[BITS] ICS 230 Midterm: Social Construction of the Internet

Rohit Khare (rohit@godzilla.ICS.uci.edu)
Mon, 23 Feb 1998 18:10:54 -0800

Dr. King --

In fear of going over the five-page limit, I kept to a high-level argument that's only
four pages. My apologies for the generalization, but I feared a fifty-page defense of
details even more.

Everyone else --

felt this would be good debate fodder.

Fire away,

Rohit Khare


[1]ICS 230 Midterm:
Social Construction of the Internet

-1. Prologue

[For those not enrolled in the course...]

The Social Constructivist viewpoint of the history of technology
focuses on the people, politics, and cultural context of innovation.
We have been studying several examples of such historiography, cases
of early Portuguese navigation, missile guidance, and bicycles.
Bakelite, for instance, is based on a synthetic resin reaction which
was discovered thirty years earlier: a case can be made that the
actual invention was stymied by close-mindedness amongst 'current'
plastic engineers (celluloid, et al). User groups are also an
important part of the picture. Pneumatic bike tires were proposed as
an anti-shock innovation, but were rejected by the bourgeois, adopted
by macho racers for speed, and took over from the back door. In the
modern era, we might consider the social construction of laser
printing and desktop publishing: an innovation born of parts lying on
the shelf for years until the proper constellation of professional
practices, corporate restructuring, and sales pitches became aligned.

0. Abstract

While Internet technology can be argued as following a technologically
deterministic development path, the emergent uses and adoption of
Internet services -- and Internet philosophy -- is more properly
viewed as a socially constructed project. In this essay, I argue that
the arc of Internet engineering was largely inherent in its origins.
However, the adoption of IP as the dominant internetworking format was
more a social process that leveraged the converging interests of
several communities. I explore this hypothesis in two contexts:
adoption of the World Wide Web and of the Internet standards
development process. Finally, these concepts help set the stage for
tracing the future evolution of the Internet technological frame.

1. Internet Technology

The Internet's origin-myths have been told and retold enough to be a
fruitful subject of anthropological and semiotic study, to say nothing
of the history of technology. The hoary claims that it was funded out
of nuclear warfighting command and control, for example. The
hagiography of the Wise Old White Men who had the foresight to capture
hippie openness in software, for another. These fragments contribute
to a heroic conception of the enterprise. Arrayed upon the other hand
are the images of inevitable Progress, riding serenely upon the
advances of processing power, storage capacity, and bandwidth
consequent from Moore's Law.

I would argue that the technological arc of Internet engineering --
packet switching; IP addressing, packet layout, fragmentation, and
tunneling; TCP streams; routing; domain naming; and the explosion of
application-layer tools -- is largely inherent in its premises (viz.,
that intelligent nodes at the periphery of the network send messages
across an unreliable internetwork). It seems predictable, in some
sense, that the 'reverse salients' would begin with the data
communications channel, then the problem of tunneling multiple
protocols within a single format, then connecting more than a few
dozen ARPAnet nodes, then numbering and naming sites, a distributed
database for name updates, then new, Internet-specific computing
services. I immediately grant a measure of post-hoc rationalization in
this depiction, but there is a sense of a "design hierarchy" of

Whereas the social constructivist viewpoint would consider suspect any
claim that a technology is advancing at its 'scientific speed limits'
-- as fast as the innovations can pour out of the lab. I would argue
that the Internet has been the focus of such concentrated resources
and broad-based participation that everything that could be invented
at each stage, was. The performance curves for chips, memory, and data
links can be plotted very reliably into the future, even though each
new stage of performance may require entirely new, noncumulative R&D.
Nevertheless, the only real question is whose name to attach to each
achievement; but it's still sure as seasons that HotChips X this
summer will have a gigahertz processor as that 1999 will herald 10Tbit
x 1,000-km fiber drivers, and so on...

The confluence of the postwar Big Science establishment, systematic
commerical exploitation of innovation, and the prosperity of the
information technology industries in particular became an ideal
incubator for the Internet technological frame. Furious
experimentation seemed to ensure that no veins went untapped:
videoconferencing, hypertext, message transfer, routing, and the whole
spectrum of current Internet applications seem to have predecessor
research dating back throughout the (brief) fossil record of Internet
development. Several of these innovations, in fact, were proposed on a
regular cycle, but only seem to have stuck at certain times, with
certain communitied, for certain purposes. The social processes
determining such adoption are the topic of the following sections.

[Short version of above: nothing really ever came outta left field in
the history of Internet technology -- at least, not a surprise to a
disciple like myself, already inculcated into the frame...]

2. Internet Adoption

As successful as it eventually became, internetworking via packet
switching was not born of demand from restless users. It could be
considered a classic case of 'giving users what the need, not what
they want' -- another motive myth that seems to be etched deep in the
professional identity of engineers, a smugly effective breed. It was
an experimental approach that slowly recruited larger constituencies
for its advancement. First, its direct developers and academic users;
managers; commercial developers; standards development organizations
(SDOs); and finally a wave of second-order mass adopters: men, college
students, the elderly, and so on. The economic arguments for IP's
eventual dominance -- network effects -- stem from these
consituencies' adoption, not the cause of it.

