Home Page








Features Contents

13th April 2002


Brian Grainger

email.gif (183 bytes)


There is no value in storing documents electronically unless we can find the information we want later.

Since the early days of DOS technical users of PCs helped the retrieval process by organising documents by directories, (or folders in Windows speak). When the more general users of PCs came along they were encouraged to do the same. The metaphor was to think of your disk space as a filing cabinet with folders on your disk drive emulating the folders in the filing cabinet.

As much as the techies tried to encourage this metaphor some users just could not get it. In the same way that some people will have desks at work covered in paper there were users who simply saved their work to wherever the save dialog box was pointing at the time. A common complaint heard by the Help Desk was, ‘I have saved my document but I cannot find it now!’ If the concept of disk folders were understood this complaint would not have been heard.

The computer equivalent to the desk with paper strewn all over it is the repository where all files are stored as a lump. To get information back from a repository one has to use search tools.

At the moment Windows allows users to follow either approach, although in recent versions folders have been de-emphasised. My Documents may be implemented by a folder but it is really a repository. The concept of segmenting My Documents into categories is not one that is followed by most general users of my acquaintance. Similarly the Inbox of the e- mail application is another repository and this tends to get overstuffed as well. The Internet File Cache is a fully blown repository. It is implemented in 4 folders but Windows Explorer will only show one complete set of files, irrespective of the folder you display.

In Windows the Find button is the search tool. In Microsoft Word the search tool is Find on the Tool button hidden within the File – Open dialog.

The World Wide Web is another repository. Searching the web to find the information you want is the top problem for most users. The web is slightly different in that hyperlinks also help us to find information. However, this can have an inherent bias, as the links reflect the personal leanings of the linker or search engine.

There is a major problem with the folder approach and it occurred with the filing cabinet as well. What do you do when a document is relevant to 2 (or more) separate folders? In the filing cabinet an easy solution, if the document was not large, was to photocopy it and place a copy in each of the relevant folders. The computer equivalent is to keep duplicate file copies in different folders. Both these solutions create a maintenance problem. It is easy to update one copy of a document and forget to do the same to another copy.

In the repository model there is a solution to this problem. You can create folders but all they contain are links to the documents applicable to the folder, which are held in the central repository. In this way there is only ever one copy of a document to maintain. It is the links that are duplicated. There is an additional process when maintenance occurs. You have to check all the folders a document is linked to in order to check whether the change is relevant to all the folders! The problem now is what to do if the change is not relevant to all the folders. Do you now have separate documents in the repository or do you keep ‘sub-documents’. One for the common bits and one for each differing bit. You then link the sub-documents to the folders accordingly.

Which is the best approach – a folder or a repository? I guess it is horses for courses. As a well organised person I like folders. Others prefer the repository approach. However, with the coming of the Internet I can see the benefits of linking and I am slowly rethinking my storage philosophy. I still like the concept of folders but using them as containers of links to documents in central repository is very appealing. I have just not though through all the requirements of my storage system to this model.

What worries me is that the tools to support the folder model are slowly being removed. The web-centric philosophy of the major software players is accelerating this. I feel that when the tools to support the folder model have gone the well organised will not be more effective than their colleagues. There will be a level playing field, but at the lower level of effectiveness rather than the higher.