Control, Change and the Internet

Patrick Maslen (patrick@quest-bird.com)

Chapter 2: Bye-Bye, Backbone

"In general when faced with obnoxiously centralized control over something that should be free and/or distributed, I look for a low overhead way around, that increases freedom in general."

- John Gilmore, co-creator of the alt.subnet on USENET

While the Protocol Wars were heating up, the Internet faced problems of its own 'at home'. Electronic mail and news grew in popularity on the Internet and related networks. The volume of traffic on networks like USENET (User's Network) also increased into an organisational nightmare for the technicians who maintained university computer networks. An influential group of these system administrators formed the Backbone Cabal, which for a few years tried to control USENET. However, as USENET increasingly intertwined with the Internet, and as access to the networks became easier, control of Internet and USENET became impossible.

The Protocol Wars presented a huge international problem for the Internet. The first years of the Internet were difficult because of national and international uncertainty about the future of networking protocols, but this was a problem faced mainly by the computer scientists and technicians who had designed TCP/IP. For the people already connected to the American network, a much more significant issue was who controlled the Internet. The Internet in the early 1980s had grown large enough to attract notice, but was small enough to be controlled by select groups. The muddied issue of control of the Internet in America did not help the resolution of the Protocol Wars. However this potential for control of the Internet declined as the decade grew older, due to the nature of the Internet Protocol itself, ever-increasing numbers on the networks, rapidly evolving technology and increasing social pressure from the new network community.

One of the most important developments in computer networking was the invention of electronic mail, or 'e-mail'. Few people foresaw the social importance of e- mail. It could be used by people on any computer network, not just the Internet. The Internet, however, allowed e-mail (among other things) to be passed between networks. Connection to the Internet meant that many people in previously unconnected organisations could communicate quickly, cheaply, using electronic mail. They could already do so using the telephone network. However, electronic mail possessed a feature which the telephone did not share, and this was the ability to transmit a message to multiple recipients.

With e-mail, people found that they could broadcast news to others on the networks. So they did. As the idea caught on (and as numbers on the network grew and grew), the methods of broadcasting became more sophisticated; news became distinct from copied mail. One innovation used in ARPANET was the 'mailing list'. Mailing lists were organised by discussion subject; people would electronically 'subscribe' to a mailing list (though this did not involve money changing hands in most cases.) The mailing list would then send copies of any mail to every other person on the list. One of the first mailing lists on ARPANET was called "SF-Lovers", for science-fiction fans.[1] It was unofficial, and very popular.

In 1979, graduate students Steve Bellovin and Tom Truscott established a network (originally between Duke University and the University of North Carolina) for the transport of news, called USENET. Although USENET was not part of ARPANET or the Internet (which didn't even exist in 1979, except on paper), it evolved in a similar environment: that of American universities using the UNIX operating system on their computers. The USENET network spread across North America on an ad hoc basis, between individual universities and computers, using a protocol called UUCP (UNIX-UNIX Copy Protocol).

USENET was intended originally to carry mainly 'UNIX bug (fault) reports'[2] but as it grew in size and popularity, newsgroups 'without any technical meaning'[3] began to show up. Like mailing lists, USENET organised news into different subjects, called newsgroups. The 'articles' in the newsgroups were stored on a computer in the USENET network, then forwarded to the next computer or computers in the network on request, together with any replies or new postings. In this way, news could be passed at varying speeds around the country.

USENET Growth
Date USENET Sites No. of groups Articles per day
1979 3 3 2
1980 15 3 10
1981 150 3 20
1982 400 3 35
1983 600 3 120
1984 900 3 225
1985 1300 3 375
1986 2200 241 946
1987 5200 259 957
1988 7800 381 1933
(source: Zakon, Hobbes' Internet Timeline v.2.4a, 22 Feb 1996.)

Originally, these newsgroups were themselves sorted into two broad categories: "mod", for those groups with 'moderated' or edited articles and "net", for other, unmoderated newsgroups. Later, the "fa" ('from ARPANET') group of newsgroups was added.[4] Throughout the early eighties, USENET grew in complexity and size. The news software it ran went through several revisions, the work of enthusiastic amateur programmers on the network. These revisions allowed the growing USENET to cope with its own expansion. The numbers of newsgroups increased dramatically, making the original three newsgroup categories huge and disorganised.

Participation in USENET was voluntary and non-official. Individual system administrators decided whether a site would accept a news 'feed' from USENET. Not all system administrators (also known as "sysops") wanted all of the various newsgroups on their systems. What is more, USENET was not free, and had no sponsorship from any government organisation. To participate in USENET, individual system administrators and universities had to foot the (sometimes large) phone bills associated with the regular transport of news on the network. System administrators had a large say in which newsgroups would or would not be carried by their computers.

