On November 29th of 1996, Prof. Roger Needham of the University of Cambridge received an honorary doctorate for his contributions to computer systems engineering, made over a period of more than thirty years.
Needham attributed this partly to serendipity. Serendipity, in his own definition, is `looking for a needle in a hay stack and finding the farmer's daughter'. But it is too easy to say that he just happened to be in the right place at the right time. We are all in that position lots of times, but we rarely realize it at the time; sometimes we realize it much later, but then it is too late.
Needham solved several important problems before others even realized they existed. He never looked for a needle in a hay stack; he's always been looking for the farmer's daughter.
I first met Roger Needham in 1980 or so. I was a Ph.D. student at the time and I was somewhat apprehensive to meet one of the grand old men of computer science. He turned out not to be intimidating at all and immediately invited me into that hallowed place where he carries out much of his best research -- the pub.
Later, when I did a sabbatical in Cambridge, I discovered the vital function of the daily hour in the Eagle. The whole systems research group, both staff and students meet there for one or several pints of India Pale Ale. Here, the progress and frustrations of the day are discussed by a group of extremely bright people and, magically, the pub is the place where some very clever ideas are born.
Needham and I were among the eight lecturers in the Advanced Summer Course on Distributed Systems. The first of these courses was held at the University of Tromsø, in 1988, under the midnight sun. It was a great success and the course was repeated five times, in Europe, the United States and Japan. The last of these courses was held in Seattle, in 1993, at the research labs of Microsoft. Soon after that course, Microsoft started selling Windows/NT, which is a vast improvement on MS-DOS by anybody's standards.
Needham's lecture on naming in distributed systems was, and still is, my favourite in that course. I now teach it to my students and they are always delighted when I repeat Needham's words that distributed name servers must always be accessible and updatable and that it would be nice if they were consistent and correct too. Finally, they have it on authority that programs need not be correct.
Needham stood at the cradle of at least four important areas of research: computer networks, distributed systems, computer security and multimedia and he has contributed significantly to each.
He was involved in the construction of one of the first local-area networks, the Cambridge ring and its successor, the Cambridge Fast Ring. In the middle 1970s, the Cambridge Ring already used some of the features that are now hailed as revolutionary in Broadband ISDN networks.
Distributed systems concerns the analysis and construction of computer systems with multiple independent interconnected processors in which a failure of one component can be masked by other components. It is the study of computer systems that are more reliable than their parts. Needham worked on the Cambridge Model Distributed System and two famous distributed naming services, Grapevine and GNS.
His work on computer security concerned the design of the Cambridge CAP computer -- a machine with hardware access control for objects -- the first protocol for authentication in distributed systems and a logic system for the analysis of such protocols.
He encouraged research on multimedia systems in Cambridge and it became one of the first research laboratories in the world where teleconferencing and video mail became regular tools for research.
Needham would say that he was not the first person to work in any of these areas, that his work on the Cambridge Model Distributed System was inspired by work at Xerox PARC, and that his work on the Needham-Schroeder authentication protocol in 1977 only started because Schroeder made him do it.
And Needham would probably claim that the reason for him to work on a logic that helps one to reason about the correctness of authentication protocols is merely that the Needham-Schroeder Protocol of 1977 was wrong and that it took eight years to discover this, even though the protocol has only five, really simple messages.
Here, I must pause for a confession. Needham and Schroeder point out the error in their protocol in an issue of Operating Systems Review (next to an article by Otway and Rees, by the way, that presents an authentication protocol which has only four messages and which is absolutely correct). In that article Needham and Schroeder say that I pointed out the error in their protocol during a discussion with them. The truth is, I did no such thing. I never realized the protocol was wrong until I stumbled on their paper in 1985.
Almost without exception, Needham carried out his research in a team. Often, of course, with colleagues and students in Cambridge, but he has also been exceedingly productive doing research with members of Xerox PARC and later with members of the DEC Systems Research Center, and, off and on, with others as well. Many of his research partners did their best work teamed up with Needham.
He combined a passion for systems engineering with what appears to be complete emotional detachment from any solution at hand. This allowed him to keep the problem he is working on in a proper perspective. I think it is this that helped him steer a team towards the essence of a problem and to keep it from getting stuck on irrelevant detail.
Needham worked as a catalyst. When he was around, systems research got more focus and more vision. He brought out the best in the people around him. This helps to explain why, for as long as I can remember, the Cambridge University Computer Laboratory has been among the best systems research laboratories in the world. This is recognized even by Americans, although their national pride doesn't always allow them to admit that MIT, Stanford, Berkeley, Cornell, and the rest of them, have something to learn abroad, in Cambridge.
Needham liked to explore new fields in his research. He called it doing research with a spade. It paid off; We are grateful that Roger Needham was our colleague, that we could learn from him, and that he has helped to make computer-systems engineering a respectable and important field. He more than deserves the honorary degree bestowed on him in recognition of his lifetime devotion to computer systems engineering.
RogerNeedham, at one time, dabbled in digs as an amateur archeologist. I'm glad he recognized in time that research with a spade, in archeology, usually keeps one very close to the surface. In computer science, his digging has been deep and it has uncovered many rare treasures.
Roger Needham died on February 28, 2003, after a brief fight with cancer.
Sape J. Mullender