I can't speak for neko's department, but can explain the reasoning here at CMU. In the Computer Science department (and now in the much larger "School of Computer Science" which includes many departments, institutes, etc.), we have long had a bias against hiring our own Ph.D.s -- unless they've been out in the world for a while. The issue is not nepotism, but rather "group-think." For a research university, it would be a very bad thing to wind up with institutional blinders on; group-think increases the possibility that we will fail to investigate worthy ideas that don't fit the existing preconceptions of the current set of academics.
Some benefits of external hiring include:
- A bias against hiring our own students leads to more hiring of people from other schools, which also leads to more mixing of ideas.
- Technology transfer works best when the technology is carried by people who are familiar with the new ideas. Sending our students out away from our university helps spread their (and our) ideas more widely and increases our impact on the real world.
Xenophon
P.S. I also note that only one of the top-3 Computer Science graduate programs in the world has a bias
towards hiring its own graduates (that would be MIT). The other 2 -- Carnegie Mellon and Stanford -- share the bias
against hiring our own grads. As does U.C. Berkeley, which comes next in the rankings.
P.P.S Graduate school rankings in computer science are a bit odd. CMU, MIT, and Stanford have been in a 3-way tie for "best in the world" for about the last 30 years or more. UCBerkeley sometimes is rated #1 also (making is a 4-way tie), at other times they are ranked by themselves in the #2 spot just after the others. There's been no significant change in those top spots since the mid-late 1960s.