OK, but does that imply that they have to go on all the ferrying flights too? If the base flight failure rate is 2% and humans are limited to 1/N of the flights (N>=2), then the flight mortality failure chance goes down from 2% to 2/N%, which seems acceptable if the cost of automating is "small". It's at least "small" in terms of human cost. I thought lack of automation (insistence on the manned space flight elements of the program) has been the primary criticism of NASA by the astro-scientist set at least since Apollo.
Of course it has been the criticism. Automation is the holy grail of engineers and scientists and it is a damn good thing too BUT, there is also a visceral need for Human involvement. Does the computerized lathe operator feel the same at the end of the day that the machinist of yore felt? My problem with NASA is the earth orbit fixation. I want to see us actually EXPLORING space not sending multi-billion dollar delivery trucks into orbit.
It seems to me that retrofitting for remote control avoids many of the worst aspects of these issues.
On another tack, I wonder if the NASA investigation will reveal in hindsight that Columbia was inherently more dangerous than the other shuttles to fly. That (and bringing back freon foam, and having formal standby Soyuzes for space rescue option) would seem to permit the shuttle program to get back on track sooner (and so avoid the personnel standby costs).
(I am not a NASA engineer, etc.)