Interview with Bob Hopgood

Bob Hopgood was born in 1935 in Baldock, Hertfordshire. His father was initially an apprentice mechanic working on aeroplanes in the RAF. He was posted to the Flying School in Egypt for three years. On returning to the UK and completing his RAF employment, he moved to Letchworth to work as a mechanic for a dustcart manufacturer called Shelvoke and Drewry (S&D). In 1937, the family moved to Wales for eighteen months setting up an S&D maintenance depot. Bob explains: “At the start of the second world war, S&D decided to mothball their operation in Cardiff and my father was called back to Baldock.”

Of the move back to Letchworth, he says: “It was fairly busy place in terms of factories, there was Irvin Air Chute, S&D, and a company locally called the Tab, which was the name used for the British Tabulating Machine Company (BTM). After the war, we found out they were making bits for Bletchley Park Colossus etc that were being manufactured at the Tab. There was a company called K&R (Kryn & Lahy) which took any piece of iron they could find that wasn’t necessary and turned it into steel. The Germans tried to bomb it, but they seemed to fail totally, but quite a lot of bombs landed in and around Baldock.”

Bob attended the junior school in Baldock, his education was frequently interrupted by evacuations to the air raid shelters where they would sing songs. After the war, he passed an exam which gained him a place at Hitchin Grammar School.

At school, Bob says he played most games, adding: “I played football in the first term, hockey in the second term and cricket and tennis in the summer term there. I also played table tennis probably every night of the week, I did quite a lot of sport at school and at university.”

At grammar school he focused strongly on maths, physics and chemistry. Bob says: “I was better at the maths than the physics and the chemistry. I landed up with five A levels, which were maths, physics and chemistry, advanced maths and further maths.”

At the age of fifteen Bob did some work at Kayser Bondor, a factory in Baldock which made lingerie and which had made parachutes during the war. He says: “That was my first excursion with computers. They were starting automating their basic buying and selling part of the business; computerised in the sense that they had about six ladies whose job was to take all the orders and to punch them onto cards using 026 IBM card punches. Eventually they would have a deck of cards which showed which companies had ordered something, and what the things were that it wanted. This produced a small deck of cards. There was a variety of card punch machines, including an IBM 602, if I remember, which could add, multiply and subtract, and you could put a plugboard together which produced a program which either could add up and say these are the things that need to come out of the store or with another program which said we have low stock of these products and sent off a message to the factory saying this lot needs to be produced.

“On the output side, the cheques arrived and there was an equal number of ladies sitting on Friden Flexowriters and they punched in the things that came back from the customer. There was an office at the end which contained a lady who had a very fancy Monroe calculator which could take Flexowriter tapes in and add all of the stuff up. The only problem was that often the money that got sent or the cheque didn’t actually work out exactly as what it should be.”

It was when the money did not add up that Bob, being good at maths, was called in on Saturday mornings to try to sort out what the customer had done. He says: “That was my role initially and later on I used to go there in the holidays and then if anyone was on holiday I would take over their bit of the finance chain. I got to know the ladies on both the 026s and the Flexowriters when we were trying to work out what on earth the person had done to send this money in return for what he should have been putting in. So that was my first venture into anything resembling computing.” Later, Kayser Bondor replaced the IBM 602 with a Leo computer.

Bob did his national service between 1954 and 1956 in the RAF, before being able to take up his place at Cambridge University. He adds: “They made you do the two years’ National Service before you went up to university. I got trained on S.W.B 8s which were long-distance HF transmitters that could transmit six teleprinters’ input all on the one channel. We all got trained down in Berkshire and further on in Compton Bassett, then two people went to Australia, two went to Washington, two went to Singapore and the other two or three of us went to Sheffield.

“I landed up in Sheffield where we were supporting brand new mobile radar systems with some air-to-ground requirements. They were brand new, so we didn’t actually do very much other than sit there and just make sure everything worked. I did that for about half the time in the RAF after I was trained, and then finally got to Watton in Norfolk and there they did have a SWAB 8 transmitter, so I spent the last few months actually working on the system that we’d actually been trained for.”

