Peter Strazdins presented a seminar on "Research-Based Education in Computer Science" at the Australian National Unviersity in Canberra, 4 June 2007. He went through some definitions of what "research-led" education and "research-based" education are . This was about using educational research results to produce better courses, using research in the subject area (computer science) in courses and in having students gain experience in undertaking research.
Marketing also got a mention: there can be conflicts between what makes a course sound good and what the students actually want to do. Mentioning research in marketing a university increases its reputation and attracts students. But most of those students do not want to be researcher. Even for those students interested in research, the amount they can do in an undergraduate course is limited.
It occurred to me that many of these contradictions could be answered by broadening the definition of research. It makes little sense to teach undergraduates to conduct the sort of research a PHD postgraduate would do, and almost none of them want to or will go on to do this. But any ICT job is going to need research skills. This could be evaluating products, optimizing a system or working out what the customers want. These are all valid areas for research and valuable skills for the workplace.
The notes from the first of Peter's seminars: "A Survey of 'Best Practice' in Computer Science Teaching" is also available online.