This is an effective, simple solution if the application doesn't have transactional constraints and authorization. The user is in full control of the data - and that's a good thing.
It becomes a bit more complicated if the data has to be in a specified shape and can only partially be authored by any individual due to authorization requirements. Having a single source of truth on a server is quite a bit easier to reason about and design in this case.
I don't know where Obsidian stands on this, but I assume that if it has collaborative features or complex data constraints it has to do some quite sophisticated stuff to make it work reliably.
> if it has collaborative features or complex data constraints
It has neither of these things - it is aimed at individual (non-collaborative) use, and is intentionally very open and structureless. It is uniquely positioned to succeed with their chosen business model.
It becomes a bit more complicated if the data has to be in a specified shape and can only partially be authored by any individual due to authorization requirements. Having a single source of truth on a server is quite a bit easier to reason about and design in this case.
I don't know where Obsidian stands on this, but I assume that if it has collaborative features or complex data constraints it has to do some quite sophisticated stuff to make it work reliably.