer-term answer is to change education
  in computer science so that privacy-protection methods such as
  cryptography are taught in the earliest classes and integrated
  into the design methodologies of systems analysis -- rather than
  being stuck on in non-standard ways afterward.


  Science fiction and social choice.

  Today our society must make, very rapidly, a large number
  of difficult choices about technology.  Distributed computing
  technology is changing with particular speed right now, and the
  consequences of its diverse uses are finally starting to be felt
  on a large scale.  Technical standards and market structures are
  nearly impossible to roll back once they become entrenched, so it
  is important to try to imagine ahead of time what the future will
  bring.  This is not just -- or even mostly -- a technical matter.
  It is more a matter of social imagination: what kinds of social
  and institutional dynamics will emerging technologies enter into?

  Many technical people of my acquaintance rebut this question with
  the lame answer that such things are impossible to anticipate,
  so why try?  One "why" is that it matters, and another is
  that, colorful counterexamples notwithstanding, we actually can
  anticipate a great many things reliably enough to make important
  choices now.  This is the zone where science fiction intersects
  with sociology.  (Why does "science fiction" connote stories set
  in the future or an irretrievable past, whereas "sociological
  fiction" connotes either nothing at all or stories set in the
  present day?)  Let us call it the "technosocial imagination" and
  think hard about it, because we're going to need it.

  The technosocial imagination, like all kinds of imagination,
  is profoundly influenced by the past.  In particular, it is
  deeply influenced by precedents and by metaphors.  It is a
  notorious fact that ideas about the future, when investigated
  in retrospect, invariably turn out to have been ways of working
  through situations in the present.  This isn't so bad, but it's
  unfortunate when it distorts choices about technology that really
  do matter for something.  We can defend ourselves against this
  effect, at least to some degree, by using the tools of critical
  analysis to make explicit the structure of our ideas about these
  things, thereby making possible the conscious investigation of

  When it comes to distributed computer technology, perhaps the
  single most important precedent and metaphor conditioning our
  technosocial imagination is "Big Brother" -- George Orwell's
  fictional abstraction of the totalitarian societies of the mid
  twentieth century, drawing on historical experiences of the
  secret police going back a long ways.  The idea of computers
  maintaining extensive databases about your purchases and
  finances, your work activities, and your personal life certainly
  sounds, on first blush, like Big Brother at work.

  But Big Brother has his limitations as a structuring metaphor for
  the technosocial imagination.  In TNO 1(7) I explained one reason
  why this is so, namely that computers don't work by looking at
  things.  They work by languages, and applied computing isn't just
  a matter of watching you but of restructuring your activities
  (either obviously or subtly over time) so that they afford
  "parsing" in terms that given computer systems can represent
  to themselves.  Another problem with Big Brother is that it is
  a metaphor of state action and thus has limited applicability to
  analyses of the market.  The secret police really does exist and
  I wouldn't want to minimize it.  But it's not what's driving the
  social structuring of distributed computer technology right now.

  Here I would like to start developing some alternative forms of
  technosocial imagination to encourage reasoning about distributed
  computer technology.  To speak of "forms of imagination" rather
  than "theories" is to announce a cultural intervention rather
  than an academic exercise, though these two activities clearly
  overlap.  More concretely, to speak of "forms of imagination" is
  to beg forgiveness for extrapolation, exaggeration, abstraction,
  and simplification.  Metaphors are blunt instruments: they order
  some elements of reality in simple and compelling ways while
  pushing all the other elements to the margins.  Thoughtful people
  apply them with care, heuristically, as a stimulus to critical
  reflection and a generator of hypotheses and questions that might
  be addressed to material reality, both through formal experiment
  and through the lived experiences of the people to whom the stuff
  is happening.

  That said, I want to extrapolate some things that are going on
  in the labor market right now.  It's clear that a great deal is
  going on in labor markets, but people disagree about *what* is
  going on.  Temporary work is found in areas where it formerly was
  not.  New kinds of piecework are called "consulting".  Contract
  manufacturing has become more prominent in the global division of
  labor as low-skill (and, increasingly, high-skill) manufacturing
  has drifted from formerly industrial countries to formerly
  nonindustrial ones.  It is not an extreme exaggeration to say,
  as did the cover story of a recent issue of Fortune (9/19/94),
  that we're looking at the "end of the job".  What I think is
  safe to say is that we're looking at an increasing heterogeneity
  of contractual relationships in the workplace, both in the
  relationships between organizations (e.g., partnerships) and
  between organizations and individuals.

