Magazine

SPECIAL FOCUS: WEB USABILITY

Web Usability and Technology

Usability of the Technology by the Designers for the People?

Usability implies purpose and audience. Part of the difficulty of defining "Web Usability" is the diversity of purposes and audiences within the Internet community. The original audiences were heavily weighted in favor of academics with high levels of computer savvy. Early users were fault tolerant of unsophisticated interface design, satisfied with an absence of pictures, and happy in a world of keystrokes that could flow between the Internet and simple text editors.

An Early Form Of Usability: "Useful"

This original community of digerati understood information to be meaningful content, not entertainment in the mode of the visual arts. Content was understood to be information, not entertainment in the mode of the visual arts. And, the Web was not viewed as a comprehensive replacement for the card catalogue in the Library of Congress with a full text search.

In short, there was more concern about shared knowledge than polished style and complex interaction (read: zany images and novel navigation controls).

Now we debate tradeoffs--snazzy and semi-usable vs. boring and functional. In current design polemics, this tension seems to exist.

Style Sheets, et. al.

As often happens, the proposed solution to a design quandary is more sophisticated technology. In this case--dynamic HTML--an additional layer of complexity for web designers--the Holy Grail of adaptive interface design. Does this mean that "Look and Feel" are to be negotiated by some effervescent interaction between authors and readers?

Perhaps. The difficulty, of course, is to make the design complexity invisible--removing cognitive load from the user. The challenge is to enable a user-selected format, to create choices that feel intuitive in any context.

Why take such a complex road? Two reasons: one is to provide format variance, the morphing of content as ubiquitous computing becomes a reality, (email on your cell phone, video messages in your computer); second to provide authentic audience preference based on the usual variation of the human population. Not all users are Gen-Xers who can read 10-point type. Not all Baby Boomers need 20-point type. Also, the "unusual variations" can be significant. Despite the relatively small percent of the population that has extremely low vision or no vision at all, the inclusive focus of the ADA legislation encourages universal access.

Usability, then, may be the clear separation of content from format, an inherent part of any design paradigm. Indeed, Mark-up Languages were designed to encode sentences, paragraphs, fonts, emphasis, and some definitions of data in tabular format not the aesthetics of a specific layout. As the standard for HTML 4.0 evolves, the designer, the technology, and the user can each specify aesthetics and modality at some level of detail.

The broad issues are reasonably clear. As often happens, the devil is in the details. But before jumping into details, the broader context of universal access seems appropriate to discuss. My stock analogy is compare information access by people who are blind to physical access by people who are wheel chair users. In the early part of this century, no architect gave much thought to physical access for people with mobility impairment. Despite the special efforts required to accommodate the physical challenges of Franklin Delano Roosevelt, the Senate remained inaccessible until Max Cleland, the first Senator in a wheel chair, was elected. The obvious functions and traditional aesthetics of stairs are so deeply imbedded within the traditions of architecture that the use of ramps is often seen as a detriment to good design. Indeed, a design devoid of any stairs would be quite dull.

Both common sense and the verbiage of the 1990 ADA legislation speak of "reasonable accommodation," an awareness that universal access is not total access. No one would expect that every front door be placed at the top of a gently sloped ramp, or that the controls for a car display Braille labels. Universal access means that an awareness of the needs of people with disabilities has affected the design. And that access is an integral aspect of a structure, not an obtrusive retrofit. Universal design implies that a sensibility of inclusion has informed an entire composition.

Are there really blind Web users?

For most readers of this article the Great Browser War is about Netscape vs. Microsoft’s Internet Explorer. Arguments reference "predatory marketing practices," dialects of Java and JavaScript, etc. But for some, the question is whether to use a graphical browser at all. Lynx, the most common text based browser, predates both dominant GUI clients. Its existence is based on the idea that the Internet is a provider of HyperText, not a televisualized world encapsulated in narrow bandwidth. For blind users whose computer configuration includes software and hardware to verbalize text, such a browser is simply one more application whose content is audible text. An alternative to Lynx and speech synthesis software is pwWebSpeak, a browser specifically designed for blind and low vision users, with speech synthesis bundled as in integral function. And, to the extent that images have a textual equivalent (alt="corporate logo"), these approaches work "reasonably well." The difficulty is that the "reasonableness" is a function of several variables. Part of the solution to universal access is more sophisticated software: translation of HTML conventions and text content into speech even if the browser is a graphical one. But part of the solution lies in the awareness of the architects of web sites. A clear illustration is the use of tables.