In 1956, Bob took up his place at Cambridge to study mathematics. He says: “In those days was, basic mathematics involved learning how to do matrices and do this, that and the other, but there was a certain amount of emphasis on computers, so we got taught about trying to solve differential equations (Hartree), etc. There was quite a lot of theoretical stuff in the maths course which were relevant to computing. Wilkes and Miller gave lectures to the maths set. There were one or two practicals in the afternoon, mostly on broken Brunsvigas …. I learnt a lot about what numerical analysis was needed, but that was practically all I got on the computing side.”

After completing his degree, in 1959 Bob went to Harwell to join the Atomic Energy Authority research establishment. Harwell was focused on developing atomic piles to generate electricity, while the Aldermaston establishment, 20 miles down the road, concentrated on weapons development. Bob adds: “There was an offshoot at nearby Winfrith which was doing design for a genuine atomic pile for going in the grid, and in the north there was at Risley, a site which did the actual work of getting piles running and getting them out into industry and running up. So that’s how the thing was laid out, so Harwell was the peaceful long-distance research bit.”

Bob was recruited into the computing section. He says: “We had a Ferranti Mercury. It was small, only 1000 words of memory, but it was amazingly quite fast compared with its competitors at that time. We mostly used the Mercury Autocode for coding problems. So I did lots of little programs on Mercury for about the first six to nine months, and then I got involved in some larger projects.”

Bob also had remote access to the IBM 704 housed at the Atomic Energy Authority at Aldermaston. The machine was changed over the years to the IBM 709, 7090 and the IBM 7030 Stretch. Harwell would send their card input and outputs via taxi.

Bob explains one of the early projects he was involved with: “The research group at Harwell got the remains of any pile fuel rods that were being used professionally for testing. So when someone had finished with a fuel rod they picked it up and then they had to work out what was in it if they were going to do an experiment with it. To do this they had a program which came from Chalk River. It ran ever so slowly and as a result it was using a lot of Mercury time, and they asked me if I could make it run faster. The program was on a 5-hole paper tape which included some confidential data so any changes had to be made by punching some additional instructions on the end of the paper tape.”

Having succeeded in improving the speed of the Chalk River program, Bob next moved to work with the Quantum Chemistry Group who used the Quantum Chemistry Program Exchange (QCPE) program called MIDIAT. MIDIAT was a large Fortran program for the IBM 709/7090. He adds: “I worked on that for probably a year or so, mostly looking at manganese oxide for some reason. I basically changed it so that it would work on larger molecules than the ones that it was initially being used on.”

After the first few years at Harwell, Bob was seconded to the Aldermaston site for a year. He explains: “Aldermaston had got the IBM Stretch and Harwell was working towards getting an Atlas. The trouble with the Stretch when it arrived was the assumption that it was mainly going to get programmed in machine code, and Aldermaston had made the decision that as they kept getting new machines and different clients, they were going to concentrate all of their future programming on Fortran. A decent Fortran compiler came from IBM, but it was so slow at compilation that it was running about a quarter or a fifth slower than the 7090, which was hardly progress, so they did a quick dirty Fortran compiler called S1, which compiled very quickly, but didn’t produce great code. However, it allowed them to move a load of programs from the 7090 onto Stretch. They then decided to make another compiler, the S2, which would be an all-singing, all-dancing optimised compiler for Stretch.

However, they didn’t have all the staff they needed so I got seconded to help write the S2 Fortran compiler which is how I got into the area of computing that I eventually landed up in.”

Bob next moved to the new Atlas Computer Laboratory which would supply computing power to Harwell and Rutherford Laboratories plus university users who required more computing power than was available at their local university.

Bob explains: “The Atlas Lab was halfway between the Harwell part of the Chilton site and the Rutherford part. Harwell had made a decision that they would write a Fortran compiler, because the University at Manchester that was helping Ferranti build the Atlases wasn’t interested in Fortran. Manchester produced compilers for Atlas Autocode, Mercury Autocode and things like that using the Brooker-Morris Compiler-Compiler (BMCC) system they had developed. They’d decided they were going to produce an ALGOL compiler. ALGOL was a funny language in the sense that it was very elegant and ideal for teaching purposes, but it had a number of unusual features, like it had no input/output commands and the format of reserved words was not defined.”