  Computer networks alone did not cause these developments, and it
  is hard to determine precisely what contribution they have made.
  But in some sectors, as TNO 1(8) pointed out, "human resources"
  has encountered informational economies of scale that have made
  it possible for companies like McDonald's to contract much of
  their labor needs (finding, hiring, paying, and firing employees)
  to outside firms.  These firms employ extensive testing and
  try to "match" people to jobs in a rationalized way on a large

  Where could this go?  Business management literature in the
  last decade has placed increasing emphasis on the role of
  ideological control in workplaces.  The idea is that, in order
  to "empower" people to make decisions based on local knowledge,
  it is necessary to indoctrinate them (they often use these words,
  ideological and indoctrinate) with the company's "philosophy"
  and "values" and "culture".  Many managers now say things like
  "we hire people who share our values" -- that means that they
  employ some kind of personality testing (either through formal
  tests or more informal kinds of screening) to hire people who
  fit a certain personality profile, for example "cheerful", "work
  ethic", "team player", and so forth.

  Imagine a system in which such tests become standardized on a
  large scale.  If you want a job anywhere in the bottom four
  fifths of the economy (or, in a different way, in the top three
  percent) then you'll regularly take a batch of personality tests
  and your scores will be stored in databases.  Employers, using
  statistical evidence gathered through pervasive work monitoring
  to correlate personality factors to job performance, contract for
  labor using this database.  When this system becomes large enough
  it will move beyond the simple idea of a "match" between the job
  and the person.  Instead, different personality traits will have
  market values.  In such a world, one might reasonably expect, a
  high "cheerfulness" rating will be a fairly valuable commodity
  that will fetch a premium on the market, with a high score for
  "sullen efficiency" trading at a modest discount.

  A functioning market in human personality traits will go a long
  way to maximizing the efficiency of a wide variety of industries
  -- and not just service work.  The people who design jobs and
  the machines that hire people to fill them will be able to make
  economic calculations: one kind of restaurant, for example, might
  require its employees to be rapid learners and good at managing
  conflict, where another might require them to be docile and
  susceptible to being pressured, and the relative labor costs
  associated with these attributes might be factored into choices
  about what kind of restaurant to open where.

  In such a world, a temporary agency will operate very much
  like a commodity market like the ones that sell pork bellies
  in Chicago.  No doubt it will become possible to trade futures
  in human labor, for example futures in solemnity or aggression
  whose values might be affected by the introduction of new
  psychopharmaceuticals.  Large companies will be able to purchase
  long-term contracts to lock in their labor prices, but a spot
  market will remain for those purchasers whose sustained needs are

  In such a market, people will be able to make better economic
  decisions about altering their personalities.  Investments in
  psychotherapy will be much better quantified, and "retreats"
  and "workshops" that promise "transformational" changes in one's
  basic philosophy and values, already a substantial industry, will
  begin to advertise much more openly and lose much of their (often
  well deserved) reputations as cults.  It'll be like test cramming
  courses only much more so: "my cheerfulness on the RDPM scale is
  now in the 98th percentile, and I owe it all to Lifespring!".

  The precise structure of the new rationalized labor market will
  be affected by many things -- or, I should say more precisely, it
  will interact with many other factors.  Among these is the extent
  to which jobs -- actually, the newly fashionable phrase is "work
  that needs to be done" -- can be broken down into easily learned
  units that might be assigned to what we now call "temporary"
  workers.  In past times, and indeed for the most part in the
  present day, this end has been achieved principally through the
  fragmentation of jobs into time-and-motion-measured units.  But
  it can be achieved through a combination of many other factors of
  well: increasing standardization of the skills, standardization
  of the ideologies of work (this effect was formerly achieved by
  professions, but professions used to have a certain amount of
  autonomy and solidarity that they are rapidly losing), increases in
  the potential hiring pool for each job by permitting work to be done
  remotely via computer networks and the like, and work monitoring
  through computers and otherwise.

