Re: Scientific Linux
Scientific Linux is a good idea in theory, dreadful in practice.
The idea of a research/academic-software-focused distro is a good one: unfortunately (I say unfortunately, but it's certainly what I would do myself), increasing numbers of researchers are now developing their pet projects on Debian or Ubuntu, and so therefore often only make .deb packages available.
Anyone who has had any involvement in research software knows that if you find yourself in the position of needing to compile someone else's pet project from source, you are often in for an even more bumpy ride than usual.
And the lack of compatible RPM packages just encourages more and more researchers to go where the packages (and the free-ness) are, namely Debian and friends, which continue to gather momentum, while Red Hat continues to stagnate.
Red Hat may be very stable for running servers (as long as you don't need anything reasonably new (not bleeding edge, but at least newer than three years old)), but I have never really seen the attraction in it myself (especially as there isn't much of a "community" feeling around it, as its commercial focus gets in the way).