Control, Change and the Internet

Patrick Maslen (patrick@quest-bird.com)

Chapter 1: Protokollon

protocol n. Original draft of diplomatic document; formal statement of transaction; rigid prescription or observance of precedence and deference to rank as in diplomatic and military services; official formula at beginning and end of charter, papal bull, etc. ~ v. Draw up protocols; record in protocol. [Gk protokollon fly-leaf glued to the case of a book (kolla glue)]

The Internet is a 'non-teleological, self-organizing system that combines human and machine communication, reasoning, and associative capabilities.'[1] Throughout its short history it has been growing and changing, sometimes at a rate which clearly demonstrates that it is wild, out of control. The Internet 'seems to be both institutional and anti-institutional at the same time, massive and intimate, organised and chaotic.'[2] But however strange it appears, the Internet is not an isolated or freak phenomenon but a link in a long chain of communications technologies created by human beings. Writing, the printing press, the telegraph and telephone, radio and television; the Internet was built upon and incorporated elements of all of these previous technologies. Like them, the new medium has profoundly affected social institutions and cultural perceptions.

In late May, 1844, Samuel Morse sent the message "What hath God wrought?" down the world's first electromagnetic telegraph line, which connected Baltimore and Washington D. C. in the United States.[3] Telegraphy connected nearly all of the world's cities into a "wired world" in which information could move faster than and independently of transport or geography. This 'wired world' enabled the development of networks of telephones and later, computers. People used the telegraph in its early months to play chess, but later the invention re-organised commerce[4] and played a key role in the American Civil War. Like the Internet a century and a half later, the telegraph's effects on society, commerce and the military were unforeseen by its designers.

The military has had a historical role in developing communications systems. The Internet is no exception to this trend. In the United States, the interlocking of science and the state began in the Second World War; the first computers emerged from this union. The RAND corporation, the nation's 'foremost Cold War think-tank'[5], was set up in 1946 to preserve the research capability of the United States. In 1962 a researcher at RAND, Paul Baran, was concerned with the 'dangerous fragility' of the United States' communications network in the event of a nuclear attack, which might make it impossible for the United States to retaliate against an enemy's first strike.[6]

Baran proposed a distributed, packet-switched network of computers with no centralised (and therefore targetable) authority. For survivability, Baran recommended that each computer (or 'node') in the network should be redundantly connected to at least two other nodes. Each node would have 'its own authority to originate, pass and receive messages.'[7] Messages would be divided into "packets"[8], short pieces of information which travelled separately to their destination. No node in the network would be more important than any other and the network could not be shut down unless every node in the network was shut down. Baran's proposed network could not be destroyed at a stroke, but neither could it be centrally controlled.

Baran's ideas were rejected by the giant United States telephone company AT&T, but later discovered by the Advanced Research Projects Agency (ARPA), an organisation which funded military research. Computers were expensive in the 1960s, and ARPA was interested in sharing its computer resources, which were spread out across the country. ARPA used Baran's ideas and sponsored the creation and development of a packet-switching network called ARPANET (for Advanced Research Projects Agency NETwork) in 1969. ARPA was more interested in a reliable network than a blast-proof one; ARPANET had fewer redundant connections than Baran had suggested. ARPANET was a decentralised rather than a truly distributed network.

After a successful demonstration of ARPANET in 1972, representatives of networking projects from around the world realised that protocols would be necessary if their networks were to be interconnected, and formed the Internetwork Working Group (INWG).[9] The organisation was created in a mood of considerable enthusiasm in the afterglow of ARPANET's successful demonstration and its harmonious record of successful research collaboration. The nature of the INWG's task - to enable different computer networks to communicate with each other - encouraged the co-operation of its members. However, the emergence of rival standards for internetworking meant that researchers in the working group soon found themselves pulling in different directions.

(map of ARPANET in 1982)
ARPANET in July 1982[10]

At the heart of the Internet is a protocol, and its name is TCP/IP. It stands for 'Transmission Control Protocol / Internet Protocol'. TCP/IP was first proposed by Vinton Cerf and Robert Kahn in 1974, and adopted by ARPANET in 1982. Cerf, who was also the first chairman of the Internetwork Working Group, once said that "the history of the Internet is the history of its protocols."[11] That is not quite true, but its protocols are the key to its history. The Internet protocols allowed different computers, and more importantly, different networks of computers to 'talk' to each other.

