Recollections of the first few years of the net

While reading Bell (2001) I was reminded of the time in the late 1960s and early 1970s when cyberculture was beginning.. and how quickly it grew after basic computer access and communication became possible.  We are social beings. Let me recall a few things as stepping stones.

My own use of computers began in the days before networks were seriously developed. Computers were largely standalone big machines in air cooled facilities.  My first programming exercises at a night class at Leeds Polytechnic around 1967 were submitted on coding sheets via punch operators on punched cards and the turn round was one week to the next evening class.

Things improved when I went to the University at Lancaster in 1969… we could punch our own cards :-) and then leave them in pigeon holes to be run overnight on the University’s single ICL 1900 computer. A mistake in the program due to a simple typo was a no no if you wanted to get a result.  We could program via flip switches in octal code a DEC PDP-8 that was the size of a large upright fridge freezer.  I write some interrupt routines for a disc driver on a a basic operating system using a limit of 1K words of memory on that around 1970.

But things were changing and interactive access to the same type of computer was coming thanks to a link up between Edinburgh University AI people and Malcolm Atkinson, then the Computer Manager at Lancaster, and since then a long term colleague, co-investigator and recently Director of the National e-Science Centre based in Edinburgh.  A precocious 17 year old programmer called John Scott at Lancaster wrote a real time access version of the Edinburgh POP-2 Language for the ICL 1900 and we were away into the cyberworld for real.  We learned POP-2 for most of our programming exercises, for data structures, and for a new AI course at Edinburgh around 1970-1971 which I signed onto.  These, and the consequent links to Edinburgh researchers interested in planning using computers set the direction for my whole career.  With encouragement from Donald Michie and Jim Doran at Edinburgh I did a final year undergraduate project to build my first AI planner – Graph Traverser 4 – and used a “compilation” approach to how plans were composed. I applied it to a range of benchmark tasks that others had tried their planners on. It far outperformed the others.  In July 1972 I  was able to get a small grant to allow my continued work on this and its writeup to continue after my degree – my very first research grant!

Donald Michie in Edinburgh had offered me a PhD place at Edinburgh and I joined him at the Department of Machine Intelligence and Perception in October 1972.  Real time access terminals using POP-2 and the time sharing Multi-POP system were the order of the day.  But it was to be an exciting time… 

Within a year the DEC-10 that was used by all AI researchers across Britain was installed in Edinburgh and became connected to the ARPANet – it was the 6th or 7th node on that network. Our access terminals could now be used to “Telnet” across to log on to other DEC-10s.  I especially used the Stanford  machine.  There was rudimentary chat, and e-mail was started with the famous “@” character being used to address users on other hosts.  Working for the first 2 hours each morning UK time I was often one of the few people on the entire network and had access to 2 DEC-10s for my work.  Multi-User Dimensions (MUDs) were experimented with soon after in the machine I used at Stanford… and the rest is history…


Bell, D (2001) Storying cyberspace 1: material and symbolic stories, chapter 2 of An introduction to cybercultures. Abingdon: Routledge. pp6-29

  • Austin Tate says:

    Malcolm Atkinson added some more detail and the real facts in a message to me today…

    The machine at lancaster when you were there was an ICL 1909 with 32K 24-bit words and 4 tape decks. Jobs were submitted on punch cards.

    Andrew Colin and I added a exchangeable disk store holding 2 Megabytes. The drive cost £40K, half from EPSRC (then SRC) and half from the computer board, as a consequence of the Flower’s report (his first I think — Peter Buneman’s step dad) and we implemented a batch operating system taking 2K words, called June after the head operator, that then ran the batch service you allude to. Once we had done that, as we streamed input and output vi two of the tape decks, the rate of handling small jobs was much improved.
    With Wallace Anderson and Andy Lister, we bought a PDP8 (8K of 18-bit words?) and attached 8 teletypes to it, using it as a multiplexor and text editor, with a maximum of 256 words/teletype for the editable text buffer. We connected it to the ICL1909, and you could edit and submit jobs to it. That was working by late 1969 before I returned to Cambridge.

    I think there was another PDP8 obtained later for teaching; perhaps that is the one you used via its keys; although the replacement machine (which I had arranged to be a PDP10 and which was another ICL machine) ran a new ICL operating system: George 3, and all our work was discontinued. I worked on the DEC10 writing the POP-2 system for it with Ray Dunn for Donald Michie’s round table, and was already hooked.

    The Computer manager was Wallace Anderson, he later went to Aberdeen to be computer services director there.

    I was a PDRA for a year and then a lecturer in the Maths department, under Elwyn Lloyd. Andrew & I designed the CS course, got it off the ground, and then did a UDI from Maths.

    John Scott did most of the ICL POP2 compiler. He who entered a chess program as a schoolboy, running on our 1909, versus Greenblatt’s (?) chess program running on the MIT system, at the Machine Intelligence Workshop, 1968.

    My initial experience with interactive systems was with Cambridge Titan & Digital’s DEC 10. At that time the DMIP mindset was paper-tape in and out of an Elliot 1409. I used the Timesharing service on the DEC10 run by Timesharing Ltd in Tottenham Court Road via an acoustically coupled modem that the GPO would not approve of from my house in the country in Dolphinholme. After an hour or two, I would get the teletype covered with splodges because an operator in the Morcambe manual exchange would have heard the weird noise, when wondering why the call was so long and listening in; she would then be plaintively asking “Is there anyone there?” sometimes generating or corrupting characters.

    EMAS was much later.

    We wrote about the Lancaster operating systems in the Computer Bulletin.

    Thanks for the chance to reminisce…

  • Austin Tate says:

    Oh..I remember those accoustic dial up modems too. 10 characters per second .. could put the entire phone headset into it and close the foam filled lid to improve the signal quality… size of a shoe box.

    And my PhD thesis in 1975 was typed on a 10cps teletype that output paper tape in the Machine Intelligence Research Unit in Hope Park Square at Edinburgh. Typing mistakes were quicker to fix by splicing in the punched out holes and manually punching the new ASCII character code.

  • Siân Bayne says:

    This is a fantastically direct and personal ‘material story’ to go alongside Bell’s. You need to write a memoir – the fact that in the early days you were the first person to log on to the ‘internet’ in the morning is mind-boggling! Thanks for this Austin.