Currently I am using a file system for Linux known as ext2fs. Alternatives would be VFAT from Microsoft or HFS+ from Apple. They all use a similar system of arranging files in a hierarchy of file folders. I have found the structure of these systems maddeningly inadequate and suspect that using a hierarchy, though easily understood, narrows how programmers have conceived of organising stored content.
Content needs to be stored in one place, then the structures of how it is interrelated need to be kept separately. In this way more than one structure can be used to organise the same information. Each piece of content needs to have a unique identifier rather than being found by where it is located. In this way if a piece of content is moved, it does not lose its links to other content or programs.
While putting together the "Tronics" section of my project I was not immediately sure of how I wished to organise and present the concepts I used to create Odysseus, She. For instance certain parts of my chapter on theme could have been independent sections of their own rather than sub-sections and I did test organising them in several ways. I could not keep these various organising structures intact, just in case I changed my mind or because they both were valid ways of thinking about the material. I could only be committed to one structure. When I moved the sections about the "unities" into the "theme" file, the various programs I used to create those files now had broken links to where those files were located, and I would have to step through the hierarchy to get to them.
This problem extends to the way content is stored and delivered over the Internet. Content is found by where it is physically located under a hierarchy of place. For an address such as <http://www.glasswings.com.au/GlassWings/modern/shapes>, the content there is found by the request for that file locating the glasswings.com.au server, then the "GlassWings" file under which is filed the content of the ``Modern Adventure'' section of which one piece is the shapes file. Should the shapes file be moved to another location, even though that file still exists, any links made to it will be broken and therefore the file will not be found. Content at the moment is primarily given a Uniform Resource Locator (URL) rather than a Uniform Resource Identifier (URI). A filing system using a Uniform Resource Identifier could find content by the name given to it rather than by its location. Ted Nelson broadly discusses the need for such indeterminate structures for content storage, organisation and retrieval in his work Literary Machines [Nel80] first published 1980.
Copyright © 1999 Katherine Phelps