Lots of good ideas there, Thomas. It looks like we all have some experience with bottom end failures, but don't all draw the same conclusions from them. The first set of cases that I had break in a race bike were the early '71 cases that Farrell likes, and the crank had the original roller drive side and ball timing side races. The case cracked on the drive side at the back just above the top cradle bolt, where the early cases were very thin. I had the cracks welded, along with some added reinforcing ribs, and used that set of cases with no more problems for several years of road racing. The original crankshaft eventually cracked around the drive side mainshaft, after a few seasons of racing. Over the next couple decades I had several other cases break on the drive side half, on a line from the mainshaft bearing bore back to the case joint behind the cylinder, just above the top cradle bolt. The cases that broke included another early 750 case that I had reinforced with ribs behind the cylinder and above and below the upper cradle mount, a stock 850 case in a short stroke 750 engine, a heavily reinforced MKIII case in a 920 engine, and a heavily reinforced 850 case in another 920 engine. I also managed to break several race-prepped crankshafts, and there may have been a couple more broken cases. The breakage all happened between 1972 and 1993, and I'm only describing the ones I recall in detail. The conclusions I drew from my experience were that the engines with reinforced cases and one-piece crankshafts lasted longer than the others, and I never saw any connection between whether the timing side bearing was ball or a roller. The 920 engine I'm running now has heavily reinforced 850 cases and a one-piece Nourish crankshaft, and really seems to be bulletproof (so far). I also think that the combination of original Commando cases and crankshafts, of any model, in a road race bike that makes high horsepower and is ridden aggressively, will eventually result in broken cases or crankshaft, or both. It's just a matter of time. With modern aftermarket replacement cases, crankshafts, rods, etc., I think it is possible to build much more reliable Commando race engines. It's still possible to break them, but it takes a lot more abuse to do so.
Just my humble opinion, and it could be wrong. I'm not trying to convince anyone, or prove anyone right or wrong, just passing on some experience.
About the only thing I'm really sure of is that, all other things being equal, the more horsepower the engine makes and the harder the pilot rides, the more often things will break.
Ken