The designers of TCP/IP and its predecessors thought of the term 'protocol' in terms of 'diplomats exchanging handshakes and information.'[12] The term is apt because the metaphor of protocols permeates the physical, organisational and social structure of the Internet.

protocol n. Original draft of diplomatic document; formal statement of transaction; rigid prescription or observance of precedence and deference to rank as in diplomatic and military services; official formula at beginning and end of charter, papal bull, etc. ~ v. Draw up protocols; record in protocol. [Gk protokollon fly-leaf glued to the case of a book (kolla glue)][13]
The word 'protocol' connects the structural basis of the Internet with concepts of diplomacy and communication; the stiff formality of the military, a way to behave. In its 'observance of precedence and deference to rank' the word relates the Internet to and acknowledges its antecedents in Western technological communications history and its military tradition.

TCP/IP was a protocol in the diplomatic sense because it allowed 'foreign' computers to 'shake hands'. In the 1960s and early seventies, even different computer models made by the same company, let alone those from different manufacturers or other countries were 'foreign' to each other.[14] The designers of TCP/IP were interested in networking, in computer-to-computer diplomacy; their protocol made no distinction between 'foreign' computers in or out of the United States. Work on the developing Internet protocols was 'international from the beginning'[15], involving the Internetwork Working Group and a node outside the United States (University College London). But international protocols imply international diplomacy, and international diplomacy implies dissent and disagreement as much as co-operation.

The technical information which TCP/IP carried was a protocol in another sense: an 'official formula at beginning and end of charter, papal bull, etc.' from the Greek protokollon. The Transmission Control Protocol converts messages into 'streams of packets' - each separately addressed - at the source, then reassembles them at the destination.[16] The Internet Protocol is responsible for transporting the packets - in any order - to the destination. The Internet Protocol stores the information about where a particular message packet is going, where it has come from, how many packets are in the sequence and where it 'fits' in that sequence[17] - just like the protokollon, fly-leaf of a book which holds information about a written document.

However, TCP/IP did not fit completely into the dictionary definition of a protocol. It was against protocol as a 'rigid prescription or observance of precedence and deference to rank as in diplomatic and military services.' The Internet Protocol (and its ARPANET predecessor, Network Control Protocol) observed no precedence or deference to rank, a fact which is subtly at odds with its military background. The Internet Protocol was designed by academics and young graduates with 'no authority'[18], not military personnel, and 'despite its military origins' it was (and is) 'not a classified network.'[19]

There is no hierarchy of control in the Internet. 'In principle,' writes Sterling, 'any node can speak as a peer to any other node, as long as it obeys the rules of TCP/IP protocols, which are strictly technical, not social or political.'[20] But the Internet always was a network of people, not just machines. The structural implications of the 'strictly technical' protocols had definite social and political consequences.

There were several technical aspects of the Internet protocols which were later reflected in the social structure of Internet culture. TCP/IP was technically advanced for its time, incorporating the latest satellites and computers, technologies not readily available to civilians in the early 1970s. The Internet protocol was also very adaptable to different computing environments, and very rugged. Most important of all, the central fact of TCP/IP was the decentralised peer-to-peer network it created. All of these features of the protocol were legacies of ARPA's military involvement in the internetworking project. Later, TCP/IP's reliability, easy adaptability to a wide range of systems (with 'implementations of it for almost every type of computer on the planet'[21]) and lack of hierarchy made it appealing for civilian use, and set it up for victory in the protocol wars.

The protocol wars between competing internetworking standards dragged on for a decade and a half. The competitors were the American TCP/IP and a European standard: OSI (Open Standards Interconnection). The 'wars' were bloodless, more like commercial conflicts than wars. Commercial interests in computer networks certainly increased during the 1980s. However, the protocol wars revealed an ugly undercurrent of international distrust and jealousy, and hampered unified research.

Rivalry between Europe and America is as old as the New World, and research, even that of the Internetwork Working Group was 'not initiated, organized, executed or applied in a social/political vacuum.'[22] European members of the INWG were sceptical of the TCP (the original TCP became TCP/IP a few years later) protocol proposed by Vinton Cerf and Robert Kahn in 1974. Although the TCP proposal was 'clearly in the mainstream of INWG ideas...[it] was viewed as just one more proposal.' Partly this was because TCP seemed 'unnecessarily costly'[23] of network computing resources, but some European members of the Internetworking Working Group were also reluctant to allow the Americans the credit for a world data-communication standard.

