7) The Commission is disingenuous when it claims that safety "is not discussed in extensive detail because any concepts falling short in human safety have simply been eliminated from consideration." Similarly, the Commission was "unconvinced that enough is known about any of the potential high-reliability launcher-plus-capsule systems to distinguish their levels of safety in a meaningful way." For the Commission to dismiss out of hand the extensive analytical work that has been done to assure that Constellation systems represent the safest reasonable approach in comparison to all other presently known systems is simply unacceptable. ...
From "A Commercially Based Lunar Architecture" - Zegler, Kutter, and Barr (United Launch Alliance):
Despite the best engineering design and analysis activities it is amply clear that even highly vetted designs such as the Space Shuttle can fail catastrophically. Probabilistic analyses are spectacularly flawed in that they make sweeping assumptions about failure modes and the means to prevent them. Nature relentlessly renders these complex analyses moot when we find another hidden failure mode via flight experience. Ground testing can assure a baseline level of confidence but only extensive flight experience can truly generate a safe vehicle with high confidence in its overall reliability. Aircraft flight testing relies implicitly on this principle.
Extensive flight experience will be hard to come by with Ares 1, which only has the job of launching astronauts for NASA. Commercial rockets like Delta IV, Atlas V, and Falcon 9 have many other jobs, so they can demonstrate their safety and reliability over many launches - or fail to demonstrate that safety and reliability without astronauts on board.
Some critics point out that even if Ares 1 itself is safer than some other rocket, it may make the overall mission more dangerous (for example by causing redundancy to be removed from Orion).
Considering that safety isn't just about launch, the Augustine committee has an interesting idea with the "Deep Space"/"Flexible Path" options. These options postpone Moon and Mars surface exploration, allowing us to postpone the safety issues of landing astronauts on the surface, surface operations, and surface departure. This allows us to grapple with a perhaps more manageable safety problem, that of deep-space operations, for the time being. When we have gotten good at that, we may find the isolated problems related to surface missions to be more manageable, too. Incremental development in hardware and operations may lead to safer missions.