  What will it be like to live in such a world?  It is hard to
  say for certain.  It depends in large part on whether, and in
  what ways, and to what depths, people actually believe in the
  particular ideologies that are in demand in the market.  If they
  really do believe these things then they will be happy until
  they are driven into the ground by the stresses of overwork and
  continual changes of jobs.  I expect, though, that most people
  will be in a middle ground, neither believing the stuff nor not
  believing it, but rather driven by the market to learn to animate
  its outer forms without having much time to critically examine
  any of it.

  This would all be sad.  Will it happen?  I can point at quite
  a few signs and rumors that it's on its way, but I really don't
  know.  The point of the exercise is not to make a prediction
  that can be scored correct or incorrect someday, but rather to
  shake up our technosocial imaginations.  This unhappy scenario
  is far from Big Brother but just as unpleasant.  If it doesn't
  have quite the same bite, that's because quite a few people have
  full-time jobs justifying such things by arguing against abundant
  evidence that the unfettered market necessarily provides the best
  of all possible worlds.  The motto of this newsletter is that
  the market is like the police: of course you need it, but if it
  becomes the central organizing principle of your culture then
  you're in deep trouble.  And that's my conclusion here: computers
  are cultural artifacts as much as they are technical ones, and
  they lend themselves to cultural engineering just as much as they
  do to technical engineering.  That's not a necessary connection,
  since computers can probably be used in different ways in
  different worlds, but it's a wholly plausible connection in the
  world that is emerging before our eyes today.

  Is it what we want?  We still have time to decide.


  This month's recommendations.

  Andrew L. Friedman, Computer Systems Development: History,
  Organization and Implementation, Chichester, UK: Wiley, 1989.
  This is a detailed, utterly astute history of computer systems
  development, based on a sequence of three stages: building
  programs that satisfy specs, developing specs that correspond to
  structures of work, and designing user interfaces that people can
  use.  He predicts the emergence of a fourth stage: concern with
  interactions between structures of computing and structures of

  William Strunk Jr and E. B. White, The Elements of Style, third
  edition, New York: Macmillan 1979.  I've read a lot of books
  about how to write, but this is the book I keep coming back to.
  It contains a simple, clear ideology of language based on the
  slogan "omit needless words".  Just reading it won't make you a
  great writer.  But reading it at bedtime on writing days can work
  wonders -- once you set to writing again the next day, you'll
  start to notice the truth of its simple rules, whereupon you
  can look for your own personal way to apply the spirit of those
  rules.  Far from constraining your writing to artificial norms,
  you'll probably find that this process helps you bring out your
  own voice.  Even better, try picking just one of the rules (for
  example, its injunction to use active verbs) and go through every
  sentence of your current draft.  If you're like me (and many of
  my students), you'll be amazed at the radical changes that this
  simple procedure can bring to your prose.

  Natalie Goldberg, Writing Down the Bones: Freeing the Writer
  Within, Boston: Shambhala, 1986.  Another manual for writers,
  as different from Strunk and White as it could possibly be, but
  just as useful.  Goldberg is a Buddhist writing instructor in
  New Mexico whose speciality is a very personal kind of fiction
  writing, but whose methods can be extended to all other kinds
  of writing as well.  She suggests stirring up your writerly
  self by engaging in a kind of automatic writing that circumvents
  your efforts to make your writing "good", fashionable, according
  to the rules, and so forth.  She also offers plenty of useful
  general-purpose advice about writing that should make valuable
  bedtime reading for anyone who is, say, stuck on a dissertation.


  Company of the month.

  This month's company is:

  PR Newswire
  1515 Broadway, 32nd Floor
  New York, New York  10036

  +1 (212) 832-9400, fax (800) 793-9313

  Producing a newspaper or news program costs money.  One way
  to save money is to use predigested news "stories" from PR
  agencies.  Oscar Gandy calls this an "information subsidy".
  This practice, while commonplace since early in this century,
  has grown ever more pervasive, to the degree that a whole
  elaborate infrastructure has grown up to implement it.  Perhaps
  the single most important element in that infrastructure is PR
  Newswire, which allows organizations engaged in public relations
  to have their materials delivered to news organizations
  worldwide.  A customer can choose from a variety of "newslines",
  corresponding to a given geographic region, industry, ethnic
  group, or whatever, including both mainstream newspapers and more
  specialized publications.  Their coverage is truly amazing.  For
  example, you can get your 300-word press release to the national
  news agency and all major media in Bangladesh for US$265.
  They'll send photos, and they'll store your own letterhead in
  their computer for repeated notices.