As a result of these unusual features, each manufacturer has their own version of ALGOL. This took Bob and Bart Fossey to Manchester to see if the Manchester Algol Compiler could be used to develop an ALGOL compiler for all of the Algol dialects in use in the universities, including Stantec Zebras, Elliott 803s, KDF 9s, etc. Bob explains: “We went up to see the ALGOL compiler that was being produced by the Brooker-Morris system. They had a two-stage compiler that did the lexical analysis for their ALGOL dialect in the first stage and then the actual compilation of the Atlas code in the second stage. If we could jump in to the second stage after they’d done their bit for the seven-hole paper tape that they were using and put a front on it.” The university initially resisted the offer but eventually after several more conversations decided to allow Bob and Alex Bell to try, he explains: “So they wrote the basic ALGOL compiler for Atlas and Alex Bell and I produced a version that would accept any dialect, but which was mainly KDF 9 and Elliott 803 Algol programs, and then injected into the second path of the ALGOL compiler that had been made at Manchester, and then we put an outer block round the ALGOL program where we had all the IO procedures for whichever IO system was needed, and that worked pretty well.”

In 1965, Bob, Bart Fossey and Bob Churchouse attended the IFIP65 Conference in New York and took the opportunity to visit several US universities including MIT in Boston, CDC in Minneapolis, Stanford where they met Niklaus Wirth, who was developing Euler and later Pascal, and John McCarthy, LISP. They also visited Carnegie Mellon where Herb Simon and Al Newell were working on Artificial Intelligence.

As a result of the trip, an exchange placement was offered and Bob went to CMU (Carnegie Tech as it was then called) for a year from 1966 to 1967. Bob says: “It was good, there were a lot of impressive people around including Herb Simon, Al Perlis, Gordon Bell etc. I went to a lot of courses, and I had to do something while I was there. They were building a great big CABAL compiler-compiler system and it was soon clear that it would take at least two years for them to finish it, so instead I decided I would rewrite the Brooker-Morris Compiler-Compiler for the Bendix G-21 they had, and the idea was that it could get ported to the IBM 360 series later if they ever got an IBM system. So I did that.

“Later, I got round to writing graphical output routines for a language called Formula ALGOL that they were developing…. The only trouble was that they couldn’t print out any of the results, because printing out mathematics isn’t that simple, so I did some work whereby the Philco displays could display the results of the Formula ALGOL system, and that took me quite a while using my G21 version of BMCC.”

After returning from his trip to the US in 1967 /68, Bob began teaching at Brunel University. Bob was invited to help out one day a week for a year by Mike Pitteway who was setting up a computer science degree at the Brunel. After the first year, he was invited to do another and has ended up teaching one or two courses for over thirty years.

Bob adds: “I enjoy teaching. I also liked computer animation so I could use that, initially I taught compiling techniques and then translator writing systems. Eventually I went on to computer graphics and landed up teaching the World Wide Web, so that took me right up to 2000 at Brunel.”

Upon his return to the Altas lab, Bob found that the ALGOL system was running well and had gradually got bigger with over ten dialects and there was also French and Danish versions. The Lab had purchased an SC4020 microfilm recorder.

He explains: “The SC4020 was basically a peripheral with a graphics machine which could draw graphs onto either 16mm or 35mm film, or create some hard copy output, and therefore you got the opportunity and the ability to produce computer animation. That possibility was something that had always interested me so I decided that I’d write a system called GROATS: GRaphic Output on Atlas using the SC4020, and that would allow people on the ALGOL side to be able to write programs to generate graphs or films. Paul Nelson, who’d got the 4020 up and running did the same thing for Fortran.”

As a result of this work, Bob and Paul spent two year giving lectures at universities explaining computer animation on the SC4020. He adds: “I produced some films for the new Nuffield A level physics course on Change and Chance, which was an integral part of that particular course. The new A level course was partly dependent on the computer animation that we did. The OU was doing a new maths degree course for which they wanted lots of animation to be available to them each week. We generated a week’s worth of computer animation in a week to show it was possible”

In 1971 Bob became group leader of the Basic Software group at the Atlas Lab after Bob Churchouse left to become a professor at Cardiff. Bob says: “I was then a group leader for what was called Basic Software and Barbara Stokoe was the one who dealt with the Application side.”

