The nice thing about computers (Turing Machines) is that they let us simulate anything, including having better[1] computers (that may just run a little slower). The big idea behind turning the file system into a "database" is that you would be able to simulate any kind of file system you wanted, including better ones. This is basically what RDBMSs like SQL Server and SQLite do.

I also like the relational model the best, but it's just one of the many available simulations of "betterness".

[1] Where "better" means: easier to program, or more reliable, or has infinite memory (garbage collection), or is easier to use, or...