The birth of .uk: an oral history

This year marks the 30th anniversary of the launch of all things .uk, the home of Britain online.


The story of why we have .uk — and not, as the Americans once insisted, .gb — is a significant part of the story of the internet as a whole, from a time before Tim Berners-Lee’s invention of the World Wide Web, and a time when the protocols, the language, the rules, and the very wires the whole thing runs on were being designed and built.

To mark the anniversary of .uk’s birth, we’ve spoken to some of the figures who played a part in those first few years, to hear the story of the early internet from their British perspective — as it grew from an academic and military research network into an engine of 21st century global social and economic change.

The participants

Dr Willie Black: Played a key role in the earliest nationwide network in the UK, connecting research computers together. Was the leader of the Naming Committee, and was founder and first CEO of Nominet, its successor.

Peter Kirstein: Worked at CERN from 1959 to 1963; served as head of UCL’s computer science department from 1980 to 1994. Co-authored two important papers with Vint Cerf on the structure and protocols of the early internet. Inductee of the Internet Hall of Fame, and winner of many, many computer science awards.

Steve Dyer: Former director of Nominet; also founded CentralNIC.

Keith Mitchell: Was one of the founders of Pipex, the UK’s first commercial ISP, in 1992. Currently managing director of the UK Network Operators’ Forum (UKNOF).

The story

ARPANET is widely considered to be the precursor to the internet we know today, but it took four years before it began to expand outside of US territory

In the 1960s the US government, concerned about how to maintain communications in the event of a Soviet missile strike, encouraged the development of ARPANET — a decentralised network of research and defence contractor computers across the country. ARPANET is widely considered to be the precursor to the internet we know today, but it took four years before it began to expand outside of US territory.

Peter Kirstein: Back when I started ARPANET work at the Institute of Computer Science in London — this was somewhere around 1970/71 — I tried to get the Science Research Council to back what I was doing, because DARPA [the US military’s research division] had a link that they were putting into Norway for seismic monitoring for their nuclear monitoring research operation, and they offered a link to that to the National Physical Laboratory. But that was the time that Heath wanted to get into the European Common Market — and because we weren’t meant to be too closely linked to the Americans, the NPL was not allowed to link to them.

The ARPA Network made up of 4 nodes in 1969

I was the second choice — I knew these people very well, and I was offered the link. But they first of all asked at the highest official level of DARPA for confirmation, and they couldn’t believe my research application — this was the first that the director of DARPA had heard of it, and that nearly torpedoed it from the beginning, because he didn’t like to hear about what his own people were doing from a foreign power! [Laughs] But I solved that little problem.

Willie Black: I was offered a job in Oxford in the Nuclear Physics Department in 1974 — I worked not as a nuclear physicist but in the analysis group that was doing bubble chamber scans. It used to blast particles into chambers of supercooled hydrogen, and there would be four cameras taking pictures of the interaction. You got lots and lots of pictures, and we were integrating equipment that would scan these automatically and find the curve of the particles from the bubbles left behind. This was with computers less powerful than your mobile phone.

In those days “high speed” lines were running at about 10 kilobits, not 10 kilobytes

Towards the end of the 70s we started to link up all these computers through various networks, and that’s how I ended up getting into networking. The physics community linked our computers through to the CERN ones, to try and speed up the data flow. In those days “high speed” lines were running at about 10 kilobits, not 10 kilobytes.

We were starting to link lines between Rutherford Appleton lab up in Oxford, and a lot of the universities — Glasgow, Cambridge and Oxford — in the physics departments more than the computing science departments, which were still running monolithic computers that you went to and put your cards in. And we ran terminal access email and things like that, before it was called email.

Steve Dyer: I was at school with a chap whose father worked for [early computer manufacturer] Ferranti. At age 16 I used to get to programme a machine called the Ferranti Sirius. If you want to know what one looks like, one hangs over the locomotive room in the Science Museum. It’s about the size of a very large office desk. As I recall it had 460 words of memory. [Laughter] It did some wonderful things, but it was a very, very, very small computer.

Then I went on to Kiel University [in Germany] where I played around with various computers and we had an Artificial Intelligence department there. We were trying to do very clever things with computers in the days when they weren’t very clever. And I sort of wandered from that into programming , and along the way in that I bought a whole load of computers from a company that went bust — they just happened to be UNIX computers, and of course UNIX came with the sparks of the internet within it.

Peter Kerstein

PK: In 1973 the British, as part of coming into the Common Market, introduced VAT of 10%, and then my communication computer arrived from DARPA with a declared value of £50,000 — and customs claimed 10% for it, so £5,000. The Research Council refused my application for funding, and Donald Davies [Welsh inventor of packet switching networking, crucial to the development of the internet] offered me £5,000 — that’s the most he could do out of his own funds — so customs would have taken all the money I had, so I said I would appeal it.

