Interview with Professor David Duce

David Duce was born in 1950, the family lived in Pudsey, halfway between Leeds and Bradford. His mother was a housewife, and his father was secretary of David’s maternal grandfather’s group of companies. He says: “My grandfather’s family background was in quarrying, and then he moved into building along with another brother.  They built the house where we lived, and also built a small number of cinemas which were operated as a set of companies.  My father was company secretary to that group of companies.”

David, having passed his eleven-plus attended Pusey Grammar school where he enjoyed studying for A levels in chemistry, maths, physics, and general studies. He says: “The most famous former student in chemistry was a man called S. F. Boys, who was a well-known theoretical chemist in theoretical chemistry circles. I took chemistry, it was a subject, I took to very early on. It was probably a combination of making things: smells and bangs were always good because you were allowed to do that sort of thing in those days; and then the theoretical side of chemistry attracted me as well.”

In 1968, having competed his A levels, David went to the University of Nottingham and gained a first class honours degree in Chemistry. After completing his degree, he went on to gain a PhD in Chemisty.

David’s introduction to computing was via a short summer course in ALGOL 60, organised by Eric Foxley, Director of the Cripps Computer Centre.

Asked if he has a favourite computing language, David says: “I don’t really have a favourite language, it’s more a question of deciding what it is you want to do and then choosing the appropriate tool for the job.  I’ve programmed in Java and JavaScript for web-based things, for server-side things in Java, client side in JavaScript and I’ve done bits and pieces of Python. It tends to be dictated by the kind of libraries that I want to use, because I’m not a ‘build it all from scratch type of person’.  I like to find the bits and pieces and then put them together to achieve the task.”

Of Fortran, he adds: “I came across Fortran when I was doing my PhD to which there was a computational aspect because we were looking at molecular force fields and trying to reverse engineer force fields from spectral data. The software that the lab was using at the time was all written in Fortran, so, I learnt Fortran.  I found it a very convenient language for that kind of numerical computation.”

As he was finishing his PhD, David applied for a role at the Atlas Computer Laboratory where he joined Bob Hopgood’s Basic Software group. He says: “I was faced with the question of whether I stayed in chemistry or moved into computing, because computing had begun to fascinate me.  At the time the way into chemistry was through an unbounded sequence of post-doc appointments and there was no guarantee of positions in chemistry at the end of it.  Chemistry was one of those disciplines that went through peaks and troughs in the employment prospects and when I was leaving university it was in a trough.  I was sitting in the library of the Cripps Computing Centre at the University of Nottingham one day, and I pulled up an issue of Computer Weekly and there was an advert for jobs at the Atlas Computer Laboratory. I thought it sounded really interesting, so I applied, and they selected me.”

With the Atlas machine having been decommissioned prior to his arrival, David worked with the newly installed ICL 1906A, George 4 paged machine.

David says of his role: “The Basic Software group was responsible for all the system software on the machine.  There was a graphics component to the work, and there were other groups who were responsible for application software, for user support, operation of the machine and so on. My initial work was working on utilities for the 1906A. Graham Robinson had constructed a job submission and management system called Task, which presentd a much simpler to use user interface than if you were writing directly in the George 4 command language.  Graham then also started work on a utilities system which incorporated a range of the 6A utilities and I worked on ‘COPYIN and COPYOUT’, which were utilities for dumping things to mag tape and recovering from mag tape; basically data transfer. I was programming in PLASYD. I’d learnt a small amount of PLAN assembly language when I was doing my PhD, and PLASYD was a structured version of PLAN, so I learnt that language.”

Around 1976, when the Atlas Lab, which was run by the Science Research Council, became a division of the Rutherford Lab, David moved with it. He says: “We were then part of a much larger organisation and the remit of the organisation was not computing as it had been for the Atlas Lab, Rutherford already had its own computing division which was supporting high energy physics work, so the role changed.”

With the move across to Rutherford, David also became involved with structured Fortran. He explains: “It was an interest that grew. One of the people working in the group at the time was Rob Witty and he had a very nice approach to small-scale software engineering design, a graphical notation for representing software structure and so on, and he gave a seminar and I thought that sounded really quite interesting.  I somewhere found a structured Fortran system which I acquired and then developed. Then working with Rob, we integrated that with his dimensional design system.  It grew out of an idea but turned out to be really quite interesting and something that we pursued.”

