I think the evidence would be the manned trials that Buhlmann used to derive his M values. Those were empirical studies that derived an M value for various tissue groupings with different half times, that would result in an "acceptable" likelihood of causing DCS (if memory serves it was something like 2.5%). Of course, what isn't clear is whether the DCS outcomes were because the test subjects were riding the 100% line all the way up (across the fast, medium, and slow tissues), or if there was some specific set of tissues (e.g., the slow tissues) predominantly responsible for the DCS outcomes. We know lipid tissues store a lot more inert gas than aqueous tissues, and that the aqueous tissues are the faster tissues. We also suspect that DCS is partly the result of showers of bubbles coming through the body and shunting the heart/lung exchange in some way, rather than forming in situ in the brain/nervous system. So it could be that the DCS incidence encountered in his trials were due to the medium and slower tissues producing many more bubbles for longer. If that's the case, then getting shallower sooner, to off-gas the fast tissues closer to their M value while minimizing gas uptake in the slow tissues, and then progressively stepping back to give the slow tissues more time to off-gas their larger, slower load might be a reasonable approach.
This is speculation on my part though.