Still, with the proper safety factors added, it is the best data we've got...
I am trying to recall accidents where the POH data said it would be ok but it wasn't; seems to be pretty rare isn't it? In the majority of overruns and collisions with obstacles, even the POH showed that it wouldn't work.
So it would seem the bigger problem isn't that the POH is off but that people don't read it...