At this point the Lab was looking to purchase a new system and prefered the CDC 7600 but the government suggested it should be a British make and ICL options were brought into the mix. ICL offered a dual processor ICL 1908A and would initially deliver a 1906A. Later, ICL cancelled the 1908A development switching effort to their New Range. However, as negotiations continued, the Science Research Council made the decision to close the Atlas Lab and merge it with the Rutherford Lab. Bob says: “The Rutherford Lab had managed to persuade everyone that they needed a lot more computing and got an IBM 360/195, which was the fastest thing around, faster than Stretch. Rutherford offered 20% of that machine to university users, so over the next few years we were left with getting all of the remote job entry systems that were currently in use to access either the 360/195 or the 1906A”.

Between 1975 and 1979, Bob became part of the Interactive Computing Facility which emerged from the Rosenbrock Report that identified that engineers wanted to do interactive design and there was a need for small systems in engineering departments.

Bob says: “SRC had so much pressure from the engineers that the name was changed from SRC to SERC, Science and Engineering Research Council. The Engineering Board produced the Rosenbrock Report, which basically said what we need is small multiuser systems situated in university engineering departments and we need an interactive computing facility. This was the remit for the software side of the Atlas Lab that moved across. There was a separate Atlas Computing Division which Geoff Manning, the Rutherford deputy director ran, and basically all of the basic software group and the applications software group that Barbara Stokoe and I were running, became part of ICF.

“So one day we were looking after a new FR80 microfilm recorder that had just replaced the SC 4020 and then we were into interactive computing for engineers. We ran the ICF I think from 1975 to 1990. It was quite a long programme, and it basically involved upgrading any of the engineering facilities that existed until they worked better, then adding multiuser systems in all the main engineering departments to make sure that they’re capable of running six to eight people at a time doing interactive graphical programming; that was the ICF remit, and to get the applications side of it so that everyone else was going in the same direction.”

For the first three years the aim was to find the right computers to support the overall aim of the ICF, many makes were considered with the Department for Industry pushing hard for the use of a GEC British machine and not a US Prime one, both of which were under consideration. Bob adds: “In the end, we bought Primes and GECs. I guess there was maybe 20 systems that eventually went into universities. It was a big programme and it concentrated on putting those machines in and getting them working. We had a network to accommodate the many machines trying to access the 1906A and the 360/195 and remote job entries here, there and everywhere as well. We started off with a load of independent little bits of a network which eventually got all put together to become the academic network for the UK.” The network was the precursor to the current JANET network.

In the late seventies, with Godfrey Stafford, head of the Rutherford lab, and Bill Walkinshaw, head of the Rutherford computing division, retiring, a decision was made to merge the Atlas and Rutherford computing divisions. Bob took over the newly merged computing divisions. He says: “I took over all of the computing at Rutherford for a period, four or five years. We’d moved all of the computer equipment into the Atlas building by then, so the mainframes consisted of two oldish IBM 360/195s front-ended by an IBM 3032, which needed to be replaced. So that was one side of it and the other side was the distributed computing systems programme, run by SERC, that was still looking, as the ICF was, to gradually move away from Primes for interactive computing into single user, personal computers, as you’d call them today.”

The 360/195s were replaced initially by the IBM 3032 and an IBM 3081. Bob continues:

“We’d been talking to ICL for a while still after the 1906A and they were trying to get into bed with Fujitsu and they came up with an idea that they would rename the Fujitsu FACOM machine to an Atlas 10, which was ten times an Atlas. The DTI were very enthusiastic, money was found, we got a good discount on it, and so we landed up with a 3081 in front of an Atlas 10 and also a mass storage system that gave us the ability to effectively keep most of the processing that we needed in terms of tapes and that online for a week, which meant that the number of operators that we needed went down by one or two per shift, which is probably ten man years.”

The IBM and Atlas 10 machines were compatible and they were also able to run a Unix system which was in use. Bob adds: “The Fujitsu machine was probably the fastest thing around, apart from Vector processors, so it was a significant advance in terms of computing power for the department, and it served us well for a few years, as a result.”

