Files as UI vs API - compares attributes of iCloud vs Dropbox. It starts on an interesting note - the model of a file system in the UI is dying, and should be let go. Beyond that it looks at mappings of each system to a file system from an API point of view and compares the successes of each.
I find the initial thread the most interesting. Drop the mental model of a file system - which maps virtual concepts of files and directories to a physical model of papers, folders and file cabinets - and replace it with...what? This is a paradigm shift for me.
I have to admit, I loath, hate, nay, despise looking for things. If I can't find something easily, it's only about a minute before I start growling and muttering things my mother would disapprove of. On this basis, I like the idea that I can save myself from thinking about where to put things or, where I have already put them. But how do we do this? It's non-trivial, since humans think of "things" and once they think of a thing, it's place is a very closely associated and implicit property of an object. Look at the definition of a Uniform Resource Identifier (URI) - it's built in part upon the URL - Uniform Resource Location.
An obvious approach is to focus on applications - tools - that hide the details of a file system from you? That's not too hard to imagine, but is that a paradigm shift? And is it a strong enough model to prevent "fraying" at the edges, where every once in a while, for edge use cases, you need to think about where your stuff is?
What about "things" (formerly called "files", "data", etc.) having names, which don't include any "where" type information. It's all "what" and "how". An object has a name, and perhaps the name includes information on the type of thing it is. The how information is "Is this backed up?", "Is this available from any device or just one?", "Is this private, really private or public?"
I find the initial thread the most interesting. Drop the mental model of a file system - which maps virtual concepts of files and directories to a physical model of papers, folders and file cabinets - and replace it with...what? This is a paradigm shift for me.
I have to admit, I loath, hate, nay, despise looking for things. If I can't find something easily, it's only about a minute before I start growling and muttering things my mother would disapprove of. On this basis, I like the idea that I can save myself from thinking about where to put things or, where I have already put them. But how do we do this? It's non-trivial, since humans think of "things" and once they think of a thing, it's place is a very closely associated and implicit property of an object. Look at the definition of a Uniform Resource Identifier (URI) - it's built in part upon the URL - Uniform Resource Location.
An obvious approach is to focus on applications - tools - that hide the details of a file system from you? That's not too hard to imagine, but is that a paradigm shift? And is it a strong enough model to prevent "fraying" at the edges, where every once in a while, for edge use cases, you need to think about where your stuff is?
What about "things" (formerly called "files", "data", etc.) having names, which don't include any "where" type information. It's all "what" and "how". An object has a name, and perhaps the name includes information on the type of thing it is. The how information is "Is this backed up?", "Is this available from any device or just one?", "Is this private, really private or public?"
Comments