Why is it so hard for Android OEMs to keep supporting their devices? I don’t buy that it’s just planned obsolescence. As Apple invests in making devices usable for longer and longer this doesn’t seem like a competitive move.
Because these devices are usually being supported by teams of only a few people. Eventually someone decides that a new project needs all hands on deck to meet deadlines and people get reallocated internally from the "least important" projects.
The fixes are simple, but they all share one problem: they cost money. Hardware orgs are almost universally run by the most penny-wise pound foolish people imaginable, so that problem is completely insurmountable to them.
There's other issues too (e.g. upstream vendors don't like upgrading and charge absurd amounts of money for it), but the main problem is simply a lack of institutional priority to do the work.
Well, nothing you say is wrong, and it isn't simply that they are foolish. Hardware companies just can't figure out a way to give us free OS upgrades and make money. They generally stop making money the second the check clears. Apple has that one figured out, the longer you are a customer the more Apple services you buy. Google has a fundamental different model with Android where they are trying to make money and coerce their integrator partners to keep upgrading, but the money is way different.
Xiaomi doesn't make money from HW sales either, they rely on services revenue. They also have the same same 3yr support life every other Android OEM commits to.
It's not foolish, it's shortsided. At the scale major OEMs operate, even a dozen engineers per device generation would be a rounding error.
Except in most cases one of those years hardly counts since many non-top-of-the-line android phones ship with an already outdated OS (though this too has been slowly improving, as has the time it takes to get them up to the newest version).
> Hardware companies just can't figure out a way to give us free OS upgrades and make money
The biggest question is : why does it take so much efforts to design an upgrade ?
Unfortunately Google never made it mandatory for chipset/hardware designers to have an abstraction layer (similar to BIOS/UEFI/ACPI/system tables) that would make a single Android kernel and OS install and work on every devices...
(And I have criticized Microsoft a lot for many things, but I have to acknoledge that they have an amazing backward-compatibility track record, even for device drivers, while Linux has its stable_api_nonsense policy...)
I have no technical knowledge of Android but from a user perspective the whole system reeks of overwhelming technical debt. Updates constantly break old features and need rapid roll-forward patches. The whole OS is plagued by irritating slow-downs, screen flickers and weird interactions. It feels like no one knows how to tackle major changes so they just update the fucking volume UI every 6 months and call it a day. Android TV degrades continuously. It feels like no one cares.
I don't know anything about android tv. You can blame Google's garbage A/B testing addiction for that.
Android on most devices is very stable. If it's not (manufacturers fault), just install LineageOS (basically stock ASOP, with device drivers)
Because Microsoft wasn't and isn't an Android OEM.
The Surface Duo was supposed to be a Windows 10X companion device until at the last minute they hired a firm (Movial) to port the metal to Android. Halfway through the port they brought on that entire division as MS employees and that's who basically was running the Android porting shitshow.
When was the last time Apple released a radically different device ?
Ask differently how long was the original Apple Watch supported ? Or the original iPad ?
I don't think any of them had more than 2 or 3 years of suport as the hardware radically changed in the next revisions. Faulting Microsoft for failing to support a first gen innovative hardware concept for more that a few years is kinda harsh.
The fixes are simple, but they all share one problem: they cost money. Hardware orgs are almost universally run by the most penny-wise pound foolish people imaginable, so that problem is completely insurmountable to them.
There's other issues too (e.g. upstream vendors don't like upgrading and charge absurd amounts of money for it), but the main problem is simply a lack of institutional priority to do the work.