During the period Bob was in charge of the Computing Division, the size of the Division reached 200 people partly due to the addition of networking, the support of ICF and new initiatives. In consequence, it was decided to split the Division into two, one being responsible for the Engineering Board activities and the other responsible for the central batch mainframes and networking. Bob became Head of the new Informatics Department that was responsible for the Engineering Board activities including some that were previously in the Technology Division.

On single user system project, Bob says: “We came to the conclusion that if the Primes could get a bit smaller you could eventually afford to have a computer of quite high power as a personal computer on your desk. The trouble was no one could do such a thing.” However, by 1979, the rumour mill suggested that with processing power increasing year on year, the breakthrough in personal computers was very close.

At the Seillac II Workshop, Bob heard that the future personal computer was going to be coming out of Three Rivers in Pittsburgh called the PERQ. Several attendees had Letters of Intent to purchase an early PERQ; CMU were hoping to purchase 10. He says: “There’d been some attempts by IBM and Apple, but it was so Mickey Mouse compared with what people wanted, it wasn’t good enough. We started looking at the PERQ and it was impressive, it had an A4 display which looked as good as a sheet of printed paper, and you could move stuff around on it at an incredible speed, and it had a largish disk with quite a bit of memory, and it ran beautifully.” Bob and his team were urged to buy one of the first 10 machines for 20,000 dollars. To ensure they got one of the first 10 produced, the British Embassy in Washington was asked to make the purchase, and have it shipped to Rutherford. He continues: “We got the machine, it had been repacked into a big sturdy wooden case which meant that it was wobbling up and down inside and was bust by the time it arrived in Rutherford.” Luckily, Charlie Portman from ICL who was working on its own single user system called Project Little was able to fix the PERQ over night and had made an estimate of what it would cost ICL to build one and proposed how it could be enhanced.

As a result, ICL, decided they would like to manufacture and sell them in the UK and elsewhere.

Bob says: “It was decided that (a) ICL would make them, (b) they would have the right to sell them worldwide and SERC agreed to buy 200 of them. The only problem was that the PERQ ran POS, which was a local operating system, based on Pascal; basically it was a soft machine, whereby you could actually define the instructions set of the machine itself. It worked very efficiently and ran very well but users wanted to have Unix on it and so after a couple of years we had 200 PERQs and most of them were running Unix, but it took a lot of effort.”

Unfortunately, after ICL’s early success of manufacturing and selling the PERQs, competition from around the world began to grow as more and more companies started to make personal computers, including Apollo, and Sun Microsystems. Bob adds: “As always, once you get a bandwagon effect from the USA, the original one doesn’t get the speed of take-up that you would want and so eventually ICL stopped selling PERQs and it was probably a glorious failure in some sense, but it did mean that probably 100 or more groups of people in the UK knew what a personal computer was and how to put them together.”

As head of Informatics, Bob saw it grow from 40 to 100 people when they inherited the magnet design people from Technology which brought all of the engineering board activities into the one department. He says: “We were running that pretty well and we got involved with magnet design, finite elements, and VSLI design with some ESPRIT projects.”

However, things changed slightly when the Interactive Computing Facility (ICF) project ended and was replaced by Engineering Applications Support Environment (EASE) developed by the Engineering Board. Bob explains: “This meant that instead of telling universities what they should buy, we turned into a ‘Which?’ outfit looking at various things and reporting on them. It was less exciting for us.”

When the Engineering Board ran short on money, Informatics was forced to lose 40 people, however, they did not make anyone redundant and instead put a freeze on Rutherford computing recruitment and therefore when computer experts were needed anywhere else in the organisation, Bob lost some of the team who were redeployed. Bob says: “Informatics got a lot smaller which was quite interesting, but not very fulfilling in some ways. But eventually it was decided that Informatics couldn’t last any longer and we merged the two departments back together again. So at that stage I was doing more activities outside of Rutherford as well as those on site.”

In 1981, as the Fifth Generation Computer Project was launched in Japan, the UK launched the Alvey project, which consisted of four streams: Software Engineering; Intelligent Knowledge-Based Systems; Man-Machine iInteraction; and down on the hardware level, VLSI Design.

