It doesn't make a lot of sense to me because the software I run on my watch is nothing like the software I run on my desktop.
To use the typical car analogy, it feels like designing a motorcycle that you can attach a second set of wheels and a cabin to turn it into a car. Or a second set of wheels and a bed to turn it into pick up truck. Or wings to become an airplane. Or a hull to become a boat.
But the reason why you run different software on your watch is because it cannot be a desktop. What if it could?
All your examples sound ridiculous, because we know that there's no technological way to make it work well. But suppose there was one. Wouldn't you want a car that could turn into a monster truck, minivan, boat or airplane, if all of these actually worked well? Especially if it was all for a fraction of the combined cost of these things today?
My car is useful. An adaptable car would take more design effort, so likely cost more. I have no use for a monster truck, a boat, I can't pilot an aeroplane, and if I did sink or crash-land my car then I'm out one car that I have use for, and need. (Wheras, if I hire and crash a boat now, my car is still fine).
And I don't want to imagine the insurance, MOT, or servicing cost for a vehicle which has to meet flight safety regulations but which is also waterproof and which has brakes able to stop a full minivan.
Minivan, might have limited use - but where would I keep the minivan part while it isn't on my car? And if I had somewhere to keep it, I may as well get a trailer or a roof-rack right now to get most of the benefit for almost none of the cost, or hire a minivan for a day if I need to carry more people instead of more stuff.
I think you're stretching the analogy beyond the limit of its usefulness :)
A phone is always on you anyway, usually in your pocket. What disadvantage do you see from making it the "universal compute unit", so to speak, driving your tablet, your laptop etc - which then just become combo display/entry devices?
If you don't need a device in a particular form factor, then you just don't buy one, and the question of storage doesn't arise. People can make arrangements depending on their specific use cases - someone might do phone/tablet, another person might do phone/laptop, yet another would span the full gamut with phone/tablet/laptop/desktop etc. And if you think that you don't need something, and then later find out that you actually do, it's easy to add that capability.
Cost is not really an issue here. Phones today are already powerful enough to serve as compute devices for all desktop tasks for the majority of the users, so we're not talking about anything more expensive than usual. And those user for whom it is not the case - developers, 3D designers, hardcore gamers etc - can still have dedicated devices, where it makes sense (i.e. usually for a desktop and/or laptop).
What disadvantage do you see from making it the "universal compute unit", so to speak
The very fact that it's a compromise means it will be worse than each of the individual solutions, but it will enable something new - and I'm not seeing anything very desirable in this space to be worth the compromise.
What exactly the disadvantage is depends on the exact compromises made, but USB-C is a small socket, many more connects/disconnects will lead to failures. Wireless links will struggle to be good enough for display output. Losing your phone on a night out would be much more disruptive and render you digitally-stranded.
If it drives your tablet, you don't get to give your tablet to your kids to play with and retain your phone. If it drives your laptop, you can't lend your laptop to a guest. If it drives your TV to watch a film, you can't be sitting around with other people and receive a text message and glance at your phone like they can. If you're working on a fullscreen spreadsheet, you're not going to want to interrupt it for an incoming text - and now you have no phone to glance at.
In order to make it a universal compute unit you increase the amount of code on it - attack surface goes up, need for patching goes up, complexity goes up, room for bugs goes up.
So you have a phone which is small, personal and convenient, and you have an empty shell of a tablet, an empty shell of a laptop, and an empty shell of a desktop - all of them could have their own smartphone class ARM CPU for say 5-20% more cost, but without it they're all useless and taking up space and you can't use more than one at a time.
But you also have a phone where you have to fiddle with the software all the time when you switch roles - disable notifications, connect USB-C things, connect bluetooth device so you can carry on using it as a phone, plug mains power in, redirect audio out to be bluetooth only for phonecalls and system audio to go to speakers, deal with 'ok google/hey siri' listening so it won't fullscreen over your work...
And for what benefit? Where's the call for "I wish I was in a hotel carrying my smartphone and the husk of a laptop, that would be so much better than having an actual laptop". Where's the call for "I wish my iPad couldn't do anything unless I had a thunderbolt cable connecting it to my iPhone, that would be really conveninent". Where's the call for "I wish my main keyboard was a portable bluetooth foldup one" or "I wish I could take my employees' $400 laptop and instead give them a screen, a keyboard, a mouse, a mains cable, a port adapter, and a big sheet of instructions, and some more training".
I guess the disadvantage I see is that companies are awful at integration, awful at UI, and the experience on the user (me) is that many things related to connectivity and integration between systems don't work, much of the time.