The French in particular were non-plussed at the idea of accepting ARPA's model for an interconnected network. There were several reasons for this. CYCLADES, a French packet-switched network of the 1970s used protocols which eventually became the basis for the Open Systems Interconnection model. So the French already had at least the beginnings of an alternative protocol system to TCP/IP. Like the French, some Japanese and European governmental representatives had a jealous, 'not invented here' attitude to the American protocol, but there were also financial reasons. 'Real money was involved; future profits were at stake,'[24] and the Europeans didn't want to let American business interests dominate their own.

There was another reason for European distrust of the American Internet proposal. Although the 'network of networks' protocol was a general goal of the Internetwork Working Group, it was unclear at the time what form such a protocol would take. The full implications of a decentralised network were not clear, and no- one knew whether the network of the future would be 'a unity, a federation, or a fragmentation.' The Internetwork Working Group aimed for the middle option, a 'more or less coherent system of communicating networks'[25], but details were vague as to how 'all parts, somehow would be interconnected'[26] in a way that would not create an unworkable global, monolithic system. This sort of thinking, which ignored the actual structure and operation of TCP/IP, persisted in some quarters for most of the 1980s.

Although there was probably some anti-American feeling among researchers of the INWG, it is too simplistic to blame that sentiment alone for the divergence of United States and European research into networking protocols. In 1976, Internetwork Working Group members from French, British and United States networks proposed an 'International End to End Protocol', but the United States military had become very interested in TCP/IP's development and was not particularly interested in changing tack. So European researchers, with the International Standards Organisation, started to develop their own protocol.

The OSI protocol had some problems relative to TCP/IP. Or, to put it another way, TCP/IP had several advantages over OSI, which helped it to triumph in the protocol wars. Open Systems Interconnection was a protocol designed for civilian use, but, paradoxically, it was a much more hierarchical system than the decentralised TCP/IP. Unlike the Internet protocol, OSI was not particularly adaptable to existing networks and had not been practically tested.

The reasons for OSI's failure to compete with TCP/IP were not purely technical. International rivalries were also involved, not only between Europe and America, but also between European countries. Rivalries such as those existing between the British and French extended back centuries; the bureaucratised, pasted- over camaraderie of the European Community did little to relieve such ancient mutual antagonism. Shared history and inequalities of wealth and political influence among member states contributed to their distrust of one another. Even while they were working together, countries of the EC were pulling apart. An elaborate bureaucratic system helped to restrain this tide of divergence in Europe, but such bureaucratic structures also delayed and complicated the specifications for the Open Systems Interconnection protocol.

Even while praising its fortitude, British computer scientists R.T. Braden and R. H. Cole bemoaned the lack of 'hierarchy' in the Internet Protocol.[27] They saw the lack as a deficiency when in fact the peer-to-peer nature of TCP/IP was the key to its success and that of the Internet culture which it enabled. Such perceptions are an example of how 'information technology will impact differently on nations that differ in social tradition, economic and political regime,'[28] and perhaps another reason why the 'vulgarly democratic' American models were not embraced by Europeans.

The Open Systems Interconnection protocol divided networks into layers of operation, with a specified protocol for each layer.[29] Unlike the Internet Protocol, this layered, limited system certainly did follow the dictionary-definition of a protocol: a 'rigid prescription or observance of precedence and deference to rank as in diplomatic and military services'. Just as these layers and hierarchies of operation clashed with the non-hierarchical Internet Protocol, the committees who defined OSI's layers differed from the United States' INWG members in their working practices.

The informal, technical free-for-all attitude of the Americans contrasted sharply with the bureaucratised European committees. Each of the seven layers of OSI had its own committee, each of which, in the words of American "Protocol Warrior" M.A. Padlipsky, was 'in danger of becoming an expansionist fief.'[30] United States network researchers found the politics of ISO network design distressing. David Piscitello a U.S. INWG member who originally supported OSI, wearied of 'wading through the obligatory political purgatory associated with each incremental change'[31], and quit in frustration.

