European Computer Science Summit 2005

European computer science takes its fate in its own hands

By Bertrand Meyer, ETH Zurich and Willy Zwaenepoel, EPF Lausanne
Europe’s contribution to computer science, going back seventy years with Turing and Zuse, is extensive and prestigious. While everyone acknowledges that this role is second only to that of the US, the European computer science community is far from having achieved the same strength and unity as its American counterpart. On 20 and 21 October 2005, at ETH Zurich, the “European Computer Science Summit” brought together, for the first time, heads of computer science departments throughout Europe and its periphery. The meeting was a joint undertaking of the CS departments of the two branches of the Swiss Federal Institute of Technology: EPFL (Lausanne) and ETH (Zurich).

Initially conceived as a small workshop, a kind of get-together, the initiative quickly attracted interest far beyond its original scope. In the end close to 100 people attended, representing most countries of the European Union, plus Switzerland, Turkey, Ukraine, Russia, Israel, a delegate from South Africa; Russ Shackelford, from the US and Gordon Davies from the UK attended on behalf of the ACM. Eastern Europe was well represented. The program consisted of two keynotes and a number of panels and workshops on such themes as research policy, curriculum harmonization, attracting students, teaching CS to non-CS students, existing national initiatives, and plans for a Europe-wide organization.

The first keynote showed how much the European CS community has to learn from the US example. The avowed model for our meeting was the US “Snowbird” conferences, which for decades have been a forum for North American CS department chairs, and resulted in the creation of the Computing Research Association; the CRA has had a profound effect on shaping the North American CS community and influencing public policy. To talk about this experience we invited Ed Lazowska from the University of Washington, chairman of the CRA for many years. Ed’s talk was a goldmine of information on the CRA story and on the computer science community in North America, filled with facts, trends, curves and relevant statistics. This was an opportunity to realize how much we have to accomplish, starting with the first steps of gathering the basic data.

The second keynote addressed a subject that is getting more pressing for a number of universities: evaluation of publications and more generally of research. Increasingly, professors and researchers are asked to have their publications, their citations or both counted. While questions remain about the wisdom of such counting exercises, computer scientists in Europe have failed to make the specificity of computer science with respect to publications — the importance of conferences for example — generally understood and accepted; the risk exists of being evaluated according to criteria poorly adapted to most of our discipline, such as the number of publications in Science or Nature. A first step towards assessing publication activity is to know exactly what to count and the methodological limit of what we can count. Michael Ley, who maintains the DBLP server in Trier (Germany), has a unique perspective on the topic. The DBLP lists computer science publications, not citations, and results from an exacting effort to get the data right, through a combination of automatic and manual work. In his keynote, Michael described for example the difficulties raised by authors whose names appear in different ways (Michael Smith, Mike Smith, M Smith), by several authors sharing the same name (he applies coauthor analysis algorithms to help sort them out), or Asian names. His work is a model of rigor care, and openness, which one may only wish were followed by all those in charge of counting who publishes what and who cites whom.

Among the other topics discussed throughout the Summit we may note: the role of women in computer science, and how to attract and retain more of them; European research policy; how to attract students, CS for non-CS students, curriculum initiatives, and presentation of national efforts.

Besides the keynotes, talks, panels and workshops the most important result of the Summit was the unanimous view that European computer scientists urgently need an organization with aims and scope similar to those of the CRA, extended — in light of the peculiar situation in Europe — to cover education as well as research. An initiative to start this organization is on its way, and should culminate in an official start at the next ECSS, tentatively scheduled for early September 2006 in Lyon. Some of the immediate tasks may be:

• Starting the ground work: list of institutions, mailing list, Web site.
• Defining criteria for publication and research evaluation.
• Defining guidelines for CS curricula.
• Proposing strategic directions for CS research in Europe.
• Attracting students to the discipline.

The aim of the association will be to become the recognized voice of the European computer science community — not limited to universities, but including for example research centers and industrial research labs.

See http://se.inf.ethz.ch/events/cs_summit_2005 for more information.