On 9 January 2024, Archives of IT (AIT) is hosting the first of an annual series of Forums on the Histories of the Internet where we will review some of the major innovations and what the future holds for this revolutionary communications technology.
At the Forum you can hear and participate in discussions on a wide range of issues, about the technology, how it developed, its impact on society and where we might go with it. This article charts some of the history and current developments that form the backdrop to that event.
Looking through our archive we can map many of the internet’s significant developments from initial ideas to tests as well as its commercialisation, initial uses and benefits and risks to society.
Pioneers Vint Cerf, Peter Kirstein and Jim Norton have all been recorded by AIT about their work on packet switching and the emergence of the internet and A Michael Noll on its commercialisation. We also have interviews with internet entrepreneur Eva Pascoe, who encapsulates the pioneering spirit of the first internet teachers and users and with Professor Bill Dutton, who was the founding director of the Oxford Internet Institute, a research centre focusing on the societal opportunities and challenges posed by the internet.
It was 50 years ago that the first international network connection linked the US with Europe and heralded the dawn of the internet, which according to the Internet Society (founded by Vint Cerf in 1992 to promote its development) went on to ‘revolutionise the computer and communications world’.
Now the internet is a global network of computers that can share information and is the largest computer network in the world. There are 5.3 billion users which is almost 66% of the world’s population.
Internet pioneer, Vint Cerf, who gave an online talk for AIT in January 2022 entitled: Internet, The Early Days, described it in simple terms as a packet switching service where the nodes of each network switch packets ‘rather like postcards in the postal system but millions of times faster’.
He also made a distinction between the World Wide Web by saying: “The internet is the underlying structure that supports the World Wide Web and the World Wide Web is this substantial application space which almost everything gets done these days.”
Cerf is known as one of ‘the fathers of the internet’ and as a co-designer of the architecture of the internet for his work with colleague, Robert ‘Bob’ Kahn. The pair were awarded the prestigious A M Turing Award in 2004 for their ‘pioneering work on the internet including the design and implementation of the internet’s basic communications protocols, TCP/IP, and for inspired leadership in networking’.
Paul Baron and message blocks
When talking about the origins of the internet, Cerf firstly mentions Paul Baron, who in 1962 was working at US think tank, The Rand Corporation, and had “speculated about message blocks and their use in digitised speech distribution in a post-nuclear scenario and had the idea of a proliferated network,” says Cerf.
At the time the Cold War between the US, its allies and Russia was in full swing, the proxy war in Vietnam was heating up and the Cuban Missile Crisis had brought the world to the brink of a full-scale global conflict, so Baron’s ‘post-nuclear scenario’ wasn’t too far from people’s fears.
Also in 1962, Leonard Kleinrock at the Massachusetts Institute of Technology (MIT) completed his dissertation: Information Flow in Large Communication Nets, in which he said the two theoretic models of message switching and what would come to be known as packet switching were mathematically the same.
Donald Davies, Packet Switching and ARPANET
Baron’s work was known to Welsh computer scientist, Donald Davies, manager of the Advanced Computer Techniques Project at the National Physical Laboratory (NPL) in the suburb of Teddington, south west London. Davies subsequently began the exploration of and coined the term packet switching in the mid-1960s and by 1966 was able to build a one node network, one of the first area networks using packet switching as the primary communication technology.
In the same year work began on ARPANET (The Advanced Research Projects Agency Network) and it was funded by, and gets its name from, ARPA (The Advanced Research Projects Agency) an arm of the US Defense Department.
It was initiated, according to Cerf, by a frustrated head of the information processing techniques office at ARPA, Robert Taylor, who found he had three terminals sitting in his office: one to Systems Development Corporation Q32 (a planned military computer) in Santa Monica, one to get to Project Genie (a time-sharing system) at the University of California, Berkley, and another to Multics (an influential time-sharing operating system) at MIT.
