One-way Doors, Teleportation and Writing without Prepositions: an analysis of WWW hypertext links

Jaime Henriquez

As a method of communication, the World Wide Web is unusual in several ways - in its delivery (by computer rather than printed page, radio waves, etc.), speed, audience, multimedia aspects, interactivity, connections to databases, and so on. Attempting to assess the effects of this chimera on communication is daunting in part because it is difficult to know what new aspect to examine first. As a starting point, I have chosen to examine the web's hypertext link, in its current form.

New media create new forms of communication. Each medium both enhances and restricts different aspects of communication. The key to understanding how communication via the web differs from other forms of communication lies in the link. If we look closely at the link, we find in its traits clues to the ways in which communication changes when conducted over this new medium.

1.1. The role of links

To begin with, a World Wide Web link is a connection between some point on one web page (which I'll call the A page) and a point (usually the beginning) of any other web page (the B page). When reading the A page, with an appropriate web browser program, clicking on the link causes the B page to be displayed instead.

Links exercise a considerable influence on what material a reader sees, and in what order. While there is no limit to the length of a web page, most are no longer than a few printed pages. When the reader finishes a page, links provided by the author offer the most obvious options for continuation, though not the only ones. Any page which can be reached by link can be reached directly, without the link, so long as its address (its Universal Resource Locator, or URL) is known. Directly specifying the URL of the page you want to see requires some knowledge, however, and those who regard computers as inherently mysterious frequently choose not to take this route, especially as URLs are typically a 20-30 character near-nonsense string which must be typed in precisely as given. Clicking on a link is easier.

Web page URLs are inherently uninformative - they indicate nothing about what you will see when the page is displayed (1). Telephone numbers pose a similar problem, giving little or no indication of who will answer the phone if you dial the number, or how you might find that person by some other means (by mail, for example). The lack is corrected by telephone directories, or address books, which supply some or all of that missing information. Links do for web pages what directories or address books do for telephones, i.e. they give some information about what is on the B page. The link implies, at a minimum, that the author of the A page thought the B page to be of some interest, for some reason. One can call any phone number at random, but the majority of calls are to someone, made with the help of a directory of some sort. Similarly, most web pages are displayed with the help of a link. Links invite the reader to new pages and, as we shall see in the discussion of the link's directionality, the invitation is a forceful one.

Links, then, have a strong impact on a reader's choices and, ultimately, what s/he sees. An analysis of the World Wide Web link can shed light on the similarities and differences of communicating via the web rather than other media, and at the same time suggest guidelines for creating effective web pages. While some of these conclusions and suggestions will be familiar to developers of web content, even those who are familiar with the suggestions may profit by a better understanding of the reasons behind them.

Three aspects of the World Wide Web link have strong effects on writing. These characteristics can create problems for both authors and readers. They are:

• the link is directional; • the link is instant; • the link is generic.

1.2. The Directional Link

A link in a chain looks and acts the same regardless of which way one travels the chain to get to it. From either direction, it is just as strong or weak. A web link, on the other hand, is different when seen from the A page or the B page. On the A page, a link is indicated by highlighting of text and/or by a change in the cursor. On the B page, there is nothing. Going from A to B is following the path of less resistance, going 'downstream,' if you will. Going from B to A, perhaps via a Back button, is travelling 'upstream' — not impossible certainly, but a very different proposition.

The directional nature of links is crucial to the rapid growth of the web. It means that anyone, regardless of who or where they are, can create an enduring link to a page which interests them, without the permission or even the knowledge of the page's author. For an author, making information available in this way comes with a one-time cost. Since there is no complementary link back to the A page, links to the author's page require no work on their part. The only effort is on the part of the person who wishes to link to it. This directionality was an inspired decision by the creators of the World Wide Web, one which enhances the flow of information via computers. It does, however, have a down side.