One group of system administrators, who were willing to devote considerable resources to the prompt and stable propagation of USENET news, came to be known as the "backbone" or the "Backbone Cabal". The word 'backbone' later came to mean a particularly fast or reliable set of nodes on the Internet such as the National Science Foundation Net (NSFNET) Backbone, completed in 1986. The USENET backbone existed because the network in the early eighties was small and obscure, and news traffic had to pass through a small number of nodes controlled by one (albeit loose) group - the Backbone Cabal. The Backbone Cabal was organised by one Gene Spafford, who was a system administrator at the ARPANET-connected site "PURDUE" in Chicago. This group became very important in the creation and naming of newsgroups, because nearly all news travelling across the country, or to Europe, would travel through backbone computers. If the members of the backbone 'cabal' refused to carry a newsgroup, they could seriously reduce its potential audience.[5]

In 1986, there was only one stable USENET backbone connection to European USENET sites: "SEISMO" at the Center for Seismic Studies in Virginia, administered by Rick Adams. Transatlantic phone connections between computers were expensive, and the Europeans weren't prepared to pay for 'fluff groups'[6] which contained 'idle banter.' Adams wanted to make the chaotic newsgroup structure easier to manage; he proposed that some controversial newsgroups be put into a 'talk' category to make them easier for system administrators to manage (that is, omit). This started an enormous USENET argument, a "flame-war", while the Backbone Cabal used their considerable influence to reorganise the newsgroups into new categories. This process was called the 'Great Renaming'[7] and drew USENET attention to the power of the Backbone Cabal.

The 'Great Renaming' led to another problem. There were some newsgroups that the Backbone Cabal simply would not condone, such as those discussing sex and drugs, partly because of their conservatism, partly out of concern that their bosses would shut down all USENET feeds if they found out that the network was carrying that sort of traffic. However, disagreements among the backbone members over these 'forbidden' newsgroups caused Brian Reid (sysop of "HOPTOAD"), John Gilmore (who ran "MEJAC") and Gordon Moffett, all friends and Backbone members, to 'defect'.

After a barbecue on May 7, 1987, the three system administrators agreed to set up (using their own and public-access sites) an alternative or 'alt.' network, which would avoid the backbone (and, theoretically, ARPANET) sites. Three weeks later, in a post to the new subnetwork, John Gilmore wrote, 'if we keep connectivity among a set of private machines, we can't be cut off by bureaucrats even if the rest of the sites go away.'[8] The alt.network carried Reid's "alt.gourmand" newsgroup (he objected to the Backbone Cabal's less flamboyant "rec.food" suggestion) and Gilmore's "alt.drugs". By early 1988, the 'alt.' groups also included "alt.sex" and (for completeness) "alt.rock-n-roll".[9]

The Backbone Cabal was at its strongest during and immediately after the Great Renaming. But the Internet intervened. In the mid to late eighties, more and more USENET sites were using TCP/IP, which allowed news to be shunted all over the network, ignoring the erstwhile backbone sites. The Internet Protocol and the growth of networks outside the conservative ARPANET allowed the 'rebellious' alt.newsgroups to flourish, while avoiding regulation by the Backbone Cabal. The growth of the Internet broke the power of the Backbone Cabal to regulate public speech on the networks. The Cabal itself soon ceased to exist.[10]

Control of the Internet's social structure was linked to control of its physical structures. The TCP/IP protocols made control of the whole system impossible unless control of each of its parts was possible. However, in the early days of the Internet, before the network became very large, such control or at least influence was not out of the question. The ARPANET and the USENET backbone demonstrated some control over the computer systems of the Internet in the early eighties, because the equipment was owned by small groups. Another large part of the physical infrastructure of the Internet was initially owned by another group, the American Telephone & Telegraph Company (AT&T).

AT&T held an utter monopoly on telephone communications in America until 1982 (the year the Internet Protocol was adopted by ARPANET), when the United States Justice Department achieved an agreement with AT&T to break up the Bell System by the end of 1983.[11] As a medium for inter-computer communications, the Internet and its predecessor ARPANET had always been directly affected by AT&T, sometimes adversely. The contract for all of the cross-country links in ARPANET was awarded to AT&T by ARPA, but the Network Working Group had many problems with the company.[12] This was partially because the long distance dedicated lines required by ARPANET were a technical challenge in 1970, but also because of AT&T's aggressive protection of its monopoly on communication in America.

