A) Devices are just slaves that stream content from a cloud server and only know how to operate on basic input hardware but do no "thinking". OR
B) Subscribers are forced to net boot into a login screen that once credentials are supplied are given an obfuscated(headers, standards), randomized(load order), and even encrypted. Then the OS loads your saved content being either encrypted from the last session or decrypted being some sort of IoT standard. Now only the system understands how to communicate with itself which is where some sort of internet cloud based application layer downloads the code*not to be confused with the content* of the applications you are subscribed too that are custom to your OS. Even programmers can write non specific code and submit it to said company too in theory run on any completely "random" system.
Both of these models have been tried numerous times before, most recently with things like the Chromebook, with what could be politely described as mixed results.
The former is basically a return to timesharing
and dumb terminals, which was terrible in 1967 and would still be terrible today. The entire computer revolution got its impetus from trying to get away from that, even if it meant using a box the size of a small microwave oven
which initially had 256 bytes of RAM, a 4-slot passive backplane with the CPU and memory each filling one slot, no permanent storage, and no I/O except a set of toggle switches and red LEDs, all for the price (as an unassembled kit, soldering gun not included) of $440 US (1976 dollars - about $2000 today). Timesharing was just so bad
that the Altair sold faster than they could make the kits, despite being about as useful as a Yugo with the axles removed.
The latter are what were called 'diskless workstations' (I think you can guess what people actually called them
), 'thin clients', and 'netbooks', all of which were conspicuous and costly failures in the market. The Chromebook and the various modern tablets are the closest thing to a successful version of that, and you will note that even they have significant amounts of local storage and can (technically) operate offline.
It isn't so much that it is a bad
idea, per se, as it is an extremely impractical and onerous solution that most consumers hate with a passion. Most users would rather put up with their computers running like a dead tortoise and having their credit card information and SSN stolen about once every nine months, than deal with either of those solutions, as foolish as it might sound. The reason PC security is awful isn't because PC operating systems are awful (well, not just
because of that), but because 99% of the users are less inconvenienced by poor security than they would be be replacing what they have
. Which should tell you all you really need to know about how actual consumers see the world vs how we as engineers do - and if you think they are the crazy ones, you are probably in for a rude awakening.