Apple needs to fix their QA, and fast. No amount of marketing can undo the damage I've seen over the last few years.
Tight integration between OS and hw is mostly used to half-arse firmware interfaces, with huge amounts of "quirks" surfaced by hackintosh enthusiasts.
What I do find surprising though is that there is no bulletproof rollback mechanism.
It shouldn't be too far-fetched to imagine a room full of older hardware where it's the job of some poor intern to update each device during the alpha/beta phase.
Bricking supported hardware on a software update goes directly against Apple's core business strategy and should have been top of mind for the QA team.
You're too quick to let them off the hook.
Compared to the ecosystem of Windows-supported devices, Apple only has to serve updates to relatively few SKUs. If they can afford to buy back a hundred million dollars' worth of stock every year, they can afford to test upgrades on affected older devices before rolling them out.
When an upgrade is supported on an Apple machine, then there should be 100% success -- or at least, as you suggest, a painless way to catch the upgrade failure and roll the device back. They sell the hardware and develop the software, so they have to own the problem. No mumbo-jumbo. No excuses. No passing of the buck to the customer who trusted them at the moment of downloading/installing the upgrade.
If they can't guarantee success for an OS upgrade on a particular machine, then they shouldn't offer it for that SKU -- and they should lower the retail price of the MBP because it doesn't "just work." People who own these machines paid a premium for them over every other comparable laptop.
I have a late-2013 MBP 13 (one of the potentially affected machines), on which I am currently running Mojave because Apple has been guilty of this kind of garbage so many times in the past. Think of how horrible it would be to put your daily driver laptop in the shop for a few days when you have important work to handle. That's the reason why Apple can't just mail it in when rolling out OS upgrades.
I have had similar situations in the past, and you know what the result has always been? I stop buying hardware from the company when it becomes clear that it doesn't give a damn about what happens to the customer. Bose rolled out firmware to the QC35 that nerfed the ANC, and then pretended that nothing happened -- essentially gaslighting the customer. I never bought another pair of Bose ANC headphones after that (I had purchased 5 pairs in the year prior to that, as gifts and for myself). Google rolled out an Android upgrade (Oreo) that immediately sent my Nexus 6P into a Bootloop of Death upon installation, and then passed the buck to Huawei. I swore never again to buy a Google hardware product as a result.
I just bought a MBP 13 a few days ago, because despite all of the BS that Apple has done, they have always turned out to support the customer in the end. But if their defense sounds anything like what you are suggesting -- that there is always some collateral damage when an upgrade is rolled out because QA is not a perfect process -- then I would rethink my purchase decisions in the future.
billion* is what I meant to write.
Well, until one of these sub models gets brucked.
But putting all that aside: the new sealed system partition is a step in that direction (ROSP was partially about getting clean separation between the system and user files to make snapshot-based updates possible). With snapshots if the update fails you still have the previous snapshot to fall back on - hopefully that will make these kinds of failures less likely in the future.
By the way my HP laptop from 2014 did well with every single Ubuntu LTS from 12.04 (I bought at the beginning of the year) to 20.04.
Learned a hard lesson that day.
By the way, did you ever recover the laptop or your data?
Why does Microsoft dominate the enterprise and not Google? I think the above has almost everything to do with it.
Apple has kind of walked the middle road between Microsoft's extreme backward compatibility and Google's "kill off everything that doesn't earn at least a billion". They support old hardware a lot longer than Google, but they also break a lot of things with their OS updates. Apple also has a pretty legendary track record of successfully shifting CPU architectures without breaking popular apps at the time, something that's pretty much unheard of elsewhere.
Which other SW/HW companies have switched CPU architecture? I am not sure there is a point of comparison here, and there has only been 1 case so far, the M1 is upcoming and we don't know yet how well that will work.
Linux works pretty well with client qemu specified as the architecture handler for non-native binaries too :)
I also said 'architecture change' so by your count that would be 2. But Rosetta was only for PowerPC to x86 wasn't it? There seems to have been emulation for the 68x chips but was it seamless without breaking anything?
I mean any company could change architecture but rely on a customer using an emulator that isn't really not breaking things.
So by my count, 1, but if I am wrong and the 68x translation was equally as seamless and breakless than 2.
But to say it was unheard of you need to reference all the other companies that transitioned architecture and broke stuff. I can't think of another general computing supplier who did that but I might be wrong - can you give an example?
There's also Palm, which successfully switched from Motorola 68000 to ARM, without breaking popular apps at the time.
To me while it's nice, it's not an impossible achievement and often likely uses somebody else's code to get there :)
I think apple receive far too much praise for things they've done which are often not as amazing as they seem and too little praise for other things (e.g. their great work on the 5nm SoC hw) as well as far too little criticism for biz practices.
Trying to set things a little more on an even keel :)
This is the first time I've heard that, and I've never experienced it. If you have a compilation of "driver support disappears" issues, I'd like to read it.
It's very common for folks to be using 5- and 6-year-old Macs with a recent version of macOS. macOS Big Sur supports 7-year-old Macs.
I've done Mac IT for a long time and I've never seen this be the case.
To be honest, I've experienced this a ton of times with Windows, where the only driver available was for an older version of Windows and not the one the customer has.
I have a nearly 10-year old iMac and I never experienced what you're talking about.
Apple has warned developers for several years that support for 32-bit apps (for example) was going away. Same thing with kernel extensions.