Loss of Context

The author of a web page need not know who is setting up a link to their page; it is also true that the author cannot know this. Consequently, the reader's entry path is unpredictable. There is no way to know what web page a reader just came from, or what is likely to be in their mind because of it. Writers of books have some assurance that attentive readers are following the sequence laid out for them in previous pages. Even if readers are less than cooperative, authors can comfort themselves with the knowledge that they have provided a sequence of which readers can avail themselves if they wish. Writers of web pages may also have provided such a sequence, but they have considerably less assurance that readers followed it or are even aware of it.

To be sure, book readers may be interrupted or may lose the thread of the author's argument. But with printed material the remedy is apparent. The thread exists and its location is known - it is on the previous pages. Web page readers come to a page from an unknown and unknowable previous page. As a result, web page authors cannot count on readers having seen, or being able to easily retrieve, preparatory material. Context built up over several pages — essential for allowing an author to make complex arguments or clarify subtle concepts — suddenly becomes unreliable. Directional links reduce an author's 'reach' to a single web page, turning writers of books or articles into writers of one-page flyers. Arguably, only subjects and subject treatments capable of being covered on a single web page are truly suited to the medium as it exists.

Providing preparatory material for readers to absorb and refer back to if necessary seems hardly worth the author's time, because some percentage of readers will not enter by way of the author's A page, but rather from someone else's A page, or from a page of search results. 'As stated in the last chapter' and similar helpful asides lose their effectiveness. The overall effect on readers is to deprive them of context. The effect on authors is to limit their options for helping readers master difficult concepts. Furthermore, since readers may arrive unprepared, authors must either write for a general audience which lacks any specialised knowledge, or risk irritating those who cannot handle (and did not expect) more advanced material.

Two current aspects of the web aggravate the problem of directional links:

• full-text searching; • the way browser programs such as Navigator or Explorer display pages.

Full-text searching

A full-text search produces a page of links to pages whose text contains one or more words specified by the reader. Frequently a search yields thousands of links, the majority of which are irrelevant. The ones which are relevant are just as likely to dump the reader into the middle of a sequence as at the beginning. Search results from a subject-indexed search engine (such as Yahoo) are much more likely to lead a reader to introductory material (often home pages), and give authors some opportunity to provide context. However, full-text search engines (Lycos, AltaVista, etc.) are less labour-intensive, more current and more common.

The palimpsest display

The developers of the first web browsers decided, probably correctly, that web page creators would want as much screen space as possible to display their information. To accommodate this desire, pages linked from (A pages) are completely replaced by pages linked to (B pages). The computer screen is used as a palimpsest - each page is erased by its successor. This delivery method has its drawbacks, however, which tend to enhance the problematic nature of directional links. Preparatory pages which the reader happens upon vanish as soon as they move on to another page. Any important information must be retained in memory. So too must the URL of the vanished page, if the reader has any desire or need to return to it. Browser programs assist the reader by providing a log of links to pages travelled, but this "backing up" process, typically via the Back button, is not entirely dependable, as pages disappear from memory unpredictably.

Both directional links and delivery via palimpsest threaten context, the former by making the previous page unknown, the latter by erasing the potentially helpful previous page. Any sense of history is at risk, with the only record of where you came from in your memory or your computer's memory. Neither one is as dependable as a sequence of pages bound together.

Loss of context is disorienting. When one is in search of information in an unfamiliar field, the web can feel like a carnival fun-house - you begin in a room with several closed doors (or links); you pick a door, enter the next room (or page), and the door you by which you came in promptly disappears, replaced by one or more new closed doors. Any sense of direction is quickly lost, as your perspective is limited to your immediate surroundings. Soon your only option is to just pick a door in the hope that the exit will find you. Similarly, searching the web can be disheartening - clicking on promising links in the forlorn hope that things will become clearer, until you either lose interest in the question or forget what you were looking for.

Suggestions to alleviate the effects

The directional nature of links is inherent in the web, but that fact should not prevent us from trying to lessen its deleterious effects. Full-text search engines and palimpsest delivery, on the other hand, are implementation decisions, which can and should be re-examined when they prove to be problematic. Equally capable of change are browser implementations of existing HTML tags.