On the subject of formal methods, David says: “In engineering design you typically do some calculations first to try and reassure yourself that your design is going to be fit for purpose.  The calculations might not always give you the full answer, but they’ll give you a degree of confidence and help you at an early stage to eliminate some things that will cause you a lot of trouble later on. I see formal methods as being somewhat in that sort of category of helping you to clarify what it is that you’re going to do and giving you some confidence that it is going to do the job that you intend for it to do.”

David was part of the team who worked initially on the procurement and benchmarking programme which followed the publication of the Rosenbrock Report (Engineering Computing Requirements) on support for engineering computing which called for the purchase of a large system, upgrades of existing large systems, and smaller multi-user mini systems which could be deployed in university engineering departments.

He then started working on the basic software provision and identifying what the standard engineering provisions would be.  He says: “That’s how I got involved in the graphics side of things. The packages around at the time were things like GINO-F, and I was involved in porting that to GEC and Prime systems.  We then developed our own in-house smaller system called FINGS – Fortran Interactive Graphics System.”

One of the multi-user mini machines at the lab was the GEC 4070 of which David says: “The Lab’s involvement with GEC range arose partly on the comms side where people were using earlier models of the GEC range as remote data entry devices and communications devices, and it turned out to be a successful machine in that area. Then the 4070 was one of the workhorses then of the multi-user systems programme.”

On the subject of the Prime operating system, David adds: “Prime was fun in those days, because it was written in Fortran.  The great thing was that you had to remember to compile it with the -r flag, which was the recursion option, and if you missed off the -r flag, that was bad news for the future of the operating system! I certainly enjoyed working with that.”

In 1979, David was appointed as Technical Secretary of the Science Research Council’s Distributed Computing Systems Research Programme. He explains: “The programme started in the academic year 1977-78. It was a recognition that distributed computing was probably going to be an important topic in the future, although nobody actually gave a very precise definition of what distributed computing was, which was good.  The programme was launched as a specially promoted programme, and perhaps unusually for the time, it had co-ordinators; the first academic coordinator was Bob Hopgood and Gill Ringland from ICL was the first industrial co-ordinator, and Rob Witty was the first technical secretary.  They were working in support of a panel of the SRC Computer Science Committee. The role of the panel was to award grants and the role of the co-ordinators was to help people to put forward good grant proposals and to help shape the form of the programme.  It was a success and ran until 1984.

 “Many of the grant holders went on to do very good work in the field. Roger Needham, Robin Milner, Tony Hoare, Samson Abramsky, Gordon Plotkin, were all early grant holders.  There was a long list of grant holders and for some of them it was the start of their careers, their first opportunity.”

In 1982, David was appointed as Academic Co-ordinator with responsibility for the technical management panel of the programme. Asked if he enjoyed management, he says: “To a degree. I didn’t make a career in management.  I was still more interested in technical things as well. Management has its own set of challenges, but I didn’t see myself as a career manager.”

Having been involved at an early stage in the development of standards for computer graphics, David joined the European Association for Computer Graphics (Eurographics) when it was launched in 1980 and became chief editor of its newsletter and involved in developing several of its conferences. He explains: “The graphics standards community in Europe developed the idea for the European Association for Computer Graphics, which was founded by Professor José Encarnação in Germany, Bob Hopgood was the first vice-chair of that.  I got involved in the Association partly because, I’d gone to one of the conferences prior to the founding of the Association, and then I went to the conference in Geneva where the Association was actually founded. Amongst the member benefits was a regular newsletter.”

Having realised that he wasn’t receiving the newsletter and complained to the vice-chair, Bob Hopgood presented David with 40 pages of manuscript and asked David to typeset it. David continues: “That was kind of how I got involved in Eurographics.  Then I got involved with the Executive Committee and the newsletter, and after four issues, it turned into a journal, Computer Graphics Forum, and I was the co-chief editor of that for a period. I got involved in organising the Eurographics Association’s conferences as well.  It was a thriving young community, there’d been nothing of its kind before and it was a really dynamic, fascinating time.”

Asked if software and engineering go together, David says: “It depends how you think of engineering. I think of engineering as a principled way of constructing effective artefacts that are fit for some specified purpose.  Engineers might not think of it in that way.  It’s about producing things, but it’s about having a set of techniques for doing so, about being able to produce things of quality, about being able to know what quality is, being able to measure quality, and having firm foundations on which those techniques are based.  I would like to think that those considerations apply to the construction of software, so in that sense I think software and engineering do belong together.”