Rather than attempting to connect a bizarre profusion of different networks, as the Internet protocol did, OSI was supposed to be a standard which, endorsed by the International Standards Organisation, would be gradually accepted and implemented by networks and manufacturers. Even the United States was affected by the magic acronym "ISO". From 1988 until 1994, the U.S. Department of Defense viewed TCP/IP as a temporary protocol, pending adoption of OSI - to the frustration of champions of the Internet Protocol like Padlipsky. However, even by 1988, TCP/IP had the numbers and the inertia to defeat OSI, whatever the DoD said.

In spite of its 'elegance'[32] on paper , OSI did not mesh well with many existing networks. Often, existing networks did not conform to the ISO's 'tidy model' which specified a narrow range of hardware and protocols for use in the system.[33] Internet protocol was a lot 'messier' than OSI - but it worked. Computer and networking technology advanced rapidly in the late seventies and eighties, so the International Standards Organisation often revised standards for the OSI protocols. Because OSI 'specified before implementation...specification took forever and implementation never really happened...IP won the race.'[34]

While they were waiting for the OSI protocol to be perfected and adopted, many European researchers, especially UNIX[35] users, found that their networks were already compatible with TCP/IP, even before the protocol was officially adopted by ARPANET in 1983. Another factor which affected Europe's choice of internetworking protocol was that many of the emerging Eastern European nations chose to use TCP/IP, in spite of 'governmental pressures' to use OSI.[36] With a hodge-podge of networks and cheap machines, Eastern Europe found the ISO's standards impossible to meet, and the cheap (TCP/IP, although American, is 'considered an open, non- proprietary protocol'[37]), rugged Internet must have been an appealing alternative. OSI was never used in Japan, whose first network (JUNET - Japanese UNIX Network), like many others, 'fell into the digital embrace of the Internet, and messily adhered.'[38] Open Systems Interconnection was a model of an ideal network, but the Internet Protocol dealt directly with the 'real world' in which there was no such thing.

Throughout the 1980s the Internet grew exponentially. Its 'solid protocols on a flexible operating system at low cost' may have been 'an unbeatable combination', but still the protocol wars didn't completely end (with Internet victorious), until late 1994, after wasting much time and energy on competing standards.[39] Meanwhile, as more and more people and computers connected to the Internet, the dwindling protocol wars became less and less relevant to the emerging network society. The structure of that society, although never static, reflected that of the protocol which enabled it to exist: uncontrolled (but not quite), American (but not quite), and almost indestructible.

Because and in spite of military influence on the Internet Protocol, it has no hierarchy. This is the paradox of the Internet: its protocols made it enduring, as required for military use, but also uncontrollable, which was perfect for private civilians. The Internet's military-imparted advantages allowed it to defeat its only serious rival and become a network society which reflected its protocols, its essential peer-to-peer nature. The Internet's decentralised networking structure conflicted with OSI's model of bureaucracy and control, but that conflict was also reflected within the developing net society. As the Protocol Wars continued, various groups attempted to control the Internet.


