Wiring the world

Even before the US military funded the experimental network that would become the internet, the vision of what it would one day become was perfectly clear.
1 November 1999
Vint Cerf plays Spacewar on a PDP-1 in May 2007.

The notion of computing as a utility, or a global network of linked information, may seem thoroughly modern. Both, however, predate the birth of what would become the internet. Even more curiously, it was a totally unrelated event that changed the course of history, and made these theoretical dreams reality.

The idea of hypertext and collaborative content creation, complete with procedures for copyright management and royalty payment, existed as early as 1965, several years before the recognised birth of the internet. Known as Xanadu, it was described and developed by Ted Nelson. It would allow anyone to publish digital content, would permit reuse of existing content, and create both links to other content, and "transclusion" links designed to include text, quotations or references located elsewhere.

It was a grand and comprehensive vision, which would prove far too ambitious. By the time it was implemented, almost four decades later, the world had moved on in its more practical, less elegant, piecemeal fashion, and the Xanadu system had become an anachronism - the original vapourware. Xanadu wasn`t why the internet was created. The internet that would eventually realise something approximating Nelson`s vision was born not only at the same time, but for the same reason as the first manned moon landing.

That reason was a small object, less than 60cm in diameter and weighing as much as an adult male, which was sprung on an astonished world during the International Geophysical Year of 1957/8. Known as Sputnik 1, most people merely saw a silvery ball with antennae, which emitted meaningless beeps from space before running out of batteries and burning up in the atmosphere.

Sputnik`s signals carried information, however, and a surprising amount of it. Superficially, the dual-frequency oscillation signalled temperature and pressure conditions inside the craft, matters that now appear trivial, but which were of considerable interest to the scientists and engineers who were the first to launch an object into outer space. A layer higher, the radio signal was used to measure certain features of the atmosphere. And at a layer the US Department of Defence understood, it signalled that the USSR had significant military capability that could threaten the security of the USA and its Nato allies.

Besides causing John F. Kennedy`s famous exhortation to put a man on the moon before the 1960s were out, the Soviet satellite prompted the US Defence Department to create an organisation to fund "high risk, high reward" research, which it named the Advanced Research Projects Agency (ARPA).

The galactic network

Computer systems were already well established by the 1960s, but several new research avenues were being explored and would benefit from ARPA funding.

In 1961, Leonard Kleinrock submitted a PhD thesis to the department of electrical engineering at MIT on the dull-sounding subject of Information Flow in Large Communication Nets. This was a mathematical generalisation of a problem that had long been studied, namely how to predict congestion in telephone networks.

Interactive co-operation between man and computer... could go beyond mere automation to guide problem formulation. JCR Licklider
At the time, however, this research was novel. The behaviour of multi-node networks was well known, and single-node queuing systems with storage had also been studied, but the two had never been combined into a single theory. Kleinrock`s thesis examined the case of a multi-node network, in which each node could store messages while they waited for transmission channels to another node to become available.

His work on how to predict and manage the flow of messages in a network became the basis for what would eventually become the internet.

During that decade, another influential thinker concerned himself with computer science. JCR `Lick` Licklider had written a paper in 1960 on the "symbiosis" between man and computer, in which he observed that 85 percent of typical "thinking" time was actually spent searching for information, transforming or charting it and performing other clerical tasks. All too often, he noted, the selection of problems to tackle was determined not by the difficulty of the problem itself, but by the amount of non-intellectual drudgery it might involve. He also noted that simply using a computer to solve a problem implied that you already knew what the problem really was. Interactive co-operation between man and computer would, in his view, be a more effective use of computing power, since it could go beyond mere automation to guide problem formulation.

"It seems reasonable to envision," he wrote, "for a time ten or 15 years hence, a `thinking center` that will incorporate the functions of present-day libraries together with anticipated advances in information storage and retrieval, and the symbiotic functions suggested earlier in this paper. The picture readily enlarges itself into a network of such centers, connected to one another by wide-band communication lines and to individual users by leased-wire services. In such a system, the speed of the computers would be balanced, and the cost of the gigantic memories and the sophisticated programs would be divided by the number of users."

Licklider believed that for his symbiotic ideal to be efficient, the superior speed of computers could better be applied in a distributed, time-sharing manner. Although widely criticised by computer makers and users, who thought time-sharing to be inefficient and complex, Licklider`s view proved to be very influential. A key reason was his appointment in 1962 to ARPA, to lead its research funding programme. His work there was instrumental in the establishment of the first doctorate programmes in computer science, at Carnegie Mellon, UC Berkeley, MIT and Stanford.

A number of technical and research developments throughout the 1960s contributed to the development of Licklider`s vision. This included seminal work on distributed computing and packet switching by Kleinrock, Paul Baran of the Rand Corporation, and others. In the commercial world, the popular System/360 mainframe by IBM and the far less expensive PDP-8 minicomputer by DEC were making computers more widely available, and increased the desire to connect them together.

The first project to connect a lot of computers together was Sabre, the airline reservation system developed by IBM and introduced in 1964. It connected 2 000 terminals in 60 cities to a central transaction processing system using telephone lines. It wasn`t distributed computing, but it did prove that telephone lines could be used for data communication.

Interfaith message

Licklider, meanwhile, continued working on his vision of a "Galactic Network", and contracted to various universities and industry firms such as Bolt Beranek and Newman (BB&N). He arranged an ARPA grant for the research laboratory at MIT, known as "Project MAC", at which famous names like Marvin Minsky were already discussing computing as a utility, much like electricity, being available anywhere on demand. It was at MIT that Multics, a multi-user time-sharing operating system, was developed.

