Thanks, that was on my list. And itself could fairly easily be addressed through a window manager or contained environment which served touch-only apps.
Other factors so far as I can read them:
- Insanely good power management, particularly relative to features. A dumb/feature phone can of course see much better battery life, but offering a small fraction of the services of a smartphone. (Whether or not this is a reasonable trade-off is another question.) Much of this is through chip hardware optimisation (meaning that emulated Android environments would perform poorly), and aggressive culling of background processes (which means that a full-featured desktop OS would perform much more poorly on its own as well).
- Various HW phone features, particularly display quality and cameras. To a lesser extent, audio output.
- Extreme wireless data dependency, whether cellular networking (4/5G) or WiFi.
- Cloud-based storage of virtually all data, largely implemented/enforced at the app level. Corrolary is that it's quite difficult/challenging to share data or files between applications, not to mention that it's often a bad idea to do so.
- Identity attestation, including at the network (SIM) and hardware level. I'd think that an NFC chip, identity token (e.g., Yubikey), or worn identity token (e.g., NFCRing) could stand in for this.
- Generally an update/upgrade model that the mass public seems to find acceptable (though I ... have my doubts).
And, though I said I'd exclude it: the apps ecosystem, both in terms of popular social networking tools, and a rich market for developing highly-specific applications for particular niche needs, whether commercial, institutional, activity-specific, or recreational.
The touch-based environment really seems like it should be possible to meet within a desktop/laptop context, and given existence of AndroidOS (or similar/compatible) emulators / environments, I suspect is.
Battery life is probably the biggest direct technical challenge. While reading your reply the thought occurs that with SoC/SBC systems, it might be possible to run a low-power mobile module which is independently active when mobile-only services are required, with data sharing through storage to the main system.
Further offloading comms load (4/5G networking, voice comms) to external devices (mobile hotspot, dedicated feature phone) could provide yet further optimisations.