Since we are all speculating, I will give you mine, I think there have been a change In the quality in the hubs being supplied. We do not know the history of the rebranding and repackaging of these replacement parts, but we have been running Hoosiers and off sets for a few years yet we are seeing a recent rash of failures,
I think repacking is more important now than before, inspection, daily is also more important.
If I could Sum it it in a few words, cheap steel and poor processing specifications
It might be "some of all of the above", Frank. I'll fess up a little here - my magic hubs that I "ordered" in 2006 and have since run out of, were NTNs with the "double wide" induction hardening that I know you have seen. At the time, I sectioned them and hardness tested them against 1992 Koyos. Hardness in the outer race is only part of the equation, the rigidity and toughness of the hub section is just as critical. AT THE TIME (<2010), I believed failures were due to insufficient hardness in the outer race, or insufficient "toughness" in the 0.030" or so beneath the race surface. NOW, I am wondering if rigidity of the hub itself is the issue - specifically the outer race section becoming "bellmouthed" under load. The fact you guys are getting play and still running the bearings tells me that you are possibly plastically deforming the hub. Some are retorquing and overtorquing and getting better luck. That also is a change, and also suggests that the hub is "creeping" permanently. Still speculation, as I haven't seen a any of those parts .... but "play" in all the failures in the past were always due to spalling in the race, and this quickly degenerated into a grinding, super-sloppy, mess of a bearing. (Caveat: We never bothered "analysing" unbranded private label hubs that failed super quick. We only analysed failed hubs we thought were good to start with.)
Anyway, mine were true NTNs that came out of an NTN facility. The distributor I bought them from is no longer an NTN distributor. There is zero doubt that the grease specified for this cheap, light-duty, aftermarket, application was NOT a top-of-the-line, expensive, OEM grease. "We" are nowhere big enough to get our own part number from NTN or anyone else, so I bought all I could afford and repacked them with Infinitec 152. They were always seemingly bombproof, not just on SMs, but any Miata-based thing we put it on.
The "old" NTN part number changed to a "new" NTN part number since then. Why? I have no idea. Two years ago, Timken 513152 "looked like" the same bearing, including NTN markings, and my customers generally had good luck with those bearings - but I never tested them. I haven't gotten to the bottom of what happened to Timken 513152 since then, but Timken and Autozone appear to have divorced, and pricing, availability, and secondhand reports on the "appearance" and performance of Timken 513152 has been all over the map in the past 2 years. I've been busy with everything else, and haven't done anything of substance on hubs.
You may very well be right, in that there simply may not be an "OEM Quality" version of the Mazda NA01-33-04X anymore. This bearing may have simply been shipped off to "Aftermarket Hell" with no engineering oversight at all, whereby unless the Autozones of the world get massive warranty returns on their private-label Duralasts and such, there won't be any reason for anyone to look at one of these bearings failing or bother with improving it. If Mazdaspeed could get Mazda to lean on THEIR supplier to treat this bearing like an OEM bearing, it might help. But, realistically, why would they? For a few hundred units a year? The bearing wasn't designed for what we are doing with it.
Having said that - there may ALSO be something causal in the tires, or the wheels, or the repacking, or the studs, or the torquing, or whatever.
It is mathematical fact that higher offset, shorter tires, less grippy tires, will reduce bearings loads. I know nobody wants to spend money changing those things, but hubs and DNFs and time spent repacking and replacing are money, too. I'd love to be part of the bearing solution, but I don't think it will be cheap, and the payback will be a long time away.
Here's an idea: If Novak is the bearing killer, get him some low-mile boneyard bearings. Have someone "lucky" repack them and torque them for him. And see what happens.