[Henry Schmidt]

Well Bob, I don't know what to tell you.
Maybe the cranks you're having ground are being ground poorly. Perhaps they skip the nitriding process. Perhaps the cranks you're seeing fail, are race cranks.
One error many amateur racers make is to use parts too long. Engine parts have a life. The excesses of racing greatly shorten that life.
Most of the 911 cranks I've seen fail/crack over the last 40 years were 2.4/2.7 cranks that people tried to spin to 8K.
The flyweights are so thin that they flex causing a terminal harmonic.
Grinding a fillet radius is of course critical so when we make a "special" crank (like, 2.0, 3.2 or 964 to GT3 rod journal) we always shoot for the largest radius possible. T
his often requires us to chamfer the bearings to create the optimal clearance.
To correct the flex (improve torsional rigidity) in 2.8 and 3.0 RSR cranks they increased the radii to almost double the stock 2.7 crank, among other modifications.
The RSR rod bearings came with a noticeable chamfer.