People of ACM - Koen De Bosschere
October 23, 2014
Koen De Bosschere is professor of Computing Systems at Ghent University in Belgium, and leader of the Computer Systems Lab. He is the editor-in-chief of ACM Transactions on Architecture and Code Optimization (TACO), and coordinator of HiPEAC, the premier network of computing systems researchers in Europe. He is co-editor of the bi-annual HiPEAC vision document and organizer of the yearly ACACES summer school. At Ghent University, he manages the bachelor and master program in computer engineering, and the university-wide student-entrepreneurship project for which he received the Hermes Award in 2012. He co-authored 100 publications over the last 10 years. His research interests include computer architecture, system software, code optimization, and software security. He earned B.S. and Ph.D. degrees in Electronics Engineering and Computer Science at Ghent University.
As the coordinator of the European HiPEAC (High Performance and Embedded Architecture and Compilation) Network, how would you evaluate its impact over the last decade?
HiPEAC started in 2004. At that time, we were convinced that a research network in computing would help to strengthen the computing systems community in Europe. After 10 years, we are a network of over 1,500 researchers, active in all major European countries. We organize four well-attended networking events per year, publish a bi-annual research vision document, and we represent the European computing systems community at the level of the European Commission.
Through HiPEAC, the community has gained international visibility, and its impact has grown. The global European research funding has increased fivefold over the last decade. The fact that we are successful does not mean that our job is done, however. In the future, we want to focus on better monetization of the European research results, and on job creation. Given the disruptions that are expected in the coming decade (the potential end of Moore's, Nielsen's and Kryder's laws), there are many upcoming challenges. HiPEAC wants to help the community to tackle these and to help Europe to become more competitive in the computing systems domain.
What are the advantages of the unique journal-first publishing model used by HiPEAC conferences, which rely on papers previously published in scholarly journals for main track presentations?
In 2010, HiPEAC decided to completely change the concept of its yearly conference. Instead of a five-day publication event, we wanted it to become a three-day networking event, with a rich program of keynotes, invited talks, workshops, tutorials and poster sessions running in parallel with the main paper track.
Since we no longer wanted the conference to be a publication event, we decided to outsource the paper selection process to ACM Transactions on Architecture and Code Optimization (TACO). In order to encourage our authors to publish in ACM TACO, we keep publishing a yearly call for papers for the conference, but the submission site is the submission site of ACM TACO. We helped the journal to streamline its review process to reach decisions in at most two months. Papers submitted by the submission deadline in June get a final decision by November (after two review rounds), and are published in January, just in time for the conference.
Both the conference and the journal tremendously benefitted from this decision. The conference now currently attracts 500+ delegates instead of the 200 we attracted before—it has become a real community networking event attracting all European stakeholders in computing.
ACM TACO now attracts about 190 submissions per year, instead of 40 in 2010. We observe that most authors no longer wait until the call for paper deadline to submit their work, but submit throughout the year, which spreads the review load for the journal. Hence, this decision simultaneously made the conference and the journal more attractive, which is for me a clear proof that conference publication is not the only possible publication model in computer science.
Authors are willing to submit their original work to a journal on condition that the journal review process is of high quality and fast, and that the published work gets enough exposure in the community. This is how most other research communities work—they publish in journals, and they network in conferences. It is not difficult to change; we can simply copy the best practices of other disciplines. It only needs the courage to do it.
What are your major concerns as Editor-in-Chief of ACM Transactions on Architecture and Code Optimization?
Operationally, my biggest concern is respecting the deadlines. When a paper is submitted, we promise the authors a decision after at most two months. That means that I spend quite some time checking the deadlines, and intervening if associate editors or reviewers are not respecting their deadlines. I really want decisions in at most two months. Recently, we got a few decisions in five weeks, which is great for the authors. I am continuously experimenting with new incentives to convince editors and reviewers to respect their deadlines and to deliver high-quality reviews and decisions.
Strategically, my biggest concern is the low impact factor of many of the computing journals, compared to journals in other disciplines. I try to understand the root cause of it. One obvious reason is that most scholars prefer to publish in conferences instead of in journals, but that is not the only reason. I recently discovered that we also cite fewer recent papers compared to disciplines with journals with high impact factors. Furthermore, improving the impact factor is clearly a chicken-and-egg problem. With a high impact factor, the journal will become more attractive for high impact papers, but we need such papers in order to increase the impact factor of the journal. As with all chicken-and-egg problems, it will take time to fix them.
As a leader and visionary in the European technology community, what advice would you give to young people considering careers in computing?
I advise young people to read about the future—and by this I mean about the predictions for the next 50 years—that is the half century that maps on their professional life. Some people like Moshe Vardi (Editor-in-Chief of Communications of the ACM) state that they will experience "the end of work" in 2045, or the technological singularity also predicted in 2045 by Ray Kurzweil. These events will not happen overnight, we are already evolving towards them today. The rise of cognitive computing is in my opinion a clear example. Thinking machines will replace many middle-class jobs. I do not know the future-proof jobs, but I tell my students to invest in a wide range of future-proof skills: creativity, social skills, entrepreneurship, and of course computing.