NASA is going ahead with the launch of Discovery. When I studied engineering ethics as an undergrad, the Challenger disaster was one of the case studies. Feynman, who helped investigate that disaster, wrote about a communication breakdown between the engineers and the administrators. Were the administrators unaware of the potential problems with the launch? Not quite. The problem Feynman writes about is sometimes described as a "normalization of deviance" from original design.
The principle boils down to the following. If breaking a rule incurs no consequences one time, then there is less concern about breaking that rule the next time. For example, last week I was ten minutes late to get to my car, which was in a two hour parking spot. Luckily, I did not receive a ticket. While I was nervous about getting a ticket last week, if I ended up running late on a future occassion, I don't think I would be as concerned.
In the case of the Challenger, similar warnings in previous flights went unheeded but did not result in catasrophe. Thus, administrators ignored those warnings again, but circumstances did not work in their favor that final time. While I am not as familiar with the internal workings at NASA during the Columbia disaster, it is not hard to imagine that similar forces were at work.
Is "normalization of deviance" inherently bad? Some would argue that new endeavors always carry a risk, so it might be necessary to push the envelope on occassion. As Admiral Adama says, "Sometimes you gotta roll the hard six." In Adama's case, however, lives are at risk regardless.
The issue is whether this launch is worth the risk. The shuttle fleet has been pushed to its limits, but at the moment, it's the only fleet NASA has. The significance of the mission would have to be the test. Let's hope a successful mission does not restart the NASA tradition of marginalizing its engineers.