Why the Web Must Die

by Jack Seay - July 10, 2001

The World Wide Web is incredible, but it must die. It is amazing almost beyond belief, that is, unless you know the history of hypertext; of which, Web pages are a dismal failure. Its' greatest strength is the fact that it uses the Internet (a network of networks) to connect millions of documents. Its' greatest weakness is its' lack of flexibility.

It is difficult to edit, to show alternate views of, to display with different formatting, or add your own comments. Its' greatest strength, the Internet it runs on, could be exploited to connect millions of documents using a superior form of hypertext. The web will probably die slowly, unless a big spider can descend on it, suck its' brains out, and spit out a new stronger, leaner animal. Perhaps a program could dissect web pages and separate the text content from the formatting, and allow partial or full reuse of the format. This is probably not ideal, and would introduce new complexity, but may have some value during the transitional phase.

I was using a hypertext program called Thinker for several years before the Web existed. Thinker is not designed to be a world networked hypertext, but it does highlight several capabilities the Web lacks. Everything in Thinker is editable all the time. There are not separate browser and editor programs. It does outline processing with dozens of levels of outline that can be hidden or shown locally or globally. And creating links and labels (or anchors - the destination of links) is amazingly easy and much more powerful than Web links.

The Web doesn't map to the way I work. I like to set bookmarks at specific locations in a book, highlight and underline, and write comments in the margin. I might want to draw pictures or diagrams as I write, mark through deletions so I can see what I deleted. I may be reading ten books at a time, and want to immediately pick up where I left off in any of them. I would like instant definitions, synonyms, examples, and explanations. But more important than mapping to how people work on paper is how they think; trying to see the big picture and pull up various obscure thoughts when needed.

If you visit any university library, you will find hundreds of thousands of books filled with mathematical equations that are very difficult to incorporate in web pages or email. The vast majority of the best material written or recorded in other media is copyrighted. The web includes no good system of paying royalties and thus excludes most of the best material. Many hypertext systems handled the problems better than web pages long before the web existed. Many efforts are underway to add new features onto HTML to give it new life, but it only makes it more of a patched up and ugly monstrosity (a kludge).

As you read any document, you should be able to mark it up and annotate it as much as you like, if only for your own use. If your changes are made public, they should only be as a revision, with changes you made being visible, and the original author receiving royalties (if copyrighted) for every word you quote. Xanadu, in development, will support this.

The Web is easy to use for a beginner, point and click, and there is much to commend in being simple to learn to do simple things. The problem is when you want to do more than point and click. Then it all becomes very difficult and complex, and impossible to do many things I have mentioned here.

To build a strong house, you have to start with a strong foundation. HTML ain't it. Better would be a single unified data structure, cross-platform, that not only allows, but encourages the types of "basic" qualities I'm discussing, most of which are highly discouraged by the Web, with it's HTML foundation.

If this document were in a hypertext that allowed outline processing, the "executive summary" (this article) could be at outline level 1 and read first. Then the outline could be opened to level 2 for reading footnotes and definitions in context. Then level 3 could examine in more detail how these features are implemented in various hypertext programs, projects, and proposals. Level 4 could go into more detail of the user interaction with an ideal system. Level 5 could examine data structures and algorithms required. Level 6 could have actual code samples. Level 7 could comment on the code. Thus you could start by getting the big picture, then gradually zoom in on more and more details of structure; sort of a fractal telescope. Multi-dimensional software could allow you to select sets of elements to display in an outline, side by side, or explore in an animated 3D environment where individual dimensions viewed swap out dimension sets as needed.

In this document, I am trying to first record generalizations in a broad overview. Later, I will take more time to add references, links, definitions, exceptions, clarifications, and examples. These could be shown as pop-up windows, flown by like many bi-plane towed banners, shown faded off in the distance, or in expandible outlines.

The problems of email are many: It encourages messages with incorrect subjects, and the subjects are limited to one short line. Emails can't be cloned and placed in several categories using any of the programs I've tried. Received messages can't be annotated with personal notes, or edited to strip redundant text. The system encourages too much repetition (quoting). New messages should instead be displayed as part of the total context, highlighted as new and graphically connected to the quoted context, rather than repeating bits of messages removed from their context. As each response is pointed to, its' connections to the original and other responses are displayed. When a message is deleted, it should be possible to choose to delete all its' clones or just the current one, and allow undeletion or not. To a writer, individual letters and words are not the only important units of thought. Clauses, phrases, sentences, paragraphs, sections of chapters, chapters, character dialog, and sections of a book are also important; and should be easily rearranged, reformatted, or deleted by these groupings. Text should be able to be multi-selected for mass changes to format. Selection of vertical rectangles of text and irregular shapes should be allowed. Formatting should be fluid, to change with the focus of attention: character, theme, subplot; graphically highlighting the current priority. A more aesthetically pleasing indicator of quoting than "greater than" signs is desirable.

Chat and email are similar enough that they could be combined in a single environment, with options to mark chatting as such, since much of it will be more personal and not well thought out. This way it could be hidden if desired. But it would be kept for later reading, searching, and data mining.

What is needed is a data structure that supports a more complete set of functions: automatic version archival and comparison, royalty payments, security, authentication, encryption, multiple viewing displays, complete editability, infinite dimensional linking, two-way unbreakable links, and scalability.

The display program (front-end) can be as simple or as complex as the tasks require. The best places to look for ideas of what should replace the Web is to study hypertext programs either existing, proposed before the Web existed, or in development. Many are far better hypertext than HTML will ever be. Mirror Worlds could include a metaphor generator to automatically create clarifying examples to help you understand abstract and obscure concepts. Readers should always be given the option to be notified of revisions to a document, then shown the new and old versions side-by-side, with graphical display of exactly what has changed.

Complete means all the parts are whole.

Links to relevant web pages can be found at http://groups.yahoo.com/group/Hyperworlds/links

Back to main page