Asked what he thought of the project, Bob says: “It was the right approach and they got the areas about right as well. We got involved with the Roberts Panel that was an earlier attempt at it. But it was clear that knowledge-based systems were appearing and there was a future there, just as like it is today with AI that most people believe.

It was clear from the personal computing side that interaction between man and machine was going to change quite dramatically. In addition, there was no great standardisation or understanding of what was the best way to do things, there were a lot of possibilities including; local area networks, ethernets or Cambridge Rings, etc.” The Rutherford Lab was asked to do the infrastructure for the Alvey Project. Bob says:

“The aim was that most of the Alvey stuff was going to be done in industry with research in the universities, and they thought they really ought to have a standard infrastructure and the decision was made that it should be Unix, so we landed up by buying a load of machines for them. Quite a few GEC machine running Unix formed the basis for many of the projects that ran. They were also interested in having some of our staff help out, so we had projects or just went round and looked at the researchers after they’d been awarded.

“As well as putting the machines in, we had a hardware pool which allowed users to put together things like additional tape decks etc onto people’s existing machines so that groups could talk to each other, we also did some Cambridge Ring and ethernet developments; we put in all the side things to go around the researchers main facility. It worked out pretty well, we had quite a bit of manpower that was involved in one way or another with Alvey. We had grants for projects in the IKBS, MMI and Software Engineering areas. I personally didn’t get much involved and the people who worked for the projects reported to someone in London, so I used to go and talk to them every now again to find out what was going on.”

Bob’s next role was as Chair of the Executive Committee of ERCIM, an organisation that Rutherford had joined as the fourth member alongside CWI in the Netherlands, GMD in Germany, and INRIA in France, the leading government funded IT organisation in each country.

ERCIM set out to expand its cover across the whole of Europe and so Bob and representatives of CWI, GMD and INRIA spent time touring Europe to talk to potential ERCIM members and invite equivalent organisations to join.

He says: “We went to Greece and decided that FORTH in Crete was the right one. Then we added SINTEF in Norway, SICS in Sweden, VTT in Finland, SZTAKI in Hungary. Eventually we’d grown to about 10 or 12 members, the Directors met once a year and at the same time there were ERCIM working group meetings that used to come together and have a separate little conference.

“With this amount of activity, ERCIM decided that they needed someone to run this and so they set up an Executive Committee, one person per organisation, and I took over as the first chair of that.”

Speaking about ERCIM’s contribution, Bob says: “ERCIM had at least one, if not two, partners in 50% of the awards of research grants in ESPRIT, so it was a significant activity. One main thing it did was to try and get young researchers around Europe to work on projects together.” As a result of asking the European Commission if they would help fund this work, ERCIM was encouraged to become a European Economic Interest Group (EEIG) which would allow ERCIM to gain research grants as a single organisation.

Bob adds: “So all the ERCIM partners, apart from one, joined the ERCIM EEIG, and then we had an enormous grant from the Commission which allowed us to recruit researchers for one of the many projects that we were running and if there were two or three members in the same project then the guy would spend six months in one, six months in another and six months in a third. We did that for quite a while, and it really did mean that there was quite a lot of inter-movement between the various members. We set up working groups and that was how we got into the World Wide Web as well.

“The World Wide Web in the first four years didn’t do very much. There were less than 500 websites worldwide, but Rutherford was 12th on the list of websites initially.

“Tim decided to move to the USA and to set up the World Wide Web Consortium, W3C, and the idea was to have three major hosts: one in Europe, one in America and one in Japan. It was clear that the European one was going to be in CERN, the American one was going to be funded by the US government at MIT, and Keio in Japan was the third host and those three hosts would effectively run the World Wide Web.”

However, CERN pulled out as they were too busy with the Large Hadron Collider, so it was decided the INRIA would become the European host with Jeff Abramatic as the head of the W3C European Host funded by ESPRIT (WebCore). Bob continues: “The first thing Jeff did was to come to a meeting of the ERCIM Executive Committee and ask for help, so Rutherford, GMD, CWI, and a few others, provided early support for W3C activities. We provided the XML support for the W3C browser and also the Computer Graphics Metafile (CGM) Profile for the Web, which was a similar transfer body and an ISO standard. The only thing good about it was I wrote the CGM Web Profile in HTML, which is, I think, the only ISO standard that was defined in HTML rather than something else. It was probably the largest HTML page up to that point.”

