This is an idea from the HumaneInterface by JefRaskin as implemented in the CanonCat:

Assume the computer is only used to edit text. Why bother with the concepts of files, directories, drives, windows? Let the interface be just the text itself. At the beginning, the text descrives the menu for the user, and then the remaining documents follow one after the other, separated by special markers. The keyboard provides special keys to jump from document to document in the one text. Documents are not arranged hierarchically (in folders) but linearly: One after another.

A hierarchy could easily be implemented by providing more special markers and more special keys to jump to them. Like a book may have chapters and sections and subsections, the one text may have have categories or topics.

Note that you can linearize a hypertext such as this wiki by just appending one page after another, allowing local links to point to anchors within the same document. The SiteMap of this wiki has a link to such a linearized version. Clearly, a linearized monolithic hypertext is just as usable as the original hypertext split into various nodes that can only be viewed one at a time in a web browser. The main drawback is loading time.

The same is true for editing. If monolithic hypertext allowed in-place editing, being linearized would not be a drawback to editing. Without in-place editing, the loading time is the main impediment.



No, the cognitive complexity and uglyness of that construct is the main impediment. This would be more confusing, not less, to most users. Computerized text would be seen as a makeshift, nerdy expedient rather than an exact analog of printer documents, books, etc. People would constantly want to print out stuff so that they could "keep it separate" by putting it in physical filing folders. People would be afraid of mixing up their personal diaries with their corporate email.

Better to make the presentation of the information either match physical analogs, or to match mental representations.


Raskin's major point is that metaphors will always confuse users. The current desktop metaphor and filesystem organization requires far too much effort from a person only interested in writing text. He argues that if you want to write various documents on your computer (and nothing but that), then the computer should boot up and allow you to type text immediately, and it should allow you to find old documents easily. None of this requires a desktop or a traditional filesystem.


I did this about 5 years ago. [1] Note that it features Multiple Categorization. You can:

  1. Find something you like.
  2. Look at it's category tree.
  3. Search the rest of the document for similarly categorized items.

After about a year of having this online, a casual web browser saw it, and wrote an intelligent parser and browser for it with GTK+. Needless to say, there were a lot of redundant / mis-spelled categories in the tree.

I agree with Bayle though; I wouldn't want my entire computer to be organized like this, though.

EditNearLinks: CanonCat JefRaskin HumaneInterface