Two years later I got a call from the UCL secretary to say that my appeal had been refused and I’d have to pay my £5,000. By this time I was running things for the Research Councils, the Ministry of Defence, for the Atomic Energy Authority, for the British Library — I was running lots and lots of things.

I was summoned by some very senior people in the Treasury and I was given a set of questions, a catechism almost: “What you’re doing is of no interest to the Ministry of Defence?” No, said I, I had a big contract from them. “What you’re doing is of no interest to the British Library?” No, said I, I had a contract with them. “It’s of no interest with Atomic Energy Authority?” Yes, it’s given me a computer, etc — went through about seven ministries, and they said thank you. Two weeks later I got a letter permitting me to import that equipment and any future equipment under the same project, free of tax and duty.

For the next 10 years, people looked at my research group running services like the international services between the UK and US for God knows everybody, and everybody wanted to take it over. So I said, “fine but you will have to pay tax and duty all over the place”. They’d look at it, decide it was too expensive, and that’s why I ended up running everything.

“Peter Kirstein at UCL set up a gateway between the two, between the ARPANET as it was and the JANET”

WB: As the US Department of Defence was doing the ARPANET, they were developing networked communications, the equivalent of what the physicists in this country were doing, you know, talking to each other. And of course, ARPANET quickly started to draw more and more universities, and other companies started to see the benefit of being able to use it — and Peter Kirstein at UCL set up a gateway between the two, between the ARPANET as it was and the JANET.

JANET

JANET — Joint Academic Network — emerged in the late 1970s as different regional research networks began to link together across the UK. It went properly live as a unified network in April 1983 after a nationwide standardisation programme — but these standards were different to those that were used by ARPANET (TCP/IP). JANET still exists, and its parent organisation JISC is responsible for gov.uk and ac.uk domains.

PK: [BT] had a 9.6 kilobit service and 48 kilobit service in those earliest days. I wrote to them, saying I don’t know whether I should take the 9.6 or the 48, I have to get my traffic to Boston, New York, Chicago, Washington, Seattle, Los Angeles, San Francisco and a few other places, could you tell me how many packets a second I get to each of those on your 9.6 and your 48 kilobit services? There was a six month silence. They then responded: “Dear Peter, it’s an interesting question you ask, we can’t actually answer it, but if we gave a free 48 kilobit line, do you think you’d do some measurements for us?”

For the next five or six years I had a free 48 kilobit line. That lasted until about 1985, when British Telecom said “oh sorry, but as a result of new legislation, we have to withdraw the free service”. I said “OK, fine, but would you then please give me regular bills for what I’m now paying for?” Every three months I would ask for a bill, until after two years I finally got a bill for £35,000. [Laughter] I wrote back and said, “look, I’ve been asking for a bill regularly, you never gave it to me, I’m sorry, I only have £6,530 available for this whole project”. They came to see me, they apologised and they settled for that.

Shortly after that there was a major evening in the National Opera where lots of powerful people were invited — it was one of these champagne cocktail and dinner things. I wasn’t clear why I was there. The head of BT got up and gave a talk, and said “well you all know why you’re here, you’re our best customers”. If I was their best customer with the loss they made on me, then that couldn’t have been very good

UNIX Machine Image: Magnus Manske

WB: Peter was linking these networks using UNIX type protocols between UNIX machines, but there were also other independent networks that were linking machines using their own protocols — it was a menagerie.

PK: There were so many people, so many organisations involved; on the UK side it was the JANET, and the Ministry of Defence were paying for half the link, and DARPA, the National Science Foundation, and NASA were paying their side of it. To organise that between the five agencies was probably my supreme act of diplomacy. It was very difficult.

“Internet” just basically means a linking up of all the networks together

WB: You can see why we wanted to get to a kind of unified standard. The ARPANET became the internet somewhere around the early 80s as well, because in the US the same thing was happening as here — commercial firms wanted to start using the same network the Defence people had used. They started to try to have commercial internet services, that we call an ISP nowadays, and linked them into, basically, the internet as it was starting to be called. “Internet” just basically means a linking up of all the networks together. And there was a great problem about the gateway between, if you like, these non-standard networks.

As long as the number of computers on the proto-internet of the 1980s remained only in the hundreds or thousands, it was possible to jerryrig fixes for incompatible standards — and for academics and researchers, who had been running things so far, to manually take care of issues like registering new domains.

