Regarding the Uberlingen incident, Bert Ruitenberg (human factors expert) recently presented a report on that incident at a conference. An interesting comment on TCAS (not verbatim):
There is a known software "glitch" in TCAS II Version 7 (the current iteration of the software) where a "mode reversal" (climb, then descend) can ultimately confuse TCAS to the extent where it no longer helps to resolve the conflict, but only makes it worse. This software glitch has been known to the developers of TCAS for many years, but they (probably the bean conters) decided some time ago that additional development of TCAS to resolve the problem was not warranted given the remote chances of its occurrence.
Remember, SOPs are written by human beings. NO HUMAN BEING IS INFALLIABLE!!!
So is TCAS.
I don't know whether this is related at all to the Darwin incident, but it was the last line of defence at Uberlingen ... and it failed.