“And he said ‘why can’t I have a network so I only have to have one terminal?’”, says Cerf. “The problem then was getting a whole bunch of computers to communicate with each other, despite the fact they had different operating systems, word lengths and other parameters, and figuring out what standards to use. Then [it would be possible] for a single terminal to talk to all three or more computers through some kind of network.”
Larry Roberts and Bolt, Baraneck and Newman build the ARPANET
In 1967 Larry Roberts, from MIT, was recruited to build the network and the implementation of the packet switching ARPANET was overseen by Bolt, Baranek and Newman (BBN), a research company from Cambridge, Massachusetts, which included Bob Khan and JCR Licklider, the latter of which conceived the concept of an ‘Intergalactic Computer Network’ four years earlier.
In October 1967, Donald Davies’s research had a direct impact on the ARPANET project through his colleague, Roger Scantlebury, who attended the first Association for Computing Machinery Symposium on Operating Systems Principles in Tennessee, where he presented the NPL’s results on packet switching for data communications.
Roberts was at the conference and he presented a proposal for the ‘ARPA net’, which incorporated the packet switching concepts into the ARPANET design.
At the conference Scantlebury met Roberts and impressed the importance of using high speed links ????? for reducing transmission delay. Roberts had previously thought that 2,400 bits a second would be all that was needed but Scantlebury convinced him to go to the maximum he could get, which at the time in the US was 50 kilobits a second by binding a dozen analogue channels together using a Bell 303 modem.
The first packet-switched computer network
In January 1969 the first computers were connected between Stanford University and The University of California Los Angeles (UCLA) through a public packet-switched computer network. By the end of the year more connections were made at The University of California Santa Barbara and the University of Utah.
“It also turned out that my colleague, Bob Khan was very much involved in the ARPANET design at BBN but in late 1972 he moved to DARPA (Defense Advanced Research Projects Agency) and started the internet project there along with several other networking projects, all of which were stimulated by the idea that we could use computers in command and control.
“So, the success of the ARPANET led to the defence department’s decision to invest in computer networking for command and control. In order for that to work the computers would end up being on ships at sea, mobile vehicles and aircraft. And all we’d done at the Arpanet was build fixed installations connected by dedicated telephone circuits.
The Internet Problem
“Bob showed up in my office at Stanford University in the spring of 1973 telling me that we had a problem, and the problem was how to interconnect mobile packet switching. And he said in order to use computers and command and control we’re going to have to use mobile, radio and satellite networks to connect to ships at sea, ship to ship and ship to shore.
“He had three different networks; the ARPANET, mobile packet radio net and the packet satellite net, and they all had different characteristics, error rates, delays, packet formats and structures. And he says: ‘how we gonna pull all those together in a uniformed way?’. That was the internet problem and by September 1973 we had concluded a particular design, which we briefed to a Nato-sponsored meeting at the University of Sussex and called it the Transmission Control Protocol/Internet Protocol (TCP/IP).
“We continued working with a collection of colleagues in the International Working Group, which I ended up chairing, founded in October 1972 in the course of an international computer communications conference in Washington DC where the Arpanet was demonstrated and which many of our colleagues from Europe participated.”
International connections and protocols
By June 1973 an ARPANET tip, capable of supporting terminals, was installed at UCL and connected by a landline up to the Norwegian Defence Research Establishment, which housed the Norwegian Seismic Array sensor system that was tracking underground nuclear tests by the Russians to make sure they didn’t exceed the test treaty constraints (The Limited Test Ban Treaty, signed by the US, UK and Russia in 1963).
In the course of that work British computer scientist, Peter Kirstein’s group at University College London (UCL) became deeply involved in thinking about the implementation of TCP/IP.
“From January to December of 1974 my research group at Stanford developed the detailed specification for the TCP protocols,” says Cerf. “In January 1975 implementations were undertaken at UCL and BBN.”
Peter Kirstein CBE – research rebel and father of the European Internet
Together with Vint Cerf and Bob Kahn, Kirstein, interviewed by AIT in 2019, began research on the internet from 1975 onwards and is known as the father of the European Internet.