WB: When JANET was formed we had the Name Registration Scheme, the NRS. And the NRS had given names the other way round — as uk.ac.oxford, or something like that.

PM: The British had chosen to do it the opposite way round to the Americans. So whereas my address here is at cs.ucl.ac.uk, on the NRS, which the British were running, it was .uk.ac.ucl.cs. I had to create an algorithm to translate it one way or the other — it was a nightmare

Keith Mitchell: Peter Kirstein was the supervisor of the UCL Computer Science course that I was on in 1985. Then I went to work for a company in Edinburgh, and when I was running the email system there some of the interesting issues started to come up because the UK had a different kind of naming system from the rest of the internet. For probably around a decade the UK had the academic way of domain names, where they went uk.hq.institution.department, and there was the internet way, where they went department.institution.hq.uk, which all caused quite a lot of interesting conflicts when you were configuring your email system and you were trying to work out whether CS stood for computer science or Czechoslovakia.

The British always think they know best, and they didn’t like .com because that was really assuming that everything commercial was American

PK: The British always think they know best, and they didn’t like .com because that was really assuming that everything commercial was American. It was also related to the fact that JANET was responsible for various things, and around about the same time one was starting to get the question of the Health Service, Ministry of Defence, they wanted to have some control of their own things. So they decided, on the one side there would a .gov, and so that would separate out all the government things. Then there’d be .ac to have all the academic things taken. And then there would be of course the rest. So you put a .co before the .uk. And .co.uk was born.

The British decision, from 1985, to use .uk for web domains clashed with the wishes of the Internet Assigned Numbers Authority (the IANA). Its director, UCLA computer scientist Jon Postel, was known affectionately as “the god of the internet” among those in the industry for his role in establishing unified standards across the developing network. In his office he kept a list of who was responsible, in each country, for handing out domain names.

WB: By the late 80s the IANA [the Internet Assigned Numbers Authority, set up in 1988 to manage global IP address allocations] was trying to get all those countries that were trying to join the internet to use the ISO 3166 standard for country codes. It was used for all sorts of things — you see it on cars, “GB” for the UK.

John Postel — Photo by Irene Fertik, USC News Service. © 1994, USC

At that point, we’re faced with a problem that Jon Postel would like to have changed it to .gb to be consistent with the rest of the world. Whereas .uk had already been established, with a few tens of thousands of domain names with .uk on them. I remember chairing one of the JANET net workshops that were held every year, and the Northern Irish were adamant that they were part of the UK — so the consensus was, we’d try and keep .uk, we’d park .gb and not use it.

PK: I didn’t particularly want to change to .gb because I was responsible for Northern Ireland as well. And what’s more, there was a certain question as to whether a research group in the US should be allowed to tell the British what to do. So this argy-bargy continued for a little while and, in the meantime, one of my clients was the Ministry of Defence, and they decided they couldn’t wait this long, and they decided I was going to lose the battle, and so bits of MOD went over to .gb — I didn’t care, as I was running .gb and .uk in any case.

The Ministry of Defence’s .gb sites were for many years the only prominent sites on the .gb namespace, and even though they have now moved over to .uk, there are still a handful of unused .gb MoD site registrations — probably the only .gb registrations left.

WB: I don’t know whether Ukraine was a bit upset at that, but .uk was reserved for the United Kingdom in the standard, which was our saving grace. So, although .gb was “official”, nevertheless, .uk was reserved so nobody was going to steal it.

PK: There was a big meeting of all the interested parties to discuss this question about .uk and .gb. By then it was probably ‘87/’88, it would have cost £25 million to change, and nobody was willing to pay the £25 million. So they argued that Jon [Postel] had no right to insist on this, and so I had to tell him that. And then I added, I guarantee to change within one week of the US going to .us, and Jon didn’t see the humour of that at all!

So .gb stayed — it stayed because nobody killed it, and nobody’s very interested and nobody’s bothered. I think I may still be in charge of .gb, I’m not sure!

By the 1990s, even before the invention of the World Wide Web, it was clear that the internet was beginning to grow faster than anticipated by its creators and caretakers — and that there would need to be a more professional, automated system for dealing with domain registrations.

By the end of ‘91, there were about six ac.uk, or uk.ac, hosts. And about a thousand uk.co

WB: By the end of ‘91, there were about six ac.uk, or uk.ac, hosts. And about a thousand uk.co. So, you know, it wasn’t much compared to today, but these were growing up and by the early 90s more and more people wanted the uk.co. But it became very difficult for these external ISPs to actually get in and register their customers’ names, in the NRS, which was basically an academic thing for the JANET servers.

