[Published originally in the September 2004 edition of Computing Research News, Vol. 16/No. 4]

The UK Grand Challenges Exercise

By Tony Hoare and Robin Milner

A Grand Challenge for scientific research pursues a goal that is announced one or two decades in advance; its achievement is a major milestone in the progress of knowledge or technology, celebrated not only by the researchers themselves but by the wider scientific community and by the general public. An essential part of the Challenge is that it should define its own clear criteria for success or failure. It has no guarantee of success. A Challenge that failed was Hilbert's programme, formulated in 1900, for the foundations of mathematics; nevertheless its failure inspired previously unimagined advances in logic and mathematics by Church, Gödel, and Turing. A Challenge that recently succeeded was the mapping of the human genome; it set directions and provided a completely new methodology for biology in the twenty-first century.

The technology of computing is distinguished both by the breakneck improvement of hardware capacity and performance, and by the world's voracious appetite for new applications. Against this background it is vitally important that the computing research community define the long-term aims and opportunities for their discipline, independently of the push of technology and the short-term pull of market demand. Inspired by a similar initiative of the Computing Research Association in the United States, the UK Computing Research Committee (UKCRC) mounted a Grand Challenges Exercise in 2002. Full details can be found on the website http://www.nesc.ac.uk/esi/events/Grand_Challenges/. The BCS will shortly produce a forty-page pamphlet summarizing the overall outcomes of the Exercise so far.

History of the Exercise

To ensure that the title ‘Grand Challenge’ is reserved for truly farsighted and ambitious scientific goals, the UKCRC began the Exercise by defining a list of stringent criteria for judging the maturity of a Grand Challenge proposal. A Programme Committee was then appointed to organize and conduct the Exercise. It issued an open call to the UK academic computing research community to contribute ideas that would meet the criteria. The initial workshop was held in Edinburgh on November 2002, where 109 submissions were discussed. The workshop identified a set of possible topics for Grand Challenges, and these led, via panel discussions, to seven draft proposals. After the workshop these were mounted on the website for public email discussion, moderated by the champions for each proposal. The discussion is still open.

A primary principle of the Exercise is that, in formulating Grand Challenge proposals, no questions are raised about sources of funding. In this way the independence of pure scientific ideals and judgment can be maintained. A second principle is that the Exercise attributes no less importance to basic exploratory research and to research aimed at more immediate goals than to research associated with a Grand Challenge project. Such a project seeks to unite suitable research directions that contribute to the same long-term aspirations.

The next coordinated step in the Exercise was a Conference on Grand Challenges for Computing Research, held in Newcastle on 29-31 March 2004. Its stated aims were to:

• encourage UK researchers in computing to articulate their views about long-term prospects and progress in their academic discipline;

• discuss the possibility of speeding progress by broader collaboration, both nationally and with the international community;

• facilitate the pursuit of more ambitious scientific and engineering goals; and

• work towards the definition of a Grand Challenge project, where this is an appropriate means to achieve the goals.

Again an open call was issued for submissions. Some fifty were received (remarkably, almost all linked to an existing Challenge), and these were again the subject of panel discussions. The Conference attracted over 150 attendees. The keynote speaker was Anita Jones, who reported on the progress of the US Grand Challenges Exercise.

The Grand Challenge Proposals

The titles of the seven draft proposals are:

GC1 In Vivo–In Silico: The Virtual Worm, Weed and Bug
GC2 Science for Global Ubiquitous Computing
GC3 Memories for Life: Managing Information over a Human Lifetime
GC4 Scalable Ubiquitous Computing Systems
GC5 The Architecture of Brain and Mind
GC6 Dependable Systems Evolution
GC7 Journeys in Non-Classical Computation

Two of the Challenges, GC1 and GC5, aim at modeling life forms. GC1 seeks a detailed and predictive informatic model of plants and animals; it may lead to an understanding of developmental and regeneration processes in organisms, with potentially dramatic implications for disease and accident victims. GC5 tackles a long-established challenge to unify research on the relationship between the human mind and the brain; its discoveries may lead to radical advances in the benefits obtained from computers in society and in personal life. Both of these Challenges will involve close collaboration with life scientists. A third Challenge, GC3, also life-oriented, aims to equip people with personal digital memories; it includes many scientific issues, for example the retrieval of pictorial and musical information. It will explore means to help people manage and use such electronic memories; for example, it may answer an apparently simple request like ‘find a picture of me playing with my nephew Peter when he was a toddler.’