However, Kirstein was ordered to stop working on the internet, an order he ignored. “The British did not approve of the ARPANET, because they regarded that as an experiment, and although the academics liked it, and they tolerated it, they certainly didn’t regard it as mainstream, and, were perfectly happy to have me connecting in the ARPANET, but they didn’t want people working on the internet side of things.
“I was actually ordered to stop working on the Internet Protocols. I said I was sorry but my links to both the Ministry of Defence and the US were far too important, whether they funded me or not, was their problem, but I was not going to stop working on it.”
As a result, by the time the ARPANET was moving into the internet, Kirstein was the only person working on it in the UK.
“This is presumably why I became known as the father of the internet, because in fact, we adopted the Internet Protocols in 1982, before the ARPANET did in general, not because I wanted to be a pioneer, but because my poor little PDP-9 computers were running out of power, and the internet protocols were already working on the PDP-11s which were their successors, and I had lots of them.
“By adopting the Internet Protocols for my services early, I could go on providing the service even though my computers were giving up the ghost. As a result, I experienced all these early problems, before the US people had to do it in earnest.”
Testing the ‘internet’ from Malvern to Boston and San Francisco
As well as running Internet Protocols running experimentally as part of the research project, Kirstein was involved in numerous other projects including SATNET, which used satellites in a multi-destination way and a US project called PRNET, a packet radio net, which he describes as mobile radio long before current mobile telephones.
“We were actually sending files from the Royal Signals and Radar Establishment in Malvern, through me, with the US protocols, then over SATNET to Boston, over the ARPANET to Stanford Research Institute, and then by packet radio to a truck going across the Bay Bridge [San Francisco], and when it started going over the bridge, the communication failed of course, because of the shielding from the structure, but when it got to the other end it continued. It sent the whole file without any errors and restarted itself. This was a really momentous showing how successful the internet protocols could be, how rugged they could be.”
“So, we were running the Internet Protocols, and we were running the internet from our side. I can never decide whether I did good service or bad service as a result, because, through doing that, the British never really had to worry about the internet until about 1987-89 because they could go on using the Internet Protocols but as far they were concerned they were using the Open System Interconnection (OSI) protocols, and I did all the work of translating between the OSI and the Internet Protocols.
“But of course, eventually, particularly with local area networks coming in, it became clear that the only solution was the internet. Vint and I already had a paper from about ten years earlier in 1978, where he and I had disagreed about what one does about it. I talked about adaption from one to another; he was visionary and talked about everybody adopting the Internet Protocols. For the first 15 years I was right, and then he completely demolished me.”
Simultaneous with the work that was going on at NPL and the early work in the US on TCP/IP, there were a set of coloured book protocols being developed by teams in the UK including at UCL. “This meant we had multiple packet switching implementations underway in the same general timeframe (late 1960s to mid-1970s),” says Cerf.
Among those were the X25 protocols that were standardised by the International Telegraph and Telephone Consultative Committee (now the International Telecommunications Union). That group consisted of a team of people from four different network activities: Datapac – Canada, Transpac – France, Telenet – US, and EPSS (Experimental Packet Switched Service) – UK, who also worked on the subsequent X27 protocols.
First packet switches in the UK
Professor Jim Norton, interviewed by AIT in 2017, a Fellow of the UK Royal Academy of Engineering, had just graduated from Sheffield University with the best first-class honours engineering degree when he went to work at the British Post Office in 1973. And his first posting was as the clerk of works resident at Ferranti Computer Systems in Wythenshawe, looking after the (EPSS) project.
“The Post Office wanted me to be the Clerk of Works on a project nobody believed would be of any importance whatsoever, which was the development of the first packet switches in the UK. To just go away and develop what would be the world’s first in the automated compliance, and a load testing system for the packet switch networks which underpin today’s the global internet.”
Norton did his main job during the day, then went back to the Ferranti plant at night, and wrote a series of programs to turn their packet switches into generators and sinks of packets so he could test them, which had never been done before. Being an engineer, he got the manuals for the computer he was using, and looked at what it could do, and wrote an entire real-time operating system, largely in assembler, for the Ferranti Argus 700, which ran roughly ten times faster than the one that Ferranti used.