Early users of the ARPAnet seem to be enthusiastic leaders from a
small coterie of computer science departments and military research
labs. They were directly invested in porting IP software to new
platforms, developing TCP and related applications, and had a
tradition of shared infrastructure development (UNIX, Lisp Machines,
etc). This led naturally into adoption of electronic communities:
email, discussion lists, and netnews stabilized rapidly. By the early
Eighties, internetworking became commerically significant, attracting
new support from manufacturers freed to innovate new transmission
technologies below IP (Ethernet, Token Ring, etc). Commercial support
also joined the battle against other competitor peers (IPX, SNA,
AppleTalk) which IP eventually assimilated and eliminated. Managers,
the purchasers of IT equipment, refereed this battle according to
their own prejudices ("CSMA/CD randomly retries transmission?! But
we're a bank!"), culminating in the purchaser-side battle of OSI vs.

Mass adoption depended on a second wave of information consumers,
beyond the Wired generation. One of the earliest drivers of 90's
Internet adoption was campuswide email for undergraduates. Children of
middle-class Boomers with home computers spread email access to their
families through online services. Libraries became comfortable with
online catalogs, CD-ROM databases, and now online research. Young men
became enamored of yet another vehicle for high-fidelity pornography
-- which tipped the next boom, electronic commerce, the need for
privacy and accountabilty (payment). These communities also deserve
the role of "social construction", since their agendas directly
motivated the development agenda of the NII, the FCC Universal Service
fund for IP, the Communications Decency Act, multicast audio and
video, resource reservation protocol, and ...

At some point in the early 1990's, IP achieved a measure of lock-in:
new OSes were required to ship with TCP/IP stacks, OSI was finally
entombed, and ancillary technological frames acquiesced as well (IP
over power lines, IP for satellite telephony, IP for radios) This is
held up as a classic example of the network effect: surplus benefits
from joint adoption of a common standard. It remains to seen how much
the economics motivated each constituency, compared to more social
processes: the degree to which IP technology represented their
political views, cognitive models of computing, and expectations of
applications, particularly amongst the second wave (the "early
majority" 3).

3. Case: World Wide Web

Perhaps the most visibly social phenomenon on the Internet today is
the World Wide Web, a community project in form (protocols) and
substance (sites). It seem informative to ask how the Web succeeded in
promulgating global hypertext where so many came before: Xanadu,
StorySpace, Brown's IRIS, HyperCard, the Dexter Model, Windows Help,
Gopher, Hyper-G (later Hyperwave) and so on. I would argue that these
were insufficiently grounded in the Internet frame, especially as far
as free-software, intelligence-at-the-periphery, and
incremental-adoption -- and most of all, worse-is-better2, as far as
link integrity, multimedia, and copyright protection features. The Web
springs from a bridge figure: the CERN team, partially grounded in
large systems, the Internet, hypertext, and physics information
sharing. So, even though the strategic goal of global hypermedia had
been long predicted (Memex, oNLine System), the tactical maneuver was
launched from a user community. As it diffused, the artifact was
reconstituted by a series of worldviews: graphic designers, who railed
at HTML's insubstantiality; web designers, who reveled in it;
publishers, who saw online analogues to newspapers and magazines;
programmers, who saw a new platform; ...

4. Case: Internet Standards Development Process

Internet governance is intertwingled with its technical structure.
Both are built on the basic philosophy of
intelligence-at-the-periphery (end-to-end), with transparent
coordination at the center (routing). The IETF is an all-volunteer
cabal where development proceeds in working groups, coordinated by a
small steering committee which inveighs power in people and ideas, not
processes (like ISO, ITU-T and traditional SDOs). Constituencies
within IT standardization have been actively choosing one over the
other during the Internet era. Scientists and engineers have fought
for openness against SDOs with overpriced publications; developers,
for electronic communication over vast travel budgets; corporations,
for private consortia operating at a faster pace; and regulators,
opting for voluntary recommendations over administrative fiat (e.g.
PICS vs. CDA, the 'Grand Alliance'). At the edges, though, other
groups are opposed to the Internet's ad-hoc rules, particuarly for
domain names. Trademark owners and governments are pushing to reclaim
control, with technical effects that refract back into released

5. Critique

Yet, where will this frame strand us? What might a social
constructivist predict for the Net's evolution from these sketches? I
would argue IPhas become a frame with staggering momentum, probably
too much -- Vint Cerf is on record as working with JPL on
Interplanetary IP. Making TCP work across delays of light-minutes
reveals yet more presumptive anomalies, just as do IP routing in
optical networks, domain naming policies, IP mobility, and so on.
Responses within the frame are getting more and more strained: Active
Networks propose packets which are actually Turing machines, not data.

I would suggest searching for bridging figures with lower committment
to the IP frame: don't wait for SIGCOMM; watch neurobiology, instead.
'Computer people' see networks as knowable graphs of nodes and links,
all deterministic. Every node can reach every other node; routing
proceeds along adminstrative trust boundaries. Biology knows that
self-organization doesn't proceed top-down, and that's where the real
breakthroughs in mobility, management, and allocation may lie
(including that evolved behavior, economics).

6. References

1 Kim B. Clark, The interaction of design hierarchies and market
concepts in technological evolution, Research Policy, v14, 1985, pp

2 Richard Gabriel, Patterns of Software, 1997

3 Geoffrey Moore, Crossing the Chasm.

[2]Rohit Khare


1. http://www.ics.uci.edu/~king/winter98.html
2. mailto:rohit@uci.edu