Jon [Postel] was faced more and more with this same issue that we were faced with among the volunteers. Back in ‘91/’92 the method that we used to register new domains was basically, there was an email group called the Naming Committee, and when you wanted, let’s say, xyz.co.uk for your customer, you sent an email saying “I want xyz.co.uk” and if nobody objected within three days it was yours. And JANET, but me in particular, arbitrated if there was a dispute.

Steve Dyer

SD: One day I decided I would start an internet company, called Mailbox Internet. And that was where I first came across domain names, because of course I wanted Mailbox.co.uk and so I went to the Naming Committee and I said “right, I’ve got an internet company, I want it to be called Mailbox, because that’s what my wife’s company was called, and I’d like the name”. They said no, no, no, you can’t have that name, it’s too generic.

KM: In 1992 I set up Pipex, the UK’s first commercial internet provider. We were very much in the thick of this because we were working with the NRS, and trying to make this all work with the rest of the internet.

We were also learning ways in which you run an organisation which is sort of critical to international internet infrastructure, and established quite a lot of principles in terms of non-profit, public company limited by guarantee, membership organisation, mutual spaces, and basically a whole bunch of things which were moving away from the sort of public sector hierarchical control of the internet namespace for the small number of large companies back then — there was CompuServe and AOL and Prodigy and all of that. We shouldn’t be going back to that period of time where bits of the internet didn’t really inter-operate with each other very well.

The Naming Committee was being dominated by these big players and the little guys were being cut out. We needed to do something — to set up an organisation and then use .co.uk in a proper way

WB: Generally speaking folk got on quite well with the Naming Committee and it worked. But by ‘93/’94, it was growing so fast that it wasn’t working so well. I was having to deal with a lot of issues that were coming up from people, disputes. People said “I registered that name first” or “you poached my customer” — there were a lot of accusations around. A lot of people starting to grumble that it wasn’t doing a very good job. The Naming Committee was being dominated by these big players and the little guys were being cut out. We needed to do something — to set up an organisation and then use .co.uk in a proper way.

SD: We had thousands of customers hammering at our doors, and everything was terribly slow, as the Naming Committee was manual of course. The Naming Committee was losing the ability to control things. There was conflict between the ISPs who wanted to get on with it, and the naming people who were basically academics, who weren’t in the commercial world so they weren’t making money.

Keith Mitchell

KM: How do you do this stuff ? You do it in a way which shares the load and is accountable to everybody.

In 1996, Nominet was set up as a not-for-profit membership organisation to handle the .uk namespace , taking over from Peter Kirstein at the recommendation of JISC — and has been responsible for it ever since.

WB: So, I became the name in Jon Postel’s register to be responsible for .uk. And incidentally, also for .gb [laughter]. To try to get us to use the ISO 3166 standard.

It’s so tragic that he died, actually. I remember chairing a meeting in Spain [in 1998], and we got a phone call to say that unfortunately Jon had died. He was a good guy — he did everything very much for the greater good. Quite a humble, academic kind of guy. But the US is full of too much commercial interest, some really saw it as a big money spinner, and there were times when I wonder, we could have done the same with .uk and made a few million each. But I think my conscience is clear.

SD: I think, quite early on, I realised the internet was going somewhere very, very different. The very first company I set up of course it was tiny, I had very little money to do it — we got the internet into our offices as it were and we started trading. And the very first thing we did was to drop down little leaflets to all the local houses in Fulham and say: “Come along, have a glass of wine — come and see the internet”.

The very first person to take it up was a dear old lady and she sat down and said: “I don’t know anything about computers, I’ve just come to see the internet, you’ll have to do everything” and I said: “Now, this is a mouse, now hold that, see this arrow on the screen see how it moves as you move your hand”. I said “right, click there” and she clicked the thing and I said: “you are now logging on to the FBI’s website in Washington” and she said: “Oh, oh, oh!” — she was terrified. I explained that no, they want you to come on their website, they want you to look at the pages, that’s fine.

I showed her a couple of other things and then I got dragged in to see the next person and when I came back ¾’s of an hour later, she had been all over the world, she’d been to NASA, she clicked here and clicked there and I thought to myself — this is the first time I have ever seen a completely non-geeky person, a normal human being, sit down and drive a computer. I suddenly realised: my entire career had been in the computer industry, but this was the first time that the thing was really part of everyday life. And that was the thing, that dear old lady, golly, I thought — I’m in the right place doing the right thing here.


This story is one of 30 celebrating the launch of .uk domain names in 1985. To read the others visit our 30 Years of .uk hub. To start your own .uk story check out www.agreatplacetobe.uk.

Click the logo to read more stories about .uk