Special Features:

AN OPEN LETTER TO THE COMPUTER SCIENCE COMMUNITY
By Tony Hey, Director of e-Science, EPSRC

The $444 million (250 million pounds) U.K. e-Science Initiative has just held its third All Hands Meeting this year. With more than 600 participants from all the U.K. Research Council communities and from industry, there is genuine progress being made towards the creation of a vibrant multidisciplinary community. What distinguishes the U.K. Initiative from many other initiatives around the world? In my view it is that the breadth of application scientists and engineers participating in applications-inspired e-Science projects is supplemented by a healthy participation of U.K. computer scientists from essentially all of the leading U.K. research groups and by the involvement of some 80 commercial companies. The combination of exciting multidisciplinary and collaborative applications, computer scientists and IT and end-user companies makes for a thriving e-Science ecosystem. As a result, the U.K. e-Science Initiative is clearly among the world leaders in research and development of the networks and services that will constitute the emerging global e-Infrastructure -- Cyberinfrastructure in the United States. With the experience of our U.K. e-Science application projects as a guide, we are attempting to build an infrastructure comprised of a set of middleware services capable of supporting the dynamic "virtual organizations" required by scientists, engineers and industry.

As a shorthand for this core middleware infrastructure, I often use the controversial term "Grid" --which for many conjures up images of hardware ranging from supercomputers to SETI@home. Companies such as Oracle and IBM have their own definitions of the Grid to suit their business purposes. In fact, scientists in Europe and the United Kingdom are widely using Grid to mean a much more general infrastructure, with data integration, federation and mining usually the core of their applications, rather than a dominant focus on compute clusters or supercomputers. My use of the term "Grid" harks back to NASA's early attempt to build an Information Power Grid (IPG) connecting their different sites. Bill Johnston from Lawrence Berkeley National Laboratory, and chief architect of the IPG, summarized its goals as being "to promote a revolution in how NASA addresses large-scale science and engineering problems by providing persistent infrastructure for: (1) the 'highly capable' computing and data management services that, on-demand, will locate and co-schedule the multi-Center resources needed to address large-scale and/or widely distributed problems; and (2) the ancillary services that are needed to support the workflow management frameworks that coordinate the processes of distributed science and engineering problems." These still seem a good set of goals today, three years on. In the United Kingdom, we now have some very promising examples that are using the presently immature infrastructure to attack real engineering problems -- the DAME aero-engine maintenance project with Rolls Royce and the GEWiTTS project on remotely operated wind tunnel services with BAESystems.

The U.K. Initiative started by attempting to understand the problems of actually implementing distributed middleware services crossing institutional boundaries by evaluating the existing IPG software. However, it was clear even in mid 2001 that any future distributed middleware that was to be supported by the IT industry would have to be based on Web services. I am, of course, aware that the whole Web services movement is still very much a "work in progress." The presently defined Web services certainly do not constitute a wholly satisfactory basis on which to build e-Science applications. Nonetheless, in my view, this is precisely why the part of the computer science community interested in building real systems should be engaging with the e-Science application community. The computer science community has valuable insights to offer into the problems of distributed software engineering, algorithms, semantics, dependability, security and formal methods that can benefit the engineers and scientists who are today attempting to build large-scale distributed systems. Furthermore, the insights gained by working with these scientists and engineers, and from understanding their problems in detail, can expose inadequacies and suggest improvements for our computer science methodologies. This engagement of computer scientists with real problems and real systems seems to me to be directly following a great tradition dating back to the pioneering early days of computing in the United Kingdom with Alan Turing and Maurice Wilkes.

The goal of the e-Science program -- to build a viable "e-Infrastructure" to support multidisciplinary and collaborative research and innovation -- can be traced back to J.C.R. Licklider's original vision for the Internet. Larry Roberts, one of Licklider's successors at ARPA, summarized Licklider's vision as follows:

"Lick had this concept -- all of the stuff linked together throughout the world, that you can use a remote computer, get data from a remote computer, or use lots of computers in your job."

Thus, the attempt to build a robust global e-Infrastructure for e-Science -- Cyberinfrastructure in the United States -- can be seen as an ambitious attempt to realize Licklider's original vision for the Internet. So far, the "killer" applications for the Internet have been first e-mail and then the Web. It is possible to see an echo of the computer science community's response to Tim Berners-Lee and the World Wide Web -- his original paper was rejected as having little research value -- in some of its apparently hostility or indifference to the e-Science and Cyberinfrastructure agenda.

I fully respect the value of "pure" computer science research into the fundamentals of the subject. I also accept that the impact of some of this research may take up to 20 or 30 years to deliver clear benefits in the world at large. Nonetheless, as summarized in the book "Pasteur's Quadrant" by the late Donald Stokes, basic research can also be "application-inspired" -- and this is certainly the premise of the U.K. e-Science Initiative and of the U.K. Government's 10 Year Research and Innovation Investment Plan. In my view, as well as a strong "pure" computer science research program, the United Kingdom also needs a strong engineering systems approach to computer science research that is grounded in real application requirements. This all seems to me so self-evident that I am sometimes dismayed by the lack of engagement by some of the CS community in the great endeavor. However, it seems worthwhile, and indeed is probably overdue, to attempt to correct any possible misapprehensions about e-Science that are still out there in the international CS community.

- Tony Hey
Director of e-Science
Engineering and Physical Sciences Research Council