Tables: One Syntax for Three Purposes

Tables are an essential technique to represent information as a collection of rows of data. For example, a list of names and telephone numbers. This structure of textual information works well in text only browsers and can be easily verbalized correctly from the internal storage of GUI browsers. It is the natural mapping of data and structure.

The second common use of tables is a visual trick: the construction of columns [ see example], a variant of newspaper format. Despite HTML syntax for auxiliary tags to indicate that some tables represent columns, the linguistic rules are neither fully standardized nor widely supported. Consequently, stories that flow down the page and wrap into subsequent columns often do not sound linear. The assumption make by the software which provides speech is to read each "Cell" in a left to right progression. Although this approach is correct for the primary use of tables, unless each story is contained within a single cell, this technique is dysfunctional for blind readers. [ see example]

A third function frequently accomplished by table syntax is the construction of an index [ see example] of links as a separate column. This technique, if used as an innocent way to provide a vertical menu bar, can work quite well. If the table precedes the body of text within the HTML code, a text only browser will present these links first, as if they were at the top of the page. If there is an on page link to jump over the navigation table, a blind reader (using a text based browser such as Lynx) need not read the entire index on each page visit. However, if the blind reader is using a GUI browser with speech synthesis, reading horizontal lines on the screen will cause the same problem as other column table structures.

What design principles are at work here? The basic one is to separate content from presentation. The emerging HTML 4.0 standard makes a finer distinction: content (what people will read), structure (the hierarchy of a document defined by the HTML codes), and format (sophisticated appearance variables represented by Cascading Style Sheets).

In the above table examples, we have the classic challenge of defining cleanly the boundary between structure and format. Obviously, a matrix of data is a useful structure. Yet, HTML affords alternate structures to be created with the same syntax, and allows color tags (and other format parameters) to be embedded within the table definition. Clear distinctions are not created by syntax, but by designers who create an intelligible partition between media, structure, and appearance.

The use of this principle, separation of content, structure, and format, within the design phase of site construction has two advantages: it clarifies what is being presented; it opens up options for multiple representations of the site.

Based on a comprehensive understanding of a site, universal access requires one of two approaches: fully accessible structure and format (a building where all doors can accessed without steps); a graphical site and an alternate set of text only pages (a side entrance which is easy to reach in a wheel chair...).

The techniques for implementation depend on several issues: site construction software, use of cascading style sheets, nature of audience, evolution of the site over time... This becomes a classic design theorem: which approaches are better in evaluations by the user community.

How Do I Know If My Site Is OK?

One way is to try a sophisticated syntax check: "Bobby." A validator that checks syntax to see what you site does in several browser environments including Lynx. Like many syntax checkers, there a some high level of "pettiness."

http://www.cast.org/bobby/

Or to use W3C's validator which has "Levels of Pettiness"

http://validator.w3.org/

Note: Syntax checkers are often more concerned with minor variations of syntax that browsers are. That is, a page may do well, but have many non-standard features disregarded by browsers. Lynx is quite good at bypassing superfluous parameters. A complete evaluation of a complex site is subtle. Full access is the result of a deep desire to make what you create available to all, a sensibility of inclusion. And, some hard thought about which techniques to use.

Arthur R. Murphy (arthur.murphy@arch.gatech.edu) is a research scientist at the Center for Rehabilitation Technology at Georgia Tech. Current projects involve the development of interactive learning systems and the enhancement of a Web site containing disability information. He is a member of the Electronic and Information Technology Access Advisory Committee, a group of two dozen national experts who advise the US government on establishing accessibility standards for information used by government employees and for public documents used by all citizens.

Copyright © 1999 by Arthur R. Murphy. All Rights Reserved.