 Saying that software engineering has improved, he adds: “There’s a long way to go.

We need to continue improving the whole knowledge base, improving the ways that we decide what it is that we want to build, and how we build up confidence that the thing that we’re building is actually going to do the job at the end of the day.  These remarks though are made by me who could not be described as a practising software engineer constructing large systems to impossible deadlines, etc, etc, etc.  However, maybe we need some kind of culture shift to appreciate what it takes to actually build something that is fit for purpose and is going to last in the long term.”

David and his team were awarded an Alvey grant to look into applying ideas of formal specification to computer graphics, which was an unexplored field at the time.  He says: “That came about partly because of the DCS programme. I was involved in development of ISO/IEC graphics standards at the time and our working methods there were formalised, but not formal, we didn’t have any good notations for writing down and reasoning about what it was that a system was doing.  We made extensive use of structured English for expressing what functions should do and so on, but we didn’t have more formal ways of doing that. The British Computer Society (BCS) Formal Aspects of Computing Science group used to have a winter seminar at Imperial College and one of those was a talk by Cliff Jones on VDM which I thought sounded interesting. In the Distributed Computing Systems programme we had an annual conference and one of the speakers there was Tony Hoare, who talked about the Z method, and I think that was the first that any of us had heard of Z at that time. So that sounded interesting as well and I wondered if those methods had applicability in computer graphics.  That’s how I got into formulating the grant application, along with a colleague, Liz Fielding, who’d done the MSc in software engineering in Oxford and had done her dissertation with Cliff Jones, and knew VDM.”

 Asked if he considered the Alvey project a success, David says: “It was Roger Needham who once said something like technology transfer is best done on the hoof.  The Alvey Programme brought people together in a way that perhaps hadn’t been done before, and engendered new co-operations which probably bore fruit in many ways after that.  So from my perspective I thought that was a successful programme.”

In 1990, David was the acting head of Systems Engineering Division in Rutherford Appleton Laboratory Informatics Division, responsible for a research group of 18 researchers in knowledge and software engineering.  He explains: “Knowledge in those days was intelligent knowledge-based systems, which was one of the pillars of the Alvey Programme, and expert systems were one of the topics of the moment in those days.  So we had a small group of people at the lab who were basically supporting the IKBS pillar of the Alvey Programme.”

During this time, together with Gordon Ringland, David edited a book called Approaches to Knowledge Representation.  He adds: “That grew out of a series of teach yourself seminars that people in the group had given on different approaches to knowledge representation from use of logic, structured frames, etc, and we pulled that together into a book, which I think was a successful venture.”

At this time, David was awarded the individual merit promotion to Grade 6 (Senior Principal Scientific Officer). According to Bob Hopgood, David was the first computer science professional in the laboratory to be awarded this grade. The promotions were awarded on the basis of individual merit that enabled the holder to pursue a course of research for a period of five years.  David adds: “They were non-competitive in the sense that there wasn’t a quota of individual merit promotions, but you had to meet the criteria for them.”

The ARGOSI project (funded by the Esprit programme) involved the integration of graphics and OSI standards. David explains: “It was the first time our department had gone after European funding. The ARGOSI consortium was a combination of people who knew each other having all worked together in graphics standards from different organisations in the UK and across Europe, and we had some commercial partners, the main one being Thomson-CSF in France.

“There were basically two strands to the project: one provided funding to enable continued participation in standards making activities; and the second was building a demonstrator of how you could put graphics standards and OSI standards together.”

The topic the group chose was to display hazards along a route, say from Rutherford to CERN. David continues: “We had a map, a set of hazards that were represented as various icons and additional information, and they were represented as segments of a computer graphics metafile which was an ISO standard, and then at points along the route you would select any relevant hazards and display them on the map.  They were pulled from a computer graphics metafile using a facility in the file transfer protocol that allowed you to access a segment of a file.  So it was basically representing hazards as graphical structures and then being able to pull them down selectively using the file transfer protocol.  This might sound slightly reminiscent of some applications that have since taken off in the commercial world, but this was prior to the worldwide web and to quite a lot of other developments. Did we have the right idea – yes, did we realise we had the right idea – probably not.”

