Successful Software Engineering Research
- Keep aware of what is actually happening by reading industrial programs.
- Try to apply your ideas to programs that were written for some other purpose, not to programs that you made up to illustrate your ideas.
- Don't attack the symptoms, but keep looking for the causes. The developers can, and will, attack the symptoms as least as well as we can.
- Keep asking why people aren't using your ideas and don't take "stupidity" or "ignorance" as an answer. You cannot eliminate stupidity and you can do little to correct ignorance, but if there is a weakness in existing research results, you have found a solid research problem.
- Be wary of fads. [...] Research topics are particularly likely to be fads in a field where each new paper is a response to a previous paper rather than to a fundamental problem. Always look for the fundamental problem and don't jump on bandwagons. Papers about yesterday's fads are forgotten.
- Be wary of vaguely defined buzzwords. A "buzzword" is a word that everyone knows but few people can define. "Buzzword" is a buzzword. Much of today's literature is a debate about meaning of words that is a disguised as a debate about how to design software. For example, most of the debates that I see about the strengths and weakness of various OO approaches boil down to differences of opinion about what OO means. Pointing out buzzwords problems is another service that researchers can provide. The secret to successful research is picking the right problem. [...]
David Lorge Parnas. In "Successful Software Engineering Research". Software Engineering Notes, Vol. 23, No. 3, May 1998, pp. 64-68.
Link to this Page
- Miscellany last edited on 13 June 2008 at 2:51:37 pm by gate2.ensm-douai.fr