The sequence, or 'previous page' problem can be handled through an implementation of the 'LINK REL=next' or 'REL=prev' tag. Currently these are little used, as most browsers ignore them. Assigning keys to link automatically to the logically next and previous pages (the right and left arrow keys are good possibilities) would encourage authors to use these tags, and permit readers who land in the middle of a sequence to follow it forward or backward with ease.

A second approach to the unknowable 'previous page' would be to modify browser programs so that any link to a subsidiary page in a website defaults to the site's main home page, or to the page specified with the 'LINK REL=start' tag. There would have to be an easy way to override this default (option key, right mouse button, etc.) and force the specific subsidiary page to come up, but by defaulting to an 'entry' page, links would normally lead to introductory and preparatory material, providing some orientation for readers. Such a modification would automatically affect search result pages, as well. In a sense, for those not seeking a specific page, such a change would make 'books' out of the single pages of the web.

An approach to the problem of palimpsest delivery would be to offer an easy way, now implemented in some browser programs, to open a linked page in a new window, and offset the window several pixels on the screen (down would allow the title to remain visible), thereby making it easier to recall pages the reader may wish to refer back to. The inclusion of frames (multiple pages displayed at once) in the HTML 4.0 recommendation opens up the possibility of using frames for this purpose as well and this is becoming a popular approach.

1.3. The Instant Link

Perhaps the greatest wonder of the Internet's data transfer method is that data from across the world is just as accessible as data from next door, and likely to arrive just as quickly. Since web links cause data transfer, as a web page is copied from its server to the user's computer, the same novelty applies. While the link is not exactly instant, it is largely insensitive to the distance between server and client (the user's browser). The time you have to wait for a page to be displayed varies, but distance plays a relatively minor role in this variation. While this dramatically increases a person's access to information (and access to audience), it also has an interesting side-effect - where once distance carried meaning, it can no longer do so.

Loss of a meaningful continuum

The reader clicking an a web link may think in terms of 'going' to the new page. But this is not like going to the store or into town. There is no sense of distance travelled. It is more like teleportation or instantaneous relocation. One minute you're here; for an indeterminate (though usually short) period you're nowhere; then you're there.

Over the years we have become accustomed to distance and it is frequently used to carry meaning. In a book, certain material is in 'Chapter 1' by design - it is the closest to the front cover and is correspondingly important. The relegation of information to the back of the book also carries meaning for readers. The first (closest) material is treated differently than the more distant; both authors and readers know this. General interest information tends to be first, followed by more specialised material. All this meaning vanishes as readers teleport from web page to web page - each new page seems equally important (or unimportant).

On the web there is no inherent spatial arrangement of pages upon which we, as authors or readers, can rely. All points are equidistant, equally available. be created by authors, and must then be communicated to readers. What is lost to teleportation must be either replaced or done without.

No boundaries

Space involves more than simply near and far; spatial organisation also involves boundaries, i.e. the division of space into areas. As well as distance, teleportation removes boundaries and this has implications for the organisation of information and its retrieval.

Material within a boundary is related. It often has a single author or a single subject. Furthermore, it has an structure distinct from the material outside the boundary. The boundary both encloses and informs - 'herein lies something different, something new'. In addition, the boundary often provides information about its contents. The common injunction notwithstanding, books are judged by their covers. A carefully designed cover draws in the intended audience, and discourages those who are likely to be disappointed by the contents. It communicates the subject, the author, the treatment, and the expected level of knowledge.

Instant links eliminate boundaries and thus eliminate another form of preparatory information. Without a 'cover,' authors lose a prime opportunity to warn readers of prerequisites, quantity and complexity of material, area covered (or not), language, and intended audience. Either authors must provide this information in some other way, or readers must go without.

No boundary negotiation

A lack of boundaries also affects the actions of readers. Where there are no boundaries, there is no need to negotiate boundaries. Putting down one book and beginning to read another is very different from clicking a link. Not only is more preparatory information involved in the former, more preparatory actions are involved. The ease with which one can 'teleport' around the web makes the process of moving from page to page less noticeable and since all pages are equally available, the process of moving from area to area, crossing 'boundaries', is equally unmarked.