On the potential of commercialisation of these types of projects, David says: “One of our partners in the ARGOSI project was the University of East Anglia which held a showcase evening of the computer science department for local industry and the general public, and we gave a demonstration of what we could do with ARGOSI. One guy came up to me and said ‘I run a truck haulage company, when can I get it?’ I just did not have the mindset to twig what he was saying.”

In 2000, as he approached 50, David decided to move out of Rutherford and started looking for another role. He chose to move to Oxford Brookes University. He says: “I hadn’t done too much teaching up to that time so I was interested in the teaching side of things and I fancied combining teaching and research. Most of my teaching was done at MSc level.  I did less at undergraduate level.”

In 2010-11 David became the Acting Head of the Computer Science Department and started to work on how the University could simplify our courses and regain recently los BCS accreditation for its computer science degrees.

David explains: “We lost the accreditation for two reasons.  The first was that the structure of our programmes at that time did not fit the pattern that BCS were looking for.  One of the strengths of Oxford Brookes was the modular degree programme which gave students very wide flexibility in the modules that they could study and there were university frameworks that stipulated how flexible you had to be. So the degree programmes were loose in that sense. The BCS requirements just did not fit within this wider university framework.  . If you tried to do the BCS module mapping of skills, accreditation criteria onto modules, you ended up with something that was very difficult to deal with, because it then becomes very hard to say, well, all these students have this capability, because of the flexibility of the programme.  The other issue was probably we didn’t understand that well enough, and that led to a bit of a disconnect with the panel at that time.  I then got involved as head of department in how we got out of that.”

One of the elements that helped contribute to the regaining of the accreditation were the wider changes at Brookes. David continues: “There was a general move in the university to condense programmes.  That actually played into our hands, because then we could come up with a set of programmes that were much more specific.  We came up with a set of programmes where we had a common core through the three years of the degree, and then with options in each of the years, which were then tailored to particular degree titles. That was a framework and a set of programmes that was much easier to present to the BCS, and by that stage I’d become a BCS assessor and I understood a bit more about how BCS works. We regained BCS accreditation.”

David was appointed Honorary Chair of Computer Science at the University of East Anglia in Norwich in 1990. He says of teaching: “Rutherford wasn’t a teaching establishment but we did run short courses for the community and I ran quite a lot of short courses on graphics standards and I quite enjoyed doing that.  So, teaching was something that I grew into.  Most of my links with UEA were research links, but the teaching side came later on.”

David was Associate Professor at the Institut National de Communication in Paris. In this role he attended advisory board meetings and gave the occasional seminar in French.

David was a member of the BSI and ISO standards committees in the graphics area, he became secretary to the BSI committee in 1984.

In 1996, David took part in his first RAE panel. He explains: “The panel chair for that round was Robin Milner and the panel consisted of academics and some people outside the academic community.  I was possibly nominated because I knew a certain amount about how academic research was done through my time with the DCS Programme and how academic research had evolved and so on.

“It was fascinating work. We had a sizeable number of institutions, and we reviewed their submissions which were all on paper. It was a huge amount of work, but it was fascinating because it meant one was forced to read stuff one wouldn’t normally have come across.  I found out all sorts of fascinating pockets of research that were being done that I just had no idea about. Then one had to reach a consensus on how one was going to rank the submissions.”

One fascinating project that David recalls from RAE 2001 was from the University of Kent where Richard Jones wrote the definitive book on garbage collection which in extended form is still the definitive text today.

Asked if he experienced any issues with any of their work during Y2K, David says: “We didn’t have any direct problems in that. One could quite see why people could well have had very serious problems.  Was it an equipment selling opportunity or a pending technological disaster?  Who knows?  But we did encounter date problems in a later project, so dates were certainly no laughing matter.”

On the subject of AI, David says: “I’m personally not a fan of AI. On one level it’s such a slippery thing, and I’m not a fan of the hype around it and perhaps the general awareness of what it is and what it isn’t and nowadays what machine learning can and cannot do, and what the basis is. My personal preference is for things that are much more model-based where you have an understanding of why the system is doing what it’s doing.  I’m less enthusiastic about things that give me an answer and can’t tell me why they think it’s the answer.  It’s a field I haven’t worked in, so I speak as an outsider to the field.”

Interviewed by Richard Sharpe

Transcribed by Susan Nicolls

Abstracted by Lynda Feeley