  The fact is that news organizations use these materials.  Now,
  nobody can complain when the materials in question are simple
  matters of fact.  The problem starts when the news organizations
  do not -- whether to save money or to avoid offending potential
  sponsors or because they simply don't care -- dig beneath the
  "spin" on the stories they get from the PR Newswire to see what
  other framings of the story might better inform the public.

  So you might want to learn more about PR Newswire by getting
  their literature.  But don't harass them.  Only get it if you
  really want to read it.  Thanks a lot.



  Mike Sunderland <sunderland@decus.org.au> was pleased by my
  comments about language in e-mail that mistakenly tends to
  presuppose that the sender and reader are in the same country.
  Extending my main example, the phrase "this country", he points
  out that commonly used phrases such as "this summer" tend to
  presuppose that the sender and reader are in the same hemisphere.

  Al Gore's National Performance Review now has some web pages.
  Have a look at http://www.npr.gov/ and ignore the huge bit maps.

  The Office of the (US) Secretary of Defense now has a web page
  as well.  Much fun.  Aim your WWW client at http://www.osd.mil/

  In a neat bit of symmetry, the Survival Research Laboratories
  (which stages allegorical demonstrations of mechanized warfare
  on city streets) is on the net now as well.  The URL is

  The August 8th issue of the direct marketing industry journal
  "DM News" reports that "Affinity Marketing Group is offering the
  29,888-name Hillary Haters list", which consists of individuals
  who "despise anything and everything to do" with Bill and
  Hillary Clinton.  Affinity says that these folks are "proven
  supporters of patriotic and ultra-right wing causes, campaigns
  and committees and ... are pro-military, anti-gay, pro-gun,
  anti-welfare, pro-American and anti-foreigner".  You can reach
  Affinity at PO Box 2409, Fairfax VA 22031, phone (703) 978-4927,
  fax (703) 978-7832.

  The Institute for the Study of Civic Values' gopher contains
  many things relevant to community networking and community
  development.  Aim your gopher at gopher.civic.net port 2400
  From a unix prompt: gopher gopher.civic.net 2400

  Here, courtesy of dbatterson@aol.com via a very useful on-line
  newsletter on e-mail an government published by Jim Warren
  <jwarren@well.com>, is a list of e-mail addresses for groups that
  are fighting anti-gay ballot initiatives:

   * Oregon's No on 13 Campaign (formerly Save Our Communities PAC)
     is at: socpac@aol.com .

   * Idaho's No on 1 Coalition is at:  NoProp1ID@aol.com .

   * The Gay and Lesbian Victory Fund is at:  victoryf@aol.com .

   * Digital Queers is at:  kwickre@aol.com .

   * National Gay & Lesbian Task Force is at:  ngltf@aol.com .

   * Gay & Lesbian Alliance Against Defamation is at:
     glaadsfba@aol.com .

   * International Gay & Lesbian Human Rights Commission is at:
     iglhrc@agc.apc.org .

  I wandered through the local VA hospital the other day and
  happened upon an issue of "Atomic Veterans Newsletter" for
  veterans who got sick from being exposed to US nuclear weapons
  tests.  It's fascinating.  The address is NAAV, PO Box 4424,
  Salem MA 01970-6424.  Membership (including a subscription to
  the magazine) is $15 a year, but it sounds like they might only
  accept memberships from people who were actually exposed to
  radiation.  You might want to check it out anyway.

  My favorite palindrome is: Satan, oscillate my metallic sonatas.

  Phil Agre, editor                                pagre@ucla.edu
  Department of Information Studies
  University of California, Los Angeles         +1 (310) 825-7154
  Los Angeles, California  90095-1520                FAX 206-4460
  Copyright 1994 by the editor.  You may forward this issue of The
  Network Observer electronically to anyone for any non-commercial
  purpose.  Comments and suggestions are always appreciated.

Go back to the top of the file