Please note that these arguments were made on the "for" side of a debate about whether CS education could be considered a "proper" part of CS. (John Impagliazzo and Olle Willen spoke against - or at least for the "not yet" position).
The case I made addresses three common misconceptions:
As academics, we have two professional duties. One it to teach, the other is to pursue research. These endeavours are different.
Research is a public activity which we owe to our wider disciplinary community. Researchers in, say, garbage collection, expect to talk with (and work with) like-minded colleagues from other institutions - national and international. They publish work at conferences and in publications which address their speciality. In this way they gain entrance to, and demonstrate credibility within, their own community.
Teaching is a (largely) private activity which we owe to the institutions in which we work. We are expected to talk about it with all our local colleagues, but to no one else. (After all, what would be the point? The context of their teaching would be quite different).
Part of the problem with CSEd Research (I believe) arises from the confusion between research activity and teaching activity. Of course, my research *informs* my teaching - in exactly the same way as it does all my colleagues. But I do not see that my research makes me any better at the classroom "tricks of the trade" than anyone else's area of expertise.
"Computer Science" is a new discipline, created out of many others. The debt to mathematics is clear, and often seen as the "core". Formal Methods, Theoretical CS, Algorithms etc., all use mathematical methodology. Hardware interfacing is clearly akin to electronic engineering. In other areas, software engineering methodologies have come into academia from industrial practice; Operational Research and Business Processes from business schools.
Perhaps the most interesting area here is HCI, which, today, is pretty much accepted within CS as part of the discipline. It, too, uses methodologies from other subjects - notably from psychology. Michael Feldman posted to this list last year, developing this case:
>I recall Ben [Schneiderman] saying that when he approached traditional >behavioral scientists for help with empirical studies, they'd wave him >off with "read my papers; what you want to do is all in the >literature", and when he approached "mainstream" CS people, they'd >wave him off with "why are you messing with this soft stuff; go back >in your corner and write another compiler." We seem to be going >through another iteration of the same Catch-22, in a discipline that >is, in fact, quite closely related to HCI. Now that HCI is (I >guess...) now considered legit, how do we do the same with CSEd >research?
CS Education uses methodologies from other disciplines - Education for the learning theory, Psychology (and other social sciences) for experimental and analysis techniques. In this way its development has been similar to many of the other research areas within CS which have developed from other subjects.
The key must surely be in the questions that we ask. If we ask questions that are generalist, ("Do students learn better from face-to-face or Web-based interaction") or facile ("Do students learn better if A or B is their first programming language") then perhaps there is no such thing as CS Education. But if we ask questions that can only be addressed from within CS (my current favourite is "Does a knowledge of computer architecture make better - more expert - programmers?") then these questions *cannot* be addressed by someone outside of the discipline. What meaning, after all, would such a question have to someone who could not program? Who did not understand what the quality of the relationship was?
CSEd Research is an emergent area. However, it is no longer the ghetto of a few individuals. Pointers to this are:
>many of the specialty-course folks finding room within the >specialty conferences and organizations to discuss teaching of their >courses. For interest, the American Assn of Artificial Intelligence >has sponsored a number of workshops and special tracks on teaching AI >in the last few years.
And with regard to emergence, finally, everything was once emergent. CS itself was once emergent. That's not really an argument to say it's no good.
Sally Fincher, 1999
http://www.cs.ukc.ac.uk/people/staff/saf/3csergi.html