In order to uniquely reference one piece of code from another it needs to have a unique name/namespace/reference. Whatever organising principle you use for that will tend to become the hierarchy that your code is stored by.

This doesn't stop you from also accessing it in other ways. And with modern IDEs you can search across a fairly chunky codebase near-instantly, which would allow you to treat it as if it's in a database.