The actual numbers depend on a multitude of variables, but above 8 PPM seems to be the generally accepted upper threshold. Without a way to measure dissolved O2 and adjust accordingly, it's kinda moot anyway.
My point is that it's not even possible to over-oxygenate using a splashing method, but you will almost certainly over-oxygenate every time when using a pure oxygen rig. With pure O2 and a stone, it is very easy to sail way past the ideal range (as someone's measured numbers showed earlier). Under-oxygenating is rarely an issue, and never is if you're pitching enough healthy yeast to begin with, especially if you're using dry yeast, and/or supplying an adequate lipid source.
Source please.
This flies directly in the face of info from Wyeast as well as JZ and Chris White's "Yeast" book.