EPSS was the first public data network in the UK when it began operating in 1977.
TCP/IP becomes the dominant protocol
Cerf says the X25 protocol was a well-developed and well-deployed packet switching implementation in commercial use very early on in the mid-1970s and ‘80s and heavily used by the financial sector in particular.
However, there were other interests in addition to the formative work that was done at NPL that influenced the ARPANET. In the late 1970s there was an initiative in the International Standards Organisation called the Open Systems Interconnection (OSI). The OSI protocols developed a seven-layer structure and was a competitor to the TCP/IP protocols.
From 1978 these two protocols were the primary contenders for global standardisation.
“I would pick 1993 as the year when TCP/IP seemed to become the dominant choice,” says Cerf. “That’s because I had been elected president of the Internet Society and had gone to the American National Institute of Standards and Technology saying the competition had been going on for 15 years and should be resolved. I asked them to evaluate the two.
“They came back and said that because TCP/IP implementations were more widely available than the OSI protocols they would agree that the US Government should make use of TCP/IP in lieu of OSI, which it had adopted as its preferred protocol in the defence department and other government departments.”
Telenet and the commercialisation of the internet
In the early 1970s Dr A Michael Noll, interviewed by AIT in 2022, was on the team of the President’s Science Advisor at the White House and later worked at AT&T identifying opportunities for new products and services.
He says the big question following ARPANET’s success was what would follow and how to commercialise it.
“My little role in that history was via a call from Dick Bolt of Bolt Beranek and Newman in 1975,” he says. “The science community, the national science people, the academics all wanted access to the ARPANET, but they weren’t part of that defence department constellation of people. So, ARPA was going to sell access which would have made the government a common carrier and that would have been opposed to the Nixon administration’s idea of a private industry.
“So, the issue was what to do about that and I remember there was a meeting we had with the White House Office of Telecom Policy and the ARPA people.”
The result neither AT&T nor Bolt Baranek and Newman wanted to get involved with packet switching, however, Dick Bolt told Noll that they were starting a company called Telenet to carry packet switching. Telenet was an American commercial packet-switched network which went into service in 1975 as the first public data network.
“That’s how Telenet was started and that was the movement that took it away from ARPA and made it something available for the whole world.”
This led to a number of other networks, mainly academic, coming online such as Bitnet and the National Science Foundation’s CSNET both in 1981 and NSFNET in 1986.
The World Wide Web
It wasn’t until 1989, however, that the universal potential of the internet became a reality. At the scientific research establishment, CERN, in Geneva, Tim Berners-Lee had invented the World Wide Web and started developing Hypertext Markup Language (HTML).
He originally intended only to network the CERN computers to share information and data within the organisation but to do so, he created the first web server which led to the launch of the World Wide Web at CERN in 1991.
Thirty years ago in 1993, Tim Berners-Lee made the source code freely available, and that combined with the growing global communications network, the Internet, started an information revolution that, by the end of the century, would change many aspects of our lives at home and work.
It was also 30 years ago that the Mosaic web browser was launched, opening up the web to non-academic users who were able create their own HTML web pages. Consequently, the number of websites grew from 130 in 1993 to over 100,000 at the start of 1996.
Cyber cafes popularise the internet
Eva Pascoe, interviewed by AIT in 2019, has been at the forefront of the commercialisation of the internet since the 1990s, contributing significantly to retail practice, public policy, training and education. She is currently the chair of Cybersalon.org, a non-profit digital think tank she co-founded in 1997, and has pioneered women’s participation in online business, online secure payments, e-commerce fashion solutions, and electronic customer relationship management.
In September 1994, Pascoe, together with her partners, David Rowe, Keith Teare and Gene Teare, opened Cyberia Café on Whitfield Street in Fitzrovia, as London’s and the UK’s first internet cafe. It provided desktop computers with full internet access.
