Brian has had a long and illustrious career in computing, starting work at English Electrics Atomic Power Division in 1957, after graduating from Imperial College of Science and Technology. Brian worked on various programming tasks, including the creation of the EASICODE System for the DEUCE computer and an ALGOL 60 compiler for the KDF9 machine, the latter of which was the subject of his book, Algol 60 Implementation, along with Lawford Russell, which was one of the first books on compilers. Brian then joined IBM Research at Yorktown Heights in 1964, where he worked on computer and systems design. After five years at IBM, Brian became Professor of Computing Science at the Department of Computing Science at Newcastle University, where he has remained since.
Brian Randell was born in 1936 in Cardiff, in South Wales. His father ran his own business as a credit salesman, selling furniture around the Rhondda Valley. His mother was very keen on sports, and took over running the business during the war while his father was in the RAF
Brian attended a local elementary school, followed by Cathays High School. He says that he has pleasant memories of his school days and enjoyed reading. Despite being on course to study languages, his teachers noticed his ability for mathematics, and he ended up studying for pure maths, applied maths and physics, at A level.
Following his A levels, Brian completed a further year of study to gain a state scholarship to study mathematics at Imperial College, London, (then known as Imperial College of Science and Technology, University of London). He was the first one in his extended family to go to university.
It was at Imperial that Brian discovered his love of computers. He says: “Mathematics was much tougher at Imperial College than at school, so I was very glad to take refuge in computers.”
The college had its first computer – the ICCE I, Imperial College Computing Engine number one, which had been built by Keith Tocher and Sidney Michaelson. A second machine, ICCE II, was underway, but the project had been halted. Brian explains: “I learnt then or soon afterwards that the famous PMS Blackett, one of the most notable people at Imperial at that time, was somebody who laid down the notion that a maths department is not the place in which a computer should be built. So that got abandoned.”
However, Brian was shown the machine by Sidney Michaelson, who also arranged for Brian to take a programming course at City University, then known as Northampton Polytechnic, in 1956. Brian followed this course and did a third-year project writing a program in linear algebra. He says: “I found computers fascinating and, interestingly enough, I think I was the only one in my class who did.”
After university, Brian joined English Electric as a programmer to write nuclear reactor codes on DEUCE. The role meant that he received a deferment for his national service. He joined alongside fellow student Mike Kelly, who invited him to lodge with him and his wife. Mike Kelly and he worked closely together, on their own initiative, on an aid to programming they called EASICODE.
Brian says: “In those days you typically would write a computer program and then do hand calculations to try and prove that your program was generating the right answers. I used to give the impression that I used the computer to debug my hand calculations, rather than the other way around.”
When Mike Kelly left to join IBM, Brian was made head of the section writing compilers for the upcoming KDF9 computer and was joined by Lawford Russell. Having written their ALGOL compiler, based on the work of Edsger Dijkstra at the Mathematical Centre, Amsterdam, it was suggested to Brian and Lawford that they should write a book about their experience, giving others not only information they may need, but also insight into the process.
Brian explains: “The first thing we did was check that Edsger Dijkstra was happy about it and he wrote back very quickly saying he was very happy about it and gave us the wonderful advice saying that a book which just describes your compiler in detail will be of interest to a very limited number of people, but if you write a book which describes all of the design decisions that you considered and the rationale that you used for the choices that you made, and provides a review in the light of what happened of the merits of those decisions, that will make it a much more interesting book. So that’s what we set out to do.”
Knowing that there were other ALGOL compiler projects underway, Brian and Lawford set out to get the book written and published as quickly as possible to beat the competition. Brian adds: “It was only later that we found out that the others either weren’t doing a book on their compilers, or certainly weren’t working on our timescale. So, getting it out and getting it out that early, and getting out a book of the type that Dijkstra had recommended certainly had a very good impact. It meant that when I started looking to move on, the existence of that book helped.” He goes on to add that when he joined IBM, he met one of the vice-presidents who came to see him carrying a copy of the book.
It was also during his time at English Electric that Brian developed an interest in dependability in computing. Together with Mike, he formulated a rule about the sort of programs they were generating. He explains: “Because we were generating programs for other people to use, application programs, for our own benefit, if nobody else’s, we tried to write those programs so that they could cope with all sorts of nasty realities. Incompetent operators, incompetent users, incompetent key punch operators and so on. … One of our rules was that a program should deal sensibly with any input that was given it. It shouldn’t just fail, it shouldn’t just check, it should give you some information as to what the problem was.” He continues: “Even back in the days of writing EASICODE, we tried on grounds of practicality to have a system that was easy to use. When you live in the same large office as the people using your programs, then you pay a lot of attention to their problems so that they don’t become your problems.”
Brian says that dependability is not an issue that industry considers until they hit an issue. He explains: “Typically people will start worrying about dependability only when they belatedly realise they’re already depending on something. The bright new idea is pursued and developed and often only later do people realise, oops, there’s a security or there’s a dependability issue here. Present work on AI and machine learning, or on the internet of things, all of those have had this sort of trajectory. They start getting really significant and then it is belatedly realised that there are issues like the dependability issues, which are important.”