With all the ERCIM members joining the W3C as well as the EEIG itself, they all became local W3C offices with Rutherford acting as the W3C office for the UK. Bob adds: “I spent two years or more wandering around England telling everyone how marvellous the World Wide Web and W3C were, encouraging people to join. A follow-on project to WebCore was W3C-LA (Leveraging Action) which was funded by ESPRIT, and lasted about three years between 1997 to 2000. We tried to increase W3C membership in Europe. We talked to pretty well everyone; the BBC who were interested, BT, they were interested, Lloyds showed some interest, the pharmaceutical industry was interested and joined, Reuters joined, a lot of universities joined. It was a real mixture. Similar kind of things happened around Europe, and the amount of uptake in terms of W3C members in Europe was much better than in the USA for a while mainly because we were better organised with W3C-LA and we had W3C Offices right across Europe.” (A third ESPRIT project (QUESTION-HOW”) was awarded just before I retired.)

Bob retired from Rutherford in 2000 and took up a role working for W3C as Head of Offices, paid by INRIA. He says: “It worked out reasonably well. I wandered around opening other offices for W3C that they needed.” These included Offices in Morocco, Israel, and Australia. When W3C decided Bob would need to work out of the MIT in the US, he decided to retire.

Asked if quantum computing will come to fruition, Bob replies: “Quantum computing will do certain things and anything that you can do by a lot of randomness and working out, it has a use in that area. However, I can’t see it ever taking over in the mainstream, having said that, if you look at some of the things going on, they look very quantum theory-ish, if you like. So the use of neural nets and stuff like that is similar in some ways. So, I don’t think it’s going to ever be mainstream, but I may be wrong.”

Asked about the threat of quantum computing to security, Bob says: “Everything is a threat to security. I don’t think they’re the only ones. I can think of a lot worse.”

“If you have an Artificial Intelligence system then the first thing it must do is explain to you how it works, that’s the definition of an Artificial Intelligence system. That’s why we call them Intelligent Knowledge-Based Systems, not AI, which probably stands for Artificial Idiot.

“It’s a fundamental requirement and pretty well everything that is labelled AI at the moment is not AI, in my definition of the term …. To be fair to the AI people, there are some attempts at trying to get the explanation side sorted out. When we were doing IKBS you did that first and most of the systems that we put together explained why they did what they did, and the same thing’s needed for the current set of AI so-called systems around. Once you get to XAI (Explainable Artificial Intelligence) you could at least say that the decision we made on this or that is because (a) these two, backed up by this data suggests that you might be one of the people who are affected. That needs to be put out with whatever the decision is, otherwise it’s sheer lunacy.”

Asked if AI is at threat to humankind, Bob replies: ” I think everything at the moment that’s being suggested is nearly a threat. It’s incomprehensible the things that are going on at the moment. I just can’t believe anyone would be that silly to say, (a) do this and that is dependent on something that they know nothing about; it must be controlled in some way. Putting an ‘X’ in front of AI on everything that’s allowed or talked about is the starting point. Everything must be explainable, and explainable to the individual involved, and if you can’t then you should ban it.”

Asked why public IT projects are not always as successful, Bob says: “You could argue it’s a software engineering problem. If the things were tested and made absolutely certain that they did what they said they were doing, then you wouldn’t have the problem.

“Has software engineering got better or worse – the answer is almost certainly it’s got worse, and the reason probably is because 90% of the code that’s running in the world is probably in JavaScript. Normally each JavaScript program’s got about ten different libraries that it’s running, all incompatible with each other, and the stuff goes out and around, and why any of it works is a miracle. I notice the US government has started moving back and saying that in our systems only Vanilla JavaScript (Library/framework free Javascript) will be used.”

Asked about the biggest failures in his career, Bob says: “I can’t think of any great ones. You could say it was being part of the team who were involved in buying 200 PERQ machines; that certainly didn’t work out the way we anticipated. We anticipated it would be the start of ICL being a major player in the personal computing area and that never happened, instead they got bought up by Fujitsu.”

Interviewed by Richard Sharpe

Transcribed by Susan Nicholls

Abstracted by Lynda Feeley