The Voracious Discipline

Making your colleagues the subjects of your research might be a poor career decision. Research creates a clear division between the observers and the observed—in a moment, researchers go from being equal colleagues to detached scholars, retreating into their own worlds and losing their collegial connections to the eld. Yet, systematically studying computer science might help us understand how such research operates and how computer science both borrows from and contributes to other elds. Computing is a voracious discipline that has plundered much of humankind’s intellectual endeavors. In computing’s early days, the basic computation model borrowed concepts from biology, and the programming process took terms from industrial engineering. By the end of the 1950s, computing was already plucking ideas from mathematics, statistics, economics, linguistics, management, and electrical engineering. During computing research’s formative period, from roughly 2008 to 2012, I watched researchers borrowing ideas from others without necessarily understanding what they were taking. One individual used an experimental design from psychology to test a decidedly nonpsychological theory. A second adopted an industrial management model without recognizing the elds’ di erences. A third used a market model without addressing some of the basic issues with supply and demand curves. In the end, all these researchers contributed to the computing eld, but none of them received the full bene ts of collaborating with researchers in di erent elds. If we can’t study how we appropriate ideas, we can at least study the literature to get some idea of how we work. If computing is a eld that merely borrows from other elds, we might expect the motion of our literature to be inward. Eventually, researchers acquire what they need from other elds and return to the computing literature. They get most of their ideas from computing journals and conferences, publishing their results in those same places. Obviously, a full study of such a phenomenon would require a great deal of data, a contentious debate over the IEEE/ACM computing typology, and a complex stochastic model. However, we can get a quick picture of how computer scientists conduct interdisciplinary research by analyzing the bibliographies of survey articles on common subjects. These articles aren’t as plentiful as I’d like, but I stumble across them regularly enough that I can do a few simple comparisons. For example, I recently found a pair of survey articles on image processing, the rst written in the early days of the eld and the second written three years later. Some quick computation revealed the pattern I’ve come to expect. The rst survey cited many articles from journals outside the computing eld: optics, biology, television engineering, mathematics, communications, and cartography. The second survey cited many of the same elds but had a much smaller fraction of references in elds outside computing. In the rst survey, 16 percent of the bibliography came from IEEE or ACM publications. In the second, that gure was 22 percent. In addition, another quarter of the items in the more recent bibliography came from books that were published for the computer science community. Two survey articles, of course, don’t fully establish any theory—much less a theory about computing research as a whole. However, they do seem to support a strategy that we innately feel.