By 1965, three computers - one at MIT, one at System Development Corporation in California, and one at ARPA itself - were connected, without packet-switching routers, to form "The Experimental Network", and one of the MIT researchers, Lawrence Roberts, drafted a paper that would lay the foundation for a network that would become known as the ARPAnet.

The term `packet`, which is perhaps the most fundamental building block of the internet, was not coined by any of these people, however. It was first used by an Englishman, Donald Watts Davies, who, in 1967, oversaw the design of a packet network at the National Physical Laboratory in Middlesex. This group would join forces with ARPA and its pet universities, as well as the Rand Corporation, which was independently working on packet networks.

A year later, Licklider wrote a piece published in Science and Technology, in which he described a future in which man would "communicate more effectively through a machine than face to face". As if to give impetus to this future, ARPA issued contracts for the design of the first "Interface Message Processors" to BB&N. These were the forerunners of what we could call routers today. The Cold War was raging and ARPA was very interested in the idea of a network that would remain reliable even if one or more nodes failed.

US senator Edward `Ted` Kennedy, rushed to provide an early demonstration of government cluelessness about the internet, and anticipating Al Gore`s role in bringing a new religion to the people, sent a congratulatory telegram to BB&N commending it for its ecumenical efforts in developing "Interfaith Message Processors".

The first node of the ARPAnet went live at the University of California, Los Angeles, on 30 August 1969, when an SDS Sigma 7 machine was connected to an IMP. On 1 October, a TX2 computer at Stanford became the second node, and on 29 October, the first packets were sent across the network by Charley Kline at UCLA. The system crashed.

By the end of the year, however, computers at the UC Santa Barbara and the University of Utah were connected, and a four-node embryo of the internet was live. A group of researchers, including Vint Cerf, could begin work on defining the communication protocols for the ARPAnet.

Dream realised

One of the first applications written for the emerging internet was an e-mail program, designed by Ray Tomlinson at BB&N, who was the first to select the @ character to link user names with host addresses.

Another very early internet project, which survives to this day, was Project Gutenberg. Started by Michael Hart, its goal was to make copyright-free works available in a universally accessible, easily replicable digital form. The first text in the collection was the United States Declaration of Independence, mostly because it fit on the very limited storage of the day. Today, it contains some 25 000 books, scanned, edited and formatted by a global network of volunteer collaborators. Project Gutenberg, in many ways, pioneered the use of the internet as envisaged by dreamers like Ted Nelson and theoreticians like JCR Licklider.

In 1972, Cerf was appointed chairman of the International Network Working Group, which would later become the International Federation for Information Processing. A year later, the ARPAnet had an estimated 2 000 users, and e-mail comprised 75 percent of all traffic.

Ray Tomlinson... was the first to select the @ character to link user names with host addresses.
The process by which internet standards and protocols were developed was known as the RFC, or request for comment. Most had a useful purpose. RFC 1, for example, issued in 1969 by Steve Crocker at UCLA, defines the protocol for communicating between two hosts, each of which was connected to an IMP.

Others were prescient in the extreme. In 1973, for example, the famous RFC 602 was published by Bob Metcalfe, who would go on to invent the Ethernet network. In the wake of several system crashes, it raised the alarm about the lack of security inherent in ARPAnet systems, and the "lingering affection for the challenge of breaking someone`s system [which] lingers despite the fact that everyone knows it`s easy to break systems, even easier to crash them". Although it described security concerns, it is primarily remembered today for its title: "The stockings were hung by the chimney with care". It ends: "You are advised not to sit `in hope that Saint Nicholas would soon be there`."

Even more whimsical was RFC 527, posted by R Merryman at the University of California at San Diego, merely entitled ARPAWOCKY (see sidebar).

Cerf sums up the era with another RFC classic, number 968.

A network for everyone

The publication by Cerf and Bob Kahn of A Protocol for Packet Network Interconnection, in 1974, would have repercussions that still echo today. It detailed the transmission control protocol (TCP). It would eventually be separated into TCP (level 2) and IP (level 3) protocols, and it wasn`t until 1983 before the final switchover from the original ARPAnet protocol (NCP) took place, but today the internet still uses the TCP/IP protocol.

The ARPAnet was becoming ever more widely used, and in the mid-1970s, BB&N launched Telenet, the first-ever commercial version of the ARPAnet. A great deal of the early internet geek culture, so evident in RFC 527`s appropriation of Lewis Carroll, was established in early mailing lists like MsgGroup and SF-Lovers. Perhaps the most definitive record of the language and culture of the era, drawing together slang and conventions from MIT, BB&N, Carnegie Mellon and other early users of the ARPAnet, was found in the Jargon File, first published in 1975 by Raphael Finkel at the Stanford Artificial Intelligence Laboratory. It remains a fascinating read and is today edited by open source advocate Eric S Raymond.

Usenet, a protocol for posting to and accessing newsgroups, as well as the first multiplayer online game, or MUD, both happened in 1979. The internet would never be the same again. To this day, online forums and games follow the principles laid down by these pioneering systems.

Throughout the 1980s, new developments brought the internet closer to prime time. Domain name servers were introduced in 1984, as the number of internet hosts went through the 1 000 mark, and a year later, the first (and arguably most influential) "virtual community" of the internet era, the Whole Earth `Lectronic Link (WELL), was established. See sidebar.

The internet was exploding. In 1987, 10 000 hosts were live. A year later, a worm laid low ten percent of the 60 000 hosts, but the internet`s growth was unstoppable. By 1989, the number was 100 000, and in 1990, the ARPAnet was formally decommissioned.

Twas the decade `fore prime time, and all through the net, hackers waited with bated breath. The 1990s were coming and with it they`d bring CERN and Mosaic, and that WWW thing.