I don't want to read an email from the Chancellor about parking for the football game. I don't want to fill out a form telling the university my ethnicity and other demographic details. I just want to do teaching and research, research and teaching, and then some more research after that. I'll do essential, meaningful service as well, helping to put together tenure documents, mentoring junior faculty, doing peer reviews of articles. I'll suffer through meetings. Most of what is "work" aside from that, however, is just a drain on energy.
Since teaching is highly structured in time and research is highly unstructured, service cuts into research much more than teaching. Service is also like research in that it takes place in less structured time frames, and can be infinitely expandable. There are people with the same teaching load who do four times more service, just as there are scholars who publish four times more than their colleagues.
What if research was structured: you had to sit in a room and do it for so many hours a day for so many weeks (like Thomas's 16 week plan) and you had to be accountable for that time? What if, at the same time, teaching was totally unstructured: you would meet with random groups of students who wanted to see you on random occasions? In other words, what if we inverted the relation between research and teaching? Obviously a lot more research would get done, but more significantly the relation between the two activities would change. It seems rather odd that our too main activities should be so asymmetric in the way they are organized and rewarded.
It's also interesting that scholarship is communication upwards, where the intended audience consists of specialists who know as much or more than you, and that teaching is communication downwards, where the intended audience knows vastly less than the instructor. It shouldn't be too surprising that the best at communicating upwards wouldn't always be the best at communicating in the other direction, or that those that are best at planning 15 weeks of classes aren't the best at working in a less structured way.
Graduate teaching is a special case, because the instructor has to decide when to go down and when to go up.