Footnotes

  1. Hardy, 'The History of the Net', [http://www.ocean.ic.net/ftp/doc/nethist.html], 1993.
  2. Tracy LaQuey, The Internet Companion: A Beginner's Guide to Global Networking, 2nd ed., 1994, p. 32.
  3. Daniel Czitrom, 'Lightning Lines', in Crowley and Heyer (eds.), Communication in History: Technology, Culture, Society, 2nd ed., 1995, p. 128.
  4. James Carey, 'Time, Space, and the Telegraph', in Crowley and Heyer (eds.), Communication in History, 1995, p. 132-3.
  5. Bruce Sterling [bruces@well.sf.ca.us], 'Short History of the Internet', from Magazine of Fantasy and Science-Fiction, [gopher://gopher.isoc.org:70/00/internet/history/short.history.of.internet], February 1993.
  6. Hafner and Lyon, Where Wizards Stay Up Late, 1996, p. 54.
  7. Sterling, 'Short History of the Internet', [gopher://gopher.isoc.org:70/00/internet/history/short.history.of.internet], February 1993.
  8. Baran's original term for message packets was "message block". "Packet" was coined by English physicist Donald Davies who invented the packet-switching concept independently of Baran in 1965. See Hafner and Lyon, Where Wizards Stay Up Late, 1996, pp. 64-67.
  9. Alexander A. McKenzie and David C. Walden, 'ARPANET, The Defense Data Network and Internet' in Fritz E. Froelich (ed.), The Froelich/Kent Encyclopedia of Telecommunications, v.1, 1991. p.361.
  10. from McKenzie and Walden, 'ARPANET, The Defense Data Network and Internet' in Froelich (ed.), The Froelich/Kent Encyclopedia of Telecommunications, v.1, 1991. p.368.
  11. Vinton Cerf, quoted in Salus, Casting the Net, 1995, p. 131.
  12. Salus, Casting the Net, 1995, p. 42.
  13. from Oxford Illustrated Dictionary, v. 2, 1962, p. 710.
  14. McKenzie and Walden, 'ARPANET, The Defense Data Network and Internet' in Froelich (ed.), The Froelich/Kent Encyclopedia of Telecommunications, v.1, 1991. p.341.
  15. Vinton Cerf, 'How the Internet Came to Be' as told to Bernard Aboba in Aboba, The Online User's Encyclopedia, 1993, p.530.
  16. Sterling, 'Short History of the Internet', [gopher://gopher.isoc.org:70/00/internet/history/short.history.of.internet], February 1993.
  17. Jonathan Postel, 'Internet Protocol', RFC 0791, [NIS.NSF.NET/internet/documents/rfc0791.txt], 1 September, 1981.
  18. Cerf, 'How the Internet Came to Be' in Aboba, The Online User's Encyclopedia, 1993, p.528.
  19. LaQuey, The Internet Companion, 2nd ed., 1994, p. 140.
  20. Sterling, 'Short History of the Internet', [gopher://gopher.isoc.org:70/00/internet/history/short.history.of.internet], February 1993.
  21. LaQuey, The Internet Companion, 2nd ed., 1994, p. 27
  22. James D. Halloran, 'International Democratization of Communication: The Challenge for Research' in Jörg Becker, Göran Hedebro and Leena Paldàn (eds.), Communication and Domination: Essays to honor Herbert I. Schiller, 1988, p. 241.
  23. McKenzie and Walden, 'ARPANET, The Defense Data Network and Internet' in Froelich (ed.), The Froelich/Kent Encyclopedia of Telecommunications, v.1, 1991. p. 362.
  24. Salus, Casting the Net, 1995, pp. 87, 123-4.
  25. J. C. R. Licklider and Albert Vezza, 'Applications of Information Technology', proceedings of the IEEE 66(11), 1978, p. 1342, quoted in Hardy, 'The History of the Net', [http://www.ocean.ic.net/ftp/doc/nethist.html], 1993.
  26. D. L. A. Barber, 'Human Factors in Open Systems Interconnection' in M. B. Williams (ed.), Pathways to the Information Society: Proceedings of the Sixth International Conference on Computer Communication, London 7-10 September 1982, 1982, p. 824.
  27. R. T. Braden and R. H. Cole, 'Some Problems in the Interconnection of Computer Networks', in Williams (ed.), Pathways to the Information Society, 1982, pp. 969-974.
  28. F. J. M., Laver, 'Computers + Communications + People = ?' in Williams (ed.), Pathways to the Information Society, 1982, p. 3
  29. J. Postel, C. Sunshine, D. Cohen, 'Recent Developments in the DARPA Internet program' in Williams (ed.), Pathways to the Information Society, 1982, p. 976.
  30. M. Padlipsky, 'The Illusion of Vendor Support', RFC 0873, [NIS.NSF.NET/internet/documents/rfc0873.txt], 1 September 1982, p. 6.
  31. David M. Piscitello (IETF OSI Area Co-director) 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, p. 28.
  32. The 1984 ISO OSI specification (RFC 0905)is 154 pages of legalistic text. The Internet Protocol specification (RFC 0791) is less than a third that length.
  33. A. McKenzie, 'ISO Transport Protocol Specification ISO DP 8073', RFC 0905, [NIS.NSF.NET/internet/documents/rfc0905.txt], 1 April, 1984.
  34. John Quarterman, quoted in Salus, Casting the Net, 1995, pp. 110, 123.
  35. UNIX was a popular operating system for university computers, developed by Bell Labs in 1972 and later rewritten to incorporate IP. More about UNIX in chapter 2.
  36. Salus, Casting the Net, 1995, pp. 190, 226.
  37. LaQuey, The Internet Companion, 2nd ed., 1994, p. 27.
  38. Sterling, 'Short History of the Internet', [gopher://gopher.isoc.org:70/00/internet/history/short.history.of.internet], February 1993.
  39. Salus, Casting the Net, 1995, pp. 126, 131.

Where to next?