“The tools we had when we opened Cyberia were extremely limited, because the speed of the internet at that time relied on very early modems (Sportsters). We were connected, but the bandwidth was still tiny.
“People were very excited about it, but I think they were more excited about the idea than the reality. We did have email, we did have FTP, and Mosaic had just started, so you could download software and little games.”
The idea for the café came to life when Pascoe’s partners were looking for a way to launch their Easynet ISP Internet provider, the second internet provider in London and Pascoe had been inspired by a visit to San Francisco where Bay cafés ran SF NET (launched in 1991).
“We were looking to make it easy for people to understand why they want to be on the internet. So, we thought the café would be the best way to contextualise it and show them the wide range of tools; email, FTP, browser, the ability to download games, music. A lot of that was really only possible in the café where you could help people because a lot of people were completely technophobic at that time.”
With women representing just 3% of internet users in 1994, Pascoe decided to organise HTML courses and only-women tech courses, an idea she had from her PhD course where she taught nurses to use the first computers in the NHS.
The rave scene, Kylie and Bowie
They also introduced Cyberia Records with help from Nick Ryan, a composer, who together with Keith Teare, invented the concept of embedding digital rights in the code.
“We were ahead of Spotify by quite a long time.” Other start-ups were Cyberia Payments with digital cheques.
The café also became known for its post-rave Sunday breakfast club as ravers would head there to check their emails before going home.
Located next to Whitfield Recording Studio, it also started to attract visiting performers who wanted to learn about the Internet or email. Eva says: “One of the early people was Kylie Minogue, who was recording an album then, and popped into Cyberia and wanted to learn about email. Similarly, U2 came and Bono was intrigued to see if they could send their music to other people. The most interesting visitor was David Bowie. He was extremely involved in the early technology. He had his own BowieNet, and we had quite a few connections with him.”
Having started the café on their own money, Pascoe and her partners realised that they would need investors to help them finance it as it continued to grow and secured funding from Maurice Saatchi and Mick Jagger which they used to open Cyberia in Centre Pompidou in Paris; the first UK company ever to be invited to Centre Pompidou followed by an expansion of 20 further cafés across Europe and the Far East.
Cookies and the Digital Bill of Rights
In 1997, Pascoe and her partners began to be concerned about digital privacy after the invention of cookies, the growth in collection of personal data and its use by third parties. They pulled together a group of people from the hacking community and academia who shared their concerns and began Cybersalon.
“We started working with people who could lobby against it and used artists to show the harm if your personal data gets collected by the wrong people and then leaked out. At some point, it morphed into more of a general protest against large companies, because the large companies took hold of the internet very quickly.”
The group developed a Digital Bill of Rights to protect people from the risks and spearheaded a movement of online regulation.
Oxford Internet Institute – researching the effects
It was the fast-paced development of the internet and its seemingly unregulated networks that led in May 2001 to the establishment of the Oxford Internet Institute (OII). This was a new research and teaching department of the University of Oxford, based in the social science division, and dedicated to the study of the social implications of the internet.
OII was founded as a result of interest from a number of different sources, including Government, on research into the effects of the internet and was aided by philanthropic funding from Dame Stephanie Shirley.
Professor Bill Dutton, interviewed by AIT in 2022, who was appointed as its first Director in 2002, says the fact the OII was a department that was anchored in the social sciences division of the university, not in engineering or computer science, made it distinctive.
“I think it was the only major department at a major university that was anchored in the social sciences and focused on the internet. Even schools of communication and media studies were not looking at the internet at that time.”
Of the decision to concentrate on the internet rather than the world wide web, Bill says: “I think that how people define the internet — and how narrowly or broadly — will be an eternal debate. I’ve always defined it as broadly as possible and I’m not interested in just simply looking at an internet protocol. The internet was really a way of networking computers and therefore networking people and I think that everything is going to increasingly network in the future and therefore, the internet is probably the broadest way of thinking about this revolution. The web, of course, is a way of facilitating the sharing of documents online and on the internet and so, I think it is an aspect of the internet, just like social media are.”
