DO-178B was based on the consensus of the avionic software community as it existed in 1992. Twenty two years after publication, we have no publically available experimental data as to its efficacy. It appears to work extremely well, since there have been no hull loss accidents in passenger service ascribed to software failure. This is a comforting and surprising result. However, if we don't know why DO-178B works so well, there is a danger that we could stop doing something that really matters, which could lead to an accident.