AT&T staff were negative and sceptical about the ARPANET until its spectacular demonstration at the International Conference on Computer Communications in 1972.[13] AT&T's contributions to the history of the Internet were not all negative. Bell Labs (owned by AT&T) produced the UNIX operating system, and offered it to universities cheaply.[14] The TCP/IP protocols spread very quickly when combined with UNIX. So, in a way, AT&T worked to undermine its own monopoly. But then, it is not altogether surprising that different parts of such a large- scale organisation might work at cross-purposes. AT&T's Chairman in 1972 was John DeButts, who lobbied Congress to protect 'the need for a unitary telephone service.'[15] DeButts' position as head of one of the largest companies in the United States gave him great political influence; that same year ARPA received a letter from the Office of Telecommunications Policy of the White House, which urged the agency to 'divest the network to private industry.'[16] In 1972, "private industry" in telecommunications meant, basically, AT&T.

The first few companies to attempt to set up commercial computer networks based on the ARPANET model in the mid seventies, failed because of huge legal costs incurred fighting lawsuits from AT&T (and some other large American corporations, including RCA, ITT and Western Union[17]), who saw any new communications technology as a threat. This sort of behaviour leads me to believe that the fledgling network industry was improved by the AT&T divestiture; certainly the Internet (and USENET) flourished, free from AT&T's direct control of its infrastructure. Dr. Jonathan Postel, network "old boy", stated 'vendor independence'[18] as one of the reasons for the Internet's success. Throughout the 1980s, after the divestiture, fierce competition between AT&T's traditional competitors and the new "baby Bells" reduced the cost of direct Internet connection and leased lines. Even if the Internet had nothing to do with the grinding legal process which eventually dismantled AT&T, it certainly benefited from the outcome.

Cheaper telephone line connections in the wake of the AT&T divestiture, a telecommunications industry far less hostile to computer networking and rapidly improving computer technology and lower hardware prices combined to make connection to the Internet that much easier during the mid-late 1980s. Prices of computers and computer components dropped steadily throughout the eighties. Personal computers became easily available, were networked in offices into Local Area Networks (LANs) and linked into the Internet. This increased the numbers on the network and dispersed its ownership. The construction of NSFNET in 1986 also accelerated the Internet. Its sheer rate of growth made the Internet uncontrollable physically and socially, led to the breaking of the USENET Cabal and possibly influenced the outcome of the Protocol Wars.

In 1986 the National Science Foundation funded NSFNET, a backbone connecting five super-computing centres across the United States, and enormously increasing connectivity of universities to the Internet. The number of Internet hosts jumped from 5089 in November, 1986 to 28,174 in December the following year, a growth of around 550 per cent. During the same time period (1986-7), the number of USENET sites almost doubled[19] , jumping from 2200 to 5200 and causing the Backbone Cabal much administrative grief in its last days. The NSFNET backbone cemented the link between USENET and the Internet and sealed the doom of the Backbone Cabal, but it brought its own influences (if not controls) to the Internet community.

NSFNET provided a huge boost to the numbers of people on the Internet, but it influenced those people strongly through its Acceptable Use Policy (AUP):

NSFNET Backbone services are provided to support open research and education in and among U. S. research and instructional institutions, plus research arms of for-profit firms when engaged in open scholarly communication and research. Use for other purposes is prohibited.[20]
NSFNET's Acceptable Use Policy encouraged the 'high degree of intolerance for commercial postings'[21] which is still found on USENET. NSFNET discouraged commercial activity on the Internet with its policy, which made the Internet seem to businesses to be an academic institution. Since most users of the Internet in 1986 were in fact affiliated with academic institutions anyway, this effect was not too hard for the National Science Foundation to achieve. For a while, NSFNET provided the main cross-country Internet link in the United States, and its Acceptable Use Policy strongly influenced all netizens.

By the late eighties, commercial entities were seeing the advantages of connection to the Internet. By 1988, many companies were pressuring the National Science Foundation for unrestricted access to its backbone.[22] The funding for the original NSFNET backbone came from the U. S. Government and so ultimately from American taxpayers. NSFNET's subsidy gradually dried up, and private enterprises increased their interests in the backbone. It was upgraded in 1988 by Merit Computer Network, MCI and IBM, who later formed ANS, which took over control of NSFNET, an action which made other commercial networks 'a bit queasy.'[23]

However, by mid-1993, the NSFNET was only one of a mesh of interconnected commercial backbones known as the CIX or Commercial Internet eXchange. The influence of the NSFNET backbone was reduced by the growth of other backbones. In fact, the whole concept of control of the Internet by one group via backbones (USENET or otherwise) became outdated during the late 1980s and early nineties. This 'mesh' of interconnected computers was close to the distributed network envisioned by Paul Baran twenty years previously, but the Internet did not resemble his grim idea of a post-holocaust command grid.

