The idea of a document is limited because it is based on paper.

A document is a wonderful way to organize and hold content when you are printing on paper. People are used to documents. Documents are a familiar way of conceiving of a lot of related information.

But documents do not work as well when you are assembling, updating, pruning, customizing, and personalizing content on a number of web sites.

As we move onto the web, we must say goodbye to the document model, recognizing our fondness for documents as a kind of nostalgia.

We must recognize the benefits of documents, and the problems, as we move from the document model to a world of informative objects.

The joy of documents

The author says "I like documents because..."

• I can stuff everything into one gigantic document, and know that it is complete.

• I can follow common patterns, developed over centuries, but I can cheat, adjust, revise, and tweak the standard form, as I go. I can get creative with the standards.

• I can think in terms of large structures- the whole document, or the whole chapter-and fit information into those patterns, loosely.

• I can see a satisfying bulk, a solid, tangible result of all my work.

• One document fits all. We publish it, and everyone reads the same document. Each person gets what he or she wants out of this treasure chest.

• I can develop an extended argument in order, assuming that everyone has read the previous pages, as I move forward. I can build on what I have already written.

• I can format whole stretches of paragraphs my own way. I can apply the same style tag to paragraphs that have very different purposes, so they all look alike.

The user likes documents on paper because...

• Everything is there. Now if I can just find what I need.

• Because it is printed and bound, I have a handy interface to all the information about the general topic.

• I know how to get around poor organization, by flipping through pages, using the index, or turning to the table of contents.

• I can watch an argument develop over time, as I move forward page by page.

The problems with documents

Documents emerged as paper containers of information. They do not work well as a container for electronic content.

Documents frustrate users who want to take advantage of electronic text by:

• Interacting with even the finest grain of content (linking to it, displaying it, hiding it, displaying only that chunk)

• Searching with advanced filters to reduce irrelevant hits

• Searches cannot be targeted at a particular type of information (this topic, within instructions, but not conceptual overviews).
• The title of the document may not appear to have anything to do with the search criteria. Why did it come up?
• Even advanced searches do not let users examine anything smaller than a document. So their hits are all full documents, each covering dozens, hundreds, or thousands of topics beyond the one the users are looking for.

• Hiding and revealing content to show only the amount needed right now

• Personalizing the structure, format, and tone of information

• Linking to a particular fact, not a large page within which the fact lies buried

• Seeing immediate updates on the fly, without waiting for a whole document to be revised

For writers, updating grows more erratic, difficult to manage.

• On a particular page, one or two facts get out of date, but we have no way of tracking that, discovering that; we wait to edit the whole article, or document, and post it again.

• Updating tends to wait on a new release, just as we used to wait for a new edition of the book, instead of going on continuously. Hence, we have months or years of increasing obsolescence, a moment of currency, and another long period of decline.

Documents are not organized with the rigid consistency of records in a database.

The company may put standards in a styleguide, but no one really has to follow them. Result: different styles, and, worse, different structures on pages that purport to deal with the same content or offer the same publication.

Different documents on similar topics may take up vastly different space, show different organizational patterns, look different, and include different kinds of graphics, even different logos.

• Inconsistency of structure from document to document slows down navigation, discovery, understanding, and use.

• When a lot of documents get published on the same site, the inconsistency of organization (and sometimes tone, as well) becomes obvious, because the user can click back and forth very quickly. What was a reasonable freedom, allowing individual teams to develop their own structures, becomes an interface problem.

• Users see the differences between pages, sections, areas of the site, and naturally, they begin to wonder what is going on.

• They get confused.

• Their model of our structure begins to fall apart.

• They can no longer rely on their developing model of our content as a guide to movement through our space.

• They leave, or if they have to stay, they feel frustrated, angry, and resentful.

• They may even think of us as incompetent.

• Inconsistency of structure baffles software, too,

• If the robot expects a certain structure, and does not find it, the search fails, or yields lousy results.

A document gives a general answer to a specific question.

Users ask specific questions, but get general answers.

• Instead of answering a user's particular question, we provide a whole document, which, we hope, contains the answer somewhere within.

• Document sizes become unwieldy, as we cram thousands of related sections into the same volume, bursting its binding. What we gain in the convenience of a single volume begins to slip away under the inconvenience of browsing or searching through this pile.

A document often appears as a continuous flow, without much division into section.

Skimming and scanning are difficult because the material has not been divided up into a lot of discrete chunks, with subheads.

• Few people can skim text on the screen the way they do print on paper.

• Indexes from books get dropped in the move online, and are rarely replaced. The value added by a second hierarchical analysis of the material is lost. (Search results are not indexes).

• Browsing the menus and searching become the main way into a site, followed by link jumping. Few users make any effort to understand the way your site is structured, and after a few associational jumps, they generally do not know where they are in your structure.

Documents pose a particular problem for authors who want to reuse content.

• Individual elements of a page cannot easily be reused elsewhere.

• Authors may not be able to find the pieces they could reuse.

• Authors may copy, paste, and edit beyond recognition.

• Authors create the wheel again and again, each time with little variations.

Users complain about document-oriented sites:

• I have to guess which document will contain the fact I want.

• I cannot easily disregard the bulk of the document when all I want to read is one paragraph.

• When I search for a topic, I get too many results, because the same topic is mentioned in so many documents, and I cannot filter more finely than that.

• The document may not be broken up into functional components, visually, so scanning goes slow.

• I cannot tell the system to hide certain components of all documents of a certain type, or change the order of components, to meet my own preferences.

• I have to wait for someone to update the whole document, rather than having new information plugged in on the fly.