Two Challenges, GC2 and GC4, are concerned with the science and engineering of ‘the Global Ubiquitous Computer’, a useful term to describe the worldwide interlinked informatic machine that we must expect in the next two decades. There will be computers everywhere, embedded even in our bodies; they will number in the billions, and their structure will evolve even after they are deployed in service. GC2 asks what scientific concepts will enable us to understand this entity well enough to control it; for example, how do we assemble models of trust, self-awareness, privacy, and knowledge into an understandable and controllable intellectual framework? GC4 approaches the same topic from a more empirical standpoint. It will formulate general design principles pertaining to all aspects of ubiquitous systems, and justify these principles by instantiation in successful real systems. These two projects will collaborate not only with each other and with GC6, but also with designers of new technologies—for example, wireless communication and sensors.

GC6 is a Challenge that links the theory of programming with its experimental validation and its practical application in software engineering. It aims to answer questions about the safety, soundness, and security of computer programs in general. It hopes to deliver its understanding in the form of software tools, such as a ‘verifying compiler,’ that enable the computer itself to guarantee correctness of its own programs; it aims not only to advance the logical foundations of computing, but also to apply this increased understanding to the problem of inadequate program testing, estimated in 2002 to cost the US economy tens of billions of dollars per year. Lastly, GC7 is a Challenge to explore and to integrate models of computing that depart from the classical von Neumann model; it will consider massive parallelism and nondeterminacy, information flow in biological organisms, and quantum computing. The aim is for a broad science of all these forms of computation.

Each of these Challenges is the subject of a proposal which can be found on the website. The goals of each Challenge are stated with precision and are distinct; yet many research topics will be shared among them. Each Challenge is a mountain peak; many routes to the peaks are in common. The Challenges will also provide stimulus for exploratory research, pursued for its own sake and not necessarily under the banner of long-term aspiration that characterizes a Grand Challenge.

Next Steps

The Grand Challenges Exercise has now reached a stage at which responsibility for further progress is distributed to individual steering committees for each Challenge theme. These seven committees will operate independently, with a small overarching committee to look after general issues. Exploration of each theme is likely to be pursued by a series of workshops, attended by the many specialists whose skills will need to be recruited to the project. The UKCRC will provide overall coordination where appropriate; for example, it may convene a conference every few years on the Newcastle model. This will keep the Grand Challenges Exercise as a whole in the sight of the whole community.

At the end of the next stage in the project, we hope that the proponents of each theme will produce a ‘roadmap’ of opportunities for future progress. Typically a roadmap will define initial projects in detail and later projects more conjecturally, depending on the outcome of earlier ones. For some of the themes a roadmap is already in sight; in other cases it is quite clear that exploratory projects are needed before even the definition of a roadmap is attempted.

No Grand Challenge proposal is guaranteed to mature into a Grand Challenge project. Even if it fails to define a roadmap, or succeeds in this but fails to achieve its ultimate goals, the Grand Challenges Exercise is providing a valuable opportunity for discussion of long-term aspirations for computing research, and for identifying its most promising directions and methods. Scientists who have devoted their effort to this Exercise have welcomed the opportunity to think deeply about such questions, and to take responsibility for the good health of future computing research.

Professor Tony Hoare (thoare [at] microsoft.com) is a Senior Researcher at Microsoft Research Ltd. in Cambridge, UK. Robin Milner is a Professor in the Cambridge University Computing Laboratory, Cambridge, UK. For additional information, see: http://www.nesc.ac.uk/esi/events/Grand_Challenges/