M-values are chosen to produce (statistical probability of) X cases of clinical DCS per Y dives. The way probabilities work is a) 7.5K dives may well have been "his time" and b) the sample of dives and divers from which they are derived must be representative of Woody and his dive, for the numbers to work. Gradient factors (as defined by Baker, not to be confused with those of Shearwater) throw an additional wrench in the works: while the M-values chosen by Herr Dr. Buhlmann represent some X-in-Y, the effect of reducing them to .7 on GF High is "we think it's safer", and "undefined" for any value of GF Low.
Which makes all the speculation of what/why/how kinda moot: he got bent because no algorithm says you won't, they just say the likelyhood of that is "reasonably low" for somebody's definition of "reasonable".
Which makes all the speculation of what/why/how kinda moot: he got bent because no algorithm says you won't, they just say the likelyhood of that is "reasonably low" for somebody's definition of "reasonable".