However, initially there was real scepticism about the future of the internet, Bill adds: “I must say that when I first got there, in 2002, some of my colleagues in social sciences in the UK said, ‘Bill, what are you going to do in a couple of years when the internet is no longer? It’s like CB radio. Or, it’s a fad.’ I replied I’ll be right here because everything is going to be more and more networked – forever.”
Currently OII is focusing much of its work on the societal risks around the development of AI, the potential biases of algorithms and the inequalities in society tied to a lack of access to these technologies.
“I think the OII is rightly taking issue with an uncritical view about the internet, but also we have to be critical of negative as well as positive views; we can’t take negative views about the internet for granted. For example, a lot of my research more recently, has questioned this notion of disinformation as it is really a technologically deterministic view. It doesn’t stand up to empirical research. I think more and more people are realising that that’s true, they just hadn’t done research around these issues. The job of social science is not to give a positive or a negative perspective on the technology but to question, challenge, conventional wisdom with empirical research and thoughtful analysis, otherwise we just bounce back and forth from these utopian and dystopian views about technology.”
He says that his current work is around cybersecurity as well as what he has called the Fifth Estate. He adds: “I think, if we can’t solve some of the issues with cybersecurity, then this really threatens the vitality of the internet going forward.”
Bill says the issues with cybersecurity are multifaceted, he adds: “It’s technical issues in terms of having technology that is more secure but it’s also having the laws in place, that correctly and appropriately regulate issues such as privacy.
What next? – AI, Interplanetary and Quantum Internets
Since the 1990s the internet has been implemented in thousands of devices including mobile phones, smart speakers and satellite navigation systems making a reality of the term The Internet of Things.
One of the great challenges this creates is how to manage secure networks and regulate the information accessed and published.
AI poses one of the greatest challenges to how people use and interact with the internet.
On 1 November at the UK’s first AI Safety Summit 28 Countries and the EU agreed substantial risks may arise from potential intentional misuse or unintended issues of control of frontier AI, with particular concern caused by cybersecurity, biotechnology and disinformation risks.
They signed The Bletchley Declaration on AI Safety agreeing to the urgent need to understand and collectively manage potential risks through a new joint global effort to ensure AI is developed and deployed in a safe, responsible way for the benefit of the global community.
Vint Cerf has taken JCR Licklider’s 1971 vision of an Intergalactic Computer Network to the next level by devising ways to extend the internet across the solar system to enable astronauts and probes to communicate with earth and download their data.
In 2003, Cerf and a small team of researchers introduced bundle protocols and the same year an in-space demonstration of the Interplanetary Internet was carried out on board Surrey Satellite Technology’s UK-DMC1 satellite.
Quantum Computing and a Quantum Internet
A quantum internet would use quantum signals instead of radio waves to send information. Sir Peter Knight, interviewed by AIT in April 2023, says part of the problem is to work out how to make quantum resilient or quantum resistant architecture in the internet.
Another aspect he says is to start working with all of the sectors of the economy that depend upon this to make sure that they’re alert to the need to be quantum ready.
“Part of what we’ve been able to do is build this thing called a quantum readiness programme, where we are talking to the banks and to others and to the data centres and providers to make sure that they have an alertness to what is necessary to engineering this.”
The UK Quantum Technologies Programme has provided part of the funding to help the National Dark Fibre Facility (established in 2014 to enable researchers to develop the underpinning communications technologies for the future internet), to establish the UK’s first fibre-based quantum network (UKQN).
This dedicated dark fibre network now connects four university sites in Bristol, Southampton, Cambridge and London at UCL, and major internet exchanges including the UK 5G Exchange at Slough and international connections at Harbour Exchange with further connections planned with the NPL.
In 2024 a 20-strong, pan-European private consortium, supported by Horizon Europe and European Space Agency funding, will launch the continent’s first space-based quantum cryptography satellite.
Martha Lane-Fox – co-founder of lastminute.com
David Clarke MBE – businessman in the early days of internet providers
David Barker – internet and social entrepreneur
Where is AI taking us? AIT feature May 2023.