He feels that commercial pressures are often behind the reasons why dependability is not considered early enough. He adds: “The whole of the computing industry has developed at very high speed and the ability to be first to market is so important that it probably makes commercial sense to hurtle along and get something into the market and then later, when you perhaps start earning money from that, indulge in the luxury of making it dependable.”
In 1964, Brian was invited to join IBM at its Yorktown Heights laboratory but initially refused. After visiting the factory at Poughkeepsie and spending some time with John Cocke, he finally accepted but with the express wish to not work on compilers, he says: “I thought IBM was such an amazing place that I guess I was thinking that to not take advantage of that and learn something new would be an awful waste.”
The shift from English Electric to IBM exposed Brian to a much larger working environment of over 3,000 people, many of whom had PhDs. Brian initially found the change a “little overwhelming”, but his confidence grew, and he enjoyed himself.
He joined a research project (Project Y) to design a supercomputer. He worked alongside a number of people who had been involved in the Stretch computer; IBM’s attempt to build the world’s fastest computer which had been somewhat successful but came up against strong competition from CDC and Seymour Cray. Brian’s first role was to help design the order code, working with Herb Schorr and for John Cocke.
When IBM moved the project from research to development, Brian and ten colleagues were asked to move out to a West Coast laboratory to initiate the creation of a new computer development department; ACS, Advanced Computing Systems. Brian says: “All of this was highly secret. It was pretty secret from the rest of IBM, and certainly from outside.” While at ACS, Brian was involved in the invention of a dynamic instruction scheduling scheme.
Brian later returned to IBM’s Yorktown Heights laboratory and joined Manny Lehman’s project, IMP; one of the first multi-processor projects. Brian explains; “That was at a time when IBM was having troubles with OS/360 and in particular with the time-sharing system, TSS/360. That was being built at a different lab, but some of my colleagues were borrowed, or spent time there and so I knew a fair bit about what was going on and what was going wrong there. I and a colleague, Frank Zurcher, got very interested in the whole issue of how to design systems. The work that we did together got very, very intense. We could be involved in deep conversations that made no sense to anybody else, but where we could understand fully what was going on. That led to our work on multi-level modelling and that work on multi-level modelling I found myself referring back to in almost everything I’ve done since.”
In 1969, with his young daughter about to start school, Brian and his wife made the decision to return to the UK. He applied for a role at Newcastle University as a second chair in computing, under Ewan Page. Brian says: “I visited Newcastle and remember with great amusement sitting in Ewan Page’s office and we’re having this wonderfully delicate conversation, when in retrospect it’s very clear that I was trying to find out the chances of my being offered the job and he was trying to find out the chances of my accepting it.”
Ewan Page is credited with introducing computing at Newcastle University, Brian says of him: “Absolutely wonderful character, almost the archetypical benevolent despot, who was revered by just about all his own staff and feared by just about everybody else, which was great. He was very keen to build up research at Newcastle and to appoint a research professor, he was more than keen to go on running the place, and that appealed to me greatly.” Brian took encouragement from the fact that both Tony Hoare, and John Buxton, had been appointed to chairs in British universities and, like him, neither had gained a doctorate. He adds: “So, I didn’t regard that as a necessary bar, the precedent had been set as far as I was concerned, and rather arrogantly felt, if they could do it, I could do it. So, the fact that I had never given a lecture course in my life, which didn’t seem to faze Ewan, didn’t faze me either.”
With funding from the Science Research Council Jim Eve and Brian established a project on system reliability, a term that was later to be changed to dependability. Brian says: “I managed to get some very good people to join in on the work. If I look back on all of the things from then right back to the beginning of my time at English Electric, you might say what I did mainly was make sure that I worked with very good people. So that was the start of work on dependability which has carried on ever since.”
In 1981 Brian was working on what he describes as “probably the single most protracted exciting period of my research, because it pulled together both an exploitation of some very simple, very general ideas and some very practical and very clever hard work by colleagues like Lindsay Marshall to produce something which really was quite something; something we were very proud of and that we really enjoyed doing.”
The aim of the project was to find ways of producing highly dependable distributed computing systems, starting with the plan of choosing the most useful distributed computing system and finding out how to make it more dependable. However, the team were not best impressed with the then mostly Unix based systems, so they built their own distributed system. The team included Lindsay Marshall, Robert Stroud and Dave Brownbridge, among others.
On the subject of successes and contribution from that project Brian says that the Newcastle Connection software they developed and the things associated with it, including the work on security, on load leveling and triple modular redundancy were important, adding: “We had a whole Lego kit of how to build systems, thanks to the Newcastle Connection.”
At the same time Brian was also involved in establishing MARI which was a contract research organisation, which sold the Newcastle Connection. Unfortunately, competition from the Sun Microsystems Network File System, a distributed file system, which was offered free, killed off the Newcastle Connection commercially.
