Good input, and I appreciate any insight that I can get.
I'm not sure what the spec for gear lash is off the top of my head for the 5mt (I'll look it up) but I know a lot of domestic transmissions and differentials spec gear lash with a +/- .002" tolerance. And it doesn't take much flex to accomodate a few thousandths of an inch worth of movement. When gears seperate past that tolerance it doesn't neccessarily mean immediate failure either, but it can allow stress raisers and surface cracks to form which propogate to the root of the tooth over time and lead to catastrophy down the road.
The bearings do fail... Not regularly, but it's not that uncommon. My front input shaft bearing failed shortly after I bought the car used right around 80k miles.
It's certainly not an everyday thing to see a large crack in the case, but the crack uncovers a weak point. Where there's smoke, there's fire. And in aluminum, where there's a crack of this type, there's flex.
Subaru did reinforce the case in 1998. I know this because my case was made in 1997
. They went from 4 bolts at the bellhousing to 8 bolts, probably for added torsional rigidity. They also added some material along the top and and bottom, no doubt in my mind for increased torsional rigidity. I've been in manufacturing quality control helping out with R&D and warranty-claim analysis for about 4 years now and have been over this scenario more times than I can count in a couple of different companies: The current design fails too often... so you find a reason and the bean counters tell you to add just enough stregnth so that it doesn't require massive retooling or other major production expenses. Engineers being limited by cost, no new concept there. There is a financial curve involved in this kind of revision. You stregnthen the design such that warranty claim costs do not exceed the money you save by not building it stronger/better. Big limiting factor there.
Another big deal to me is the hardware involved... Subaru uses metric class 8.8 bolts to hold this thing together, which is really sub-par. Class 10.9 bolts are far stronger but much more expensive in the 180mm+ lengths needed (about 4-5 times the price of class 8.8 per bolt). So that's an obviously cut corner.
Subaru did stregnthen the case to some extent, but certainly not to the level that you would see in a 6mt or any other modern transmission in a high performance car. That would have killed their bottom line in re-tooling costs and production down time. I'd imagine that this was a move made to hold the STi models together until the 6mt was ready for production. They left some very obvious weak points in place. That's where I'm stepping in.
It takes some work for a car manufacturer to bring such a potent AWD car to market at a mid-$20k price range. But little things like using class 8.8 bolts instead of 10.9 and staying with basically a facelifted 20+ year old transmission design are what keeps the WRX affordable. Which is fine, that's why the aftermarket is alive and well
Remember that when you're building 20,000 or more units of something, an easy $10 per-unit revision means you just lost $200,000 or your retail price goes from $25,995 to $26,005 (example) and 300 fewer fickle Americans buy your car. So nope, they might not have applied what to your or me seems like an "easy fix." That's just a reality in manufacturing.