Concentric,
My reference to G-REDL would be the date it crashed (April 2009, you are correct). The original type certificate for the L2 was EASA approved 12 June 1991. Guess I was off slightly, not quite 18-19 years ... More like 17.5 years. That said, the L2 variant operated for 17.5 years with the "design" of gearbox and epicyclic before it failed in that way. Obviously REDL's epicyclic wasn't physically 17.5 years old, but the design was .... That's my point. In that 17.5 years, how many epicyclic modules went to the full TBO, the original TBO. If they never made TBO, they held together to bring everyone home safely. So what has changed ... Not the design.
When OJF crashed April 2016, the "design" of the epicyclic would have been about 20 years old in total and would have been sitting in the 225 for about 12 years.
What I'm saying is, I do not personally find a component design that has operated for 17.5 years a poorly designed component. The eggs in the basket theory may be a bit out there as well. I would say "every" gearbox out there has one rotten egg that if it cracked it would spoil the whole carton.