Brian has now been at Newcastle University for fifty years, not a move he intended. He says: “I’m not sure that I’ve ever planned things very much. I’ve taken advantage of opportunities. I may have had sort of strategic aims, not so much for myself, but attitudes as to what is important and what should be done. Certainly, I had no thought when I came to Newcastle that that was going to be my last move. There were quite a few times when things were dangled in front of me, typically to go back to the States, but almost invariably that involved a total misjudgement as to what would attract me. The idea of some prestige job, being responsible for large numbers of people and so on, was exactly what I didn’t want.”
During those fifty years, Brian has witnessed many changes to education and specifically the teaching of computing. He explains: “When I was at English Electric, universities were places we sold computers to, rather than places we expected to get anything from. In IBM, I was turned into somebody who belatedly realised he was part of a research community and was doing research, and research for its own sake, albeit motivated by views as to what research was worth doing. I used to joke, using a sentence that John Buxton had, I think, originated about having left the ivory towers of industry for the sordid commercial reality of a university computing lab. So, I probably after a while started getting an exaggerated view of how important universities and university computing was compared to industry. I certainly found through my work, particularly with Brussels (the ESPRIT programme), that it was possible to work very co-operatively between industry and university, and that I thought was a very healthy development.”
In the 1990s Brian worked on the European ESPRIT programme. He had been sceptical about ESPRIT 1, because he felt most of the work had been largely industrial with few universities involved. He says: “But things changed considerably, I think a fair part because of the creation of the ESPRIT Basic Research Project scheme, which I’m pretty proud of having had an involvement in.”
On the subject of whether higher education is preparing graduates for employment, Brian says: “I like to think it is. I assume that higher education has got a whole range of merits and to judge it as a whole, I don’t feel that I’ve got any particularly good viewpoint. I think that Newcastle in general does a good job. I think there are some aspects of what we teach or what we don’t teach which perhaps is not quite so much to my liking. But my presumption is that we have enough in the way of contacts with industry and we’ve got enough in the way of reputation that we’re well respected and we’re well respected for good reason.”
The History of Computing
Brian has a passion for the history of computing which he believes is not only interesting but also has a practical importance in helping students to understand the past and prepare for the future, as well as recognising the great work done. He explains: “I also think that a profession that doesn’t honour its sources is a pretty poor profession, that it’s appropriate for somebody to know, who are the heroes, who should I be grateful to, who should I think has had a life that I should know a bit more about. He also adds: “it’s one of the subjects that is pretty good at getting you ready for pub quizzes. It’s fun, it’s interesting, and I think that’s more than enough reasons to read and to study history.”
As well as writing the book Origins of Digital Computers, Brian is also recognised for his work in telling the story of Alan Turing and the codebreaking work at Bletchley Park. He is aware, however, that as stories gain popularity, they are at risk of losing their accuracy, highlighting the Imitation Game as an example which he says: “The Imitation Game is an example of what as a film is, I think, a really quite good film, as entertainment, but as history it’s worryingly bad. That seems to be inescapable. Almost any historical incident that you think you know about, if you go and dig, you find it’s a bit different and it’s more complicated than that, often more interesting than that.” He recommends the comic style book by Sydney Padua on Babbage and Lovelace, as an excellent read and considers BBC documentaries to be pretty good. However, he’d like to see “the swing away from internalist histories to be reversed a bit.” Adding: “I think that there are too many histories which are sociological histories of computing, rather than histories of computing itself. There’s a big place for these, but I think the pendulum has swung too far. Part of that pendulum, and rightly so, was to do with issues of the disregard for the work of the considerable number of women who’ve contributed. But now I sometimes think that it’s almost impossible to write a new book on history unless it’s somehow to do with sociology or feminism.”
Brian tells a story of how, in the 1970s, having written the Origins of Digital Computers, he was invited to speak the Institute for the History and Philosophy of Science in Toronto. He says that having sensed that he was slightly nervous, the director said to him: “Brian, you must understand, there is as much bad history of science written by historians who don’t understand science as by scientists who don’t understand history.”
When it comes to offering advice to people wanting to start a career in IT today, Brian thinks that it is an enormous field into which to jump and knowing where to start could be daunting. He believes that he’s been very lucky in that aspect in that throughout his career he has got into projects at the very beginning of the subject, from compilers, to dependability and the history of computing. This has meant that he’s had room to learn and experiment. He says: “So recommending to somebody now to jump into computing is jumping into a very big field and clearly within it there’s all sorts of new things. They’ll have to jump into a small part of something really very big, and which small part is a good one to jump into isn’t obvious to me.”
Looking towards the future, AI and the internet of things give Brian cause for concern. He says that he often challenged his PhD students when they were choosing their research topics, as to whether they would want their grandchildren to know that they were in computing. He cites both the good and bad of the internet, from fake news to good and bad aspects of social media. He says: “I decided that one of the reasons I liked working on dependability was that I felt I could defend that from future grandchildren rather better than I could defend quite a lot of other things I might have worked on.”
Despite these worries, he describes himself as an optimist, saying: “I find it very difficult to imagine how you can be a researcher without being an optimist.”
Interviewed by: Troy Astarte on the 6th December 2018 at Newcastle University
Transcribed by: Susan Nicholls
Abstracted by: Lynda Feeley