Professor Hazel L. Sive has it right:<p>"<i>Part of excellence, she says, “is that you are either the top investigator in your field, or one of the very tiny handful of top investigators in your field, in the world.</i>"<p>A professor has to be #1 or #2 in his field, <i>sometimes</i>, <i>maybe</i> #3.<p>"<i>“Competent teaching is required for promotion,” she says. Although “extreme excellence [in research] can compensate somewhat for less excellent teaching skills,” since acquiring teaching skills is “always in progress,” good teaching alone is not enough for promotion.</i>"<p>All MIT classes are taught by professors (I've heard of one exception to this, a case of a truly exceptional EECS grad student who everyone felt was the best person in the department to teach a particular class) therefore this is a necessary but not sufficient criteria. It is just about impossible to avoid having anything to do with undergraduates, I know of only one professor in the EECS department who managed that trick (let us say he was very politically powerful, although in my direct experiences with him he was a better than OK guy; we gave him booze, he gave us money :-).<p>These two above "no compromise" things are part of what makes MIT MIT.<p>Minor notes: the 50% failure/success rate is about average for the US from what I've read recently. Eric Hudson's research at the gross/highest level is I'm pretty sure a very popular area, it was back in the early '80s.