As I See It, September 2008

By Nelson Morgan, Director

The past is never dead. It's not even past.

–– William Faulkner

This issue is a singular one, as it marks the 20th anniversary of ICSI's official inauguration. ICSI was incorporated in June 1986, but by September 1988 staffing had reached critical mass and we formally launched our enterprise with an event. This October we will celebrate this anniversary with another event, which I'll describe below; but first, a bit of reminiscing.

ICSI was the brainchild of Norbert Szyperski, Managing Director of the German National Center for Computer Science (GMD) in Germany, in collaboration with Ron Kay, formerly from IBM in the U.S. Prof. (Dr.) Szyperski felt that he had benefited greatly from the year he spent in the U.S. on an Eisenhower Fellowship (an early visit), and also saw that German computer science was poorly represented in the American scene. The two colleagues saw an opportunity to start something of great significance for the advancement of computer science and international collaboration. And so, after an exploratory phase leading to the decision to affiliate with UC Berkeley, they founded ICSI (with critical assistance from a number of American and German colleagues).

Twenty years ago, ICSI was significantly smaller; we started on a single floor at 1947 Center Street, but within a few years we expanded to a second floor. At first we relied exclusively on our international program, but soon our funds were supplemented by domestic peer–reviewed grants. We had roughly 40 people then, including administrative staff, and now we have about 130. Half of the Institute was organized around a single coherent theme (massive parallelism), while now the focus has morphed over the years based on the interests of senior ICSI researchers.

Despite all of these changes, ICSI has not lost its idiosyncratic flair. While our research and the corresponding sources of support have diversified greatly, we have just as much of an international environment — we have at least as many visitors from at least as many countries as we did 20 years ago. A few of the same people are still here (Jerry Feldman, Dick Karp, and myself), but after 20 years there are many new faces. ICSI staffers currently include a number of people who were here as graduate students in earlier years (Sally Floyd, Vern Paxson, Srini Narayanan, Krste Asanovic, Andreas Stolcke). ICSI's mission, furthering computer science research through international collaboration (and furthering international collaboration through computer science research) has remained the same. And while we have a diverse range of topics, the basic program of theory and application of algorithms to machine perception, cognition, and communication (networking) has remained more or less the same.

Perhaps most importantly, ICSI really has much of the same "feel" as it did 20 years ago. It is part of campus, and yet is separate from it. It is (usually) an island of calm. We're still having "teas" twice a week, and we still have a highly effective administrative staff. Space is still comparatively ample (although look out for those plans to put in lofts to squeeze in more people!). We still have some of the most accomplished researchers in each of their respective fields, great students and postdocs, a cosmopolitan environment, and we still have strong bonds to a number of far–flung labs and researchers. And over these years, there have been many notable achievements, including: the world's first single chip vector microprocessor; seminal work on object–oriented software that had a significant impact on Java; speech processing algorithms incorporated in millions of cell phones; and one of the first and most complete solutions to provide performance guarantees for real–time Internet traffic; Tornado codes, which allow any number of heterogeneous clients to acquire bulk data with optimal efficiency at times of their choosing, even in the face of high channel loss rates; and the eXtensible Open Router Platform (XORP), an innovative and robust open–source routing platform that (we hope) will break open the closed nature of networking hardware. And individual ICSI researchers have won extremely prestigious awards throughout this time, as described in the lead article in this issue.

It is personally extremely gratifying that, not only have we survived 20 years of significant peaks and valleys in each of our support mechanisms, but that we have done so without significantly sacrificing our research lifestyle. ICSI research is still overwhelmingly investigator–driven, despite the many pressures to become more of a contract research house.

This year, we are well along our path to further expansion of our international visitor program, one of the keys to our continued success. In April, we worked with our Brazilian colleagues to begin to set up a new partnership, one that we hope will be formalized by the end of the year. And a new extension of our German visitor program was announced as well.

We will celebrate this 20th birthday on October 17. The day will include talks from ICSI's founders, feedback from representatives of new and continuing sponsor countries, and keynote lectures by a few of the accomplished ICSI researchers, past and present (and sometimes both!). The first of these will be Professor Dick Karp. Dick was here at the start, and except for a few years when he apparently was lost in Seattle, has been here since. A second talk will be from Professor Vern Paxson, who leads ICSI's activity on internet security, and who was also part of Domenico Ferrari's networking research group during ICSI's early days. Ben Gomes, an early Google employee, who received his PhD working with Jerry Feldman here at ICSI, will give a third. The final talk will be given by Krste Asanovic, who received his PhD working at ICSI, became an MIT professor, and ultimately returned to a dual role as an ICSI staff member and a UCB professor.

And what about the next 20 years? I'm putting my bet on ICSI becoming an even better place to be and work. Any takers?