The various backbones and controlling influences on the developing Internet in the 1980s each left their mark on Internet culture. From ARPANET came the sense of community and collaboration which made the original network design possible. USENET and the Backbone Cabal added (or emphasised) idealism and vociferously subversive undergraduate counter-culture. The NSFNET backbone encouraged distrust of big-business and commercial activity, helped in this regard by the university culture of USENET and the ARPANET's experience with AT&T in the early seventies and eighties. Then, in the early nineties, the CIX and lateralisation of the backbones helped Internet society to gradually accept, if not fully embrace, commercialisation of the nets. Although the potential for control of the Internet faded as the 1980s advanced, these influences lingered: the legacy of the backbones.

The 1980s were a tumultuous time for the Internet. As the Internet Protocol was adopted by more and more people throughout the decade, aided by the American State's divestiture of AT&T and the construction of NSFNET, TCP/IP's advantages over other protocols became clear, or at least clearer. TCP/IP's decentralised system destroyed the rigid hierarchy of the Backbone Cabal, and helped to fashion a unique culture. This culture was flavoured by the (sometimes contradictory) ideas of groups which had attempted to control it during its growth, but it was shaped by TCP/IP, the almost All-American Protocol.


Footnotes

  1. Hafner and Lyon, Where Wizards Stay Up Late, 1996, p. 201.
  2. Steve Bellovin and Mike Horton, unpublished report on early USENET, in Salus, Casting the Net, 1995, p. 133.
  3. Lee S. Bumgarner [LSBUMGAR@vaxl.acs.jmu.edu], 'Great Renaming FAQ 2.0 beta (text this time)', in [alt.culture.usenet], [alt.folklore.internet], 31 August 1995.
  4. Hardy, 'The History of the Net', [http://www.ocean.ic.net/ftp/doc/nethist.html], 1993.
  5. Matt Schnierle [pyld@oak.grove.up.edu], 'alt.conspiracy.usenet-cabal FAQ', [http://www.ma.iup.edu/`pyld/faq.html], 7 January 1996.
  6. Bumgarner, 'Great Renaming FAQ 2.0 beta (text this time)', 31 August 1995.
  7. Hardy, 'The History of the Net', [http://www.ocean.ic.net/ftp/doc/nethist.html], 1993
  8. John Gilmore, 26 May 1987, in Salus, Casting the Net, 1995, p. 145.
  9. Hardy, 'The History of the Net', [http://www.ocean.ic.net/ftp/doc/nethist.html], 1993
  10. Although some (completely without basis) maintain that the Cabal still secretly controls USENET. See Matt Schnierle, 'alt.conspiracy.usenet-cabal FAQ', 7 January 1996.
  11. Barry G. Cole, After the Breakup: Assessing the New Post-AT&T Divestiture Era, 1991, p. 1.
  12. Salus, Casting the Net, 1995, p. 61
  13. Cerf, 'How the Internet Came to Be', in Aboba, The Online User's Encyclopedia, 1993, p.528.
  14. Steve Holmgren, 'Network UNIX', RFC 0681, [NIS.NSF.NET/internet/documents/rfc0681.txt], 18 April, 1975.
  15. Cole, After the Breakup, 1991, p. 2
  16. Salus, Casting the Net, 1995, p. 80
  17. ibid. p. 109.
  18. Dr. Jonathan B Postel, quoted in G. Malkin, 'Who's Who on the Internet: Biographies of IAB, IESG and IRSG Members', RFC 1336, [NIS.NSF.NET/internet/documents/rfc1336.txt], 19 August, 1991.
  19. Robert H. Zakon [zakon@info.isoc.org], Hobbes' Internet Timeline v.2.4a, [http://info.isoc.org/guest/zakon/Internet/History/HIT.html], 22 February 1996.
  20. NSFNET AUP Excerpt, from Salus, Casting the Net, 1995, p. 237.
  21. Margaret L. McLaughlin, Kerry K. Osborne, Christine B. Smith, 'Standards of Conduct on Usenet' in Steven G. Jones, CyberSociety: Computer-mediated Communication and Community, 1995, p. 106.
  22. B. Kahin, 'Commercialization of the Internet: Summary Report', RFC 1192, [NIS.NSF.NET/internet/documents/rfc1192.txt], 12 November 1990.
  23. Salus, Casting the Net, 1995, pp. 199-200.

Where to next?