Most universities incur out-of-pocket costs of less than a dollar a
month for Internet use by each of their students, faculty, and staff. In
a world of ever-increasing prices, the Internet is one of the greatest
bargains around. The cost is so low, in fact, that it continues to cause
consternation in the telecommunications industry, some of whose
conservative members smell a socialist plot to undermine the Republic.
The Internet is cheap for good reasons, and as we work our way toward
universal network access, it is worth a bit of explanation why. There
are three major areas that combine to produce low costs: Internet
architecture, Internet services, and the economic structure of the
Internet.
Internet Architecture
The traditional voice network in use today is constructed based on a
system of expensive digital and analog circuit switches that sit inside
the network. At the ends of the network, we users have analog handsets
to talk and listen with. From a technical architecture perspective, this
network is smart inside and dumb at the ends. The Internet design is
essentially the reverse. In the Internet, smart computers sit at the
ends, linked together by a system of dedicated circuits and inexpensive
packet switches. The Internet design is both absolutely and relatively
less expensive than the telephony design. It is absolutely less
expensive because the network does much less work in handling packets
from source to destination than it does handling voice circuits and
services; it is relatively less expensive because the networking
functions implemented in mass-produced personal computers and
workstations are superior technically to a large, installed bases of
much older equipment used by the telephone companies.
The older design of the telephony switches creates another level of
inefficiency and cost by using fixed-frequency boundaries when creating
voice circuit paths. Those frequency bands, or channels, are tied up for
the duration of a phone call regardless of the utilization of the
channel by voice traffic. Internet traffic, on the other hand, is
composed of variable-length packets, which move from source to
destination locations on a best-effort basis, thus improving circuit
utilization but also, occasionally, resulting in delays that are
perceptible to users.
Internet Services
The three major services offered on the Internet today are electronic
mail, file transfer, and connection to remote computers. In contrast to
a telephone call, in which the expectation is that the network will
transmit each user's voice within tenths or hundredths of a second, the
Internet's services have more relaxed response requirements. At busy
times, mail can take minutes or even hours to deliver. File transfers
take place at whatever speed is available on the network and are not
infrequently "timed out" if congestion is severe. And remote
connections, or "telnet" service, may not be available at all during
congested periods.
Considered from the network design point of view, voice constitutes a
premium service, because once initiated, a telephone call must be
provided in real time so as to prevent users from experiencing
perceptible delay in their conversations. Internet service, in contrast,
is tolerant of delay, which permits the network to achieve greater
overall utilization and efficiency.
Internet Economics
The first thing that comes to mind about low Internet costs is
government subsidy, but that factor is relatively small--perhaps 10
percent--and now declining rapidly. Of possibly greater significance is
the continuing government funding of advanced network research and
applied development. Many of the most important features of the
Internet, including TCP/IP protocols and such applications as Eudora and
Mosaic, were paid for with research grant dollars and do not have to be
recovered by way of network charges.
Several other economic factors drive Internet costs down. First, the
Internet is primarily an institutional network, whose network managers
generally "get it cheaper wholesale" by aggregating demand from an
entire university campus, a corporation, or a public agency. Second, the
resources those managers are buying--dedicated circuits and packet
switches--are among the least regulated and most competitive services
provided by the telecommunications industry. Third, the Internet
community does business on a very low overhead basis. For instance,
there exists essentially no billing overhead, many intercampus and
interregional services are bartered, and many services are donated by
the participating organizations.
Taken together, the major Internet attributes of efficient design, new
technology, deferred services, wholesale purchasing, and low-overhead
management have resulted in a cost structure that is generally an order
of magnitude less than the current tariffs levied by the regulated
carriers for equivalent commercial services.
The Future
There has been a bit of a rock-throwing contest going on between
defenders of the Internet approach to costs and pricing and traditional
telecommunications carriers. Upstart Internauts delight in slinging
missiles at overstuffed (oops, I meant overstaffed) telephone companies.
During the next several years, both camps will need to adjust to many
changes as described below.
-- Deregulation, new competition, and a continuing flow of new
technology will bring the real costs (and prices) of telecommunications
networks down dramatically--by at least 50 percent by the end of the
decade and as much as 90 percent in some situations.
-- New Internet applications, especially those including voice and
imaging, will require guaranteed network response times and will
generate increased costs. The allocation of access to a multitiered set
of Internet services will represent a contentious process on many
campuses.
-- Widespread use of the Internet for business and value-added
information applications will result in a variety of new payment
arrangements, including pay-per-view, site licenses, and time-of-day
differentials.
-- Reliance on the Internet by important sectors of the economy will
require a more robust approach to operational integrity and security,
with costs somewhere between those of today's low-security Internet and
those of the regulated carriers. The deployment of new technology may
result in no net increase in cost on a per-user basis.
We are rapidly leaving the era of the one-size-fits-all Internet and
feeling the growing pains of economic reality in cyberspace. Let's give
thanks for that dollar a month, and let's vow to keep the providers
honest.
Michael M. Roberts is vice president of Educom. roberts@educom.edu