You're looking for empirical data and you probably will never get any unless you do the testing yourself.

The problem is that the alignment and tolerances of the frame bores and to a lesser extent the spindle have an enormous effect on the bearing lifespan of all press fit BB systems. To make empirical statements about any of it, you need accurate measurements of all those parameters. That requires sophisticated tooling far beyond what most mechanics have, and certainly riders as well. The limit to what most can readily do is measure the bore ID with a caliper at various points. That is very different from measuring runout from a theoretical center. And, measuring the angular alignment and axial alignment all requires serious inspection type tooling, to say nothing of the time involved.

It seems likely at this point that the real story of press fit BB bearing issues is that, depending on your perspective, either it's usually an out of tolerance frame that causes problems or the standards were not developed to match well with what factories are able or willing to turn out. With all the standards (BB30 is a good example) you can find examples of bikes where they run pretty bulletproof despite problems being so common and severe elsewhere. It's likely that way simply because only a few manufacturers get their tolerances right consistently.

You're looking for empirical data and you probably will never get any unless you do the testing yourself.

The problem is that the alignment and tolerances of the frame bores and to a lesser extent the spindle have an enormous effect on the bearing lifespan of all press fit BB systems. To make empirical statements about any of it, you need accurate measurements of all those parameters. That requires sophisticated tooling far beyond what most mechanics have, and certainly riders as well. The limit to what most can readily do is measure the bore ID with a caliper at various points. That is very different from measuring runout from a theoretical center. And, measuring the angular alignment and axial alignment all requires serious inspection type tooling, to say nothing of the time involved.

It seems likely at this point that the real story of press fit BB bearing issues is that, depending on your perspective, either it's usually an out of tolerance frame that causes problems or the standards were not developed to match well with what factories are able or willing to turn out. With all the standards (BB30 is a good example) you can find examples of bikes where they run pretty bulletproof despite problems being so common and severe elsewhere. It's likely that way simply because only a few manufacturers get their tolerances right consistently.

To put it another way: systems like DUB in a BB86/92 shell require really small individual bearing balls because of the spatial constraints. That's not necessarily a dealbreaker for longevity, but the smaller the balls get, the less durable they are individually and the more reliant the system is on all the balls being loaded evenly and correctly, which can go bad fast if the shell bores are poorly aligned to each other or if they're too tight or too lose. In contrast, if the individual balls were bigger, they can soak up some of these sorts of issues, to a degree - see for example Mid shell BBs on BMX bikes, which are frequently seen on frames with production values as poor as anything and installed by angry children with hammers, and yet generally have good bearing lifespan. The balls inside are much bigger.