A recent article in the CACM complained about the dominance of reductionists’ views in computer science research.

“We are sorry to inform you that your paper has been rejected, due to the lack of empirical evidence supporting it.” [1]

The author goes on to discuss other views of science, suggesting that the focus on empirical validation of research is too narrow a view of computer science. I concur, but would like to add that today’s empirical focus actually represents an improvement over the prior even more simplistic view of science, one in which handling of human subjects was cursory at best (at least in software engineering research).

Here is what the social sciences think of computer science. The photo was taken at WikiSym 2010, during Cliff Lampe’s keynote. He is an information systems researcher, who straddles social science and computer science.

“Sloppy methods” — that’s what makes up computer science in the eyes of this scientist. I don’t think that we are actually that sloppy, but that our notion of what constitutes computer science is too restrictive. Here is a simple picture of how I explain science to my students:

In the current prevailing view, science is only about the lower part of the figure, the theory validation part. It doesn’t matter how we came up with a theory as long as we can validate it. This is rather short-sighted, given that science is also a process of communication and building on each other’s work. The theory formation part is equally important and allows for the same rigor as the theory validation part.

This broader view of science has long been held by other sciences that deal with human subjects, most notably the social sciences. Explicit methods exist, for example, Grounded Theory, which support the process of going from data to theory. Grounded Theory, done right, is every bit as rigorous as any method aimed at validating a theory. I have seen a few Grounded Theory based submissions to program committees, and I’d expect them to grow. We should be supportive of rigorous research that goes from data to theory and not just theory to data. I’m certain that this broader view of computer science will eventually prevail.

I also understand that one might have an even broader vision of the science part in computer science. My little picture above is also simplistic in many views. However, for now, I’d be happy to see the scope of software engineering research broaden at least that far.

References

[1] Gonzalo Genova. “Is Computer Science Truly Scientific?” Communications of the ACM, vol. 53, no.7 (July 2010). Page 37pp.