File systems have to evolve. These days, file system means two things: an application-independent API to access common documents, and a hierarchical local storage. But it doesn't have that way.
The best thing I've seen in file system evolution is KDE's KIO: Any KDE application can take any KIO url and use it; all file operations are asynchronious (even if you open local files, and that's very nice even for local files that are big), and any program can use network resources as easy as local with little to no effort.
But we should improve on that: a heterogenous user file system should provide discoverability (e.g. your social network photos are automatically available in any program once you bind the account, and you know where to find them). File system branches restrict some operations on files or hint on their cost (scanning a huge photo bank is a very expensive operation; you can't access the contents of audio files inside a streaming service but you can play them in the program). There also should be other ways to organize files than just dumb hierarcheries (imagine a search box in place of a folder, you need a query and then you enter the search results; or you can have tag cloud in your file system)
There's a great deal of work of innovation here and nobody does it at the moment, so it seems.
Sorry for mistakes 'cause I'm hurrying to go to bed :)
The best thing I've seen in file system evolution is KDE's KIO: Any KDE application can take any KIO url and use it; all file operations are asynchronious (even if you open local files, and that's very nice even for local files that are big), and any program can use network resources as easy as local with little to no effort.
But we should improve on that: a heterogenous user file system should provide discoverability (e.g. your social network photos are automatically available in any program once you bind the account, and you know where to find them). File system branches restrict some operations on files or hint on their cost (scanning a huge photo bank is a very expensive operation; you can't access the contents of audio files inside a streaming service but you can play them in the program). There also should be other ways to organize files than just dumb hierarcheries (imagine a search box in place of a folder, you need a query and then you enter the search results; or you can have tag cloud in your file system)
There's a great deal of work of innovation here and nobody does it at the moment, so it seems.
Sorry for mistakes 'cause I'm hurrying to go to bed :)