The Australian National University
Canberra, ACT 0200, Australia
For some time we've needed better models for developing the structure of an organisation's information on the Web, particularly for very large diverse sites, which, more and more are exemplified by the Web presence of government departments and universities. Books on the subject of good Web page design abound (and some are even good) and there are plenty of texts which outline strategies to structure Web sites (such as David Siegel's "Creating Killer Web Sites"). But few provide any direction for the coordination of very large distributed sites.
This is a problem. It is a problem because Web surfers expect organisations to have a "home page", and from this page, they expect to be able to gain access to all parts of the organisation that are online. It is a problem because the organisations which may well provide critical information universities and government are the very organisations that have a complex Web presence that tends to hide this information.
This paper aims to describe the special challenges facing the decentralised distributed organisation on the Web and develops a conceptual site design framework which can be used by people coordinating Web publishing activities for distributed organisations to develop a stronger organisational focus. The paper will also show how the concepts can be applied practically by identifying the issues that one decentralised distributed organisation the Australian National University has faced, and the way that the concepts outlined in this paper offered a solution.
We need to define some basic terms that will be used with consistent meaning in the context of this paper. The terms and their meanings are listed below. I do not mean to suggest that these are canonical definitions of the terms. Simply, the definitions are designed to give all readers a common terminology that is consistent with the paper's.
An organisation that places its information on the Web is establishing a Web presence for itself. A Web presence can be a single Web page, a series of interlinked Web pages (a Web site) or a series of Web sites.
Web sites are difficult to define, and yet it is a term that is critical for any discussion of organising information on the Web. In this paper, a Web site is a collection of resources (HTML files, graphics, sounds, Java classes, etc) that are grouped together by some conceptual relationship. This is a very arbitrary definition, and is meant to reflect the actual nature of most Web sites rather than an ideal.
An organisation which has a very controlled Web publishing strategy may have a very well defined Web site use of graphics, navigation and careful selection of URLs all working together to re-enforce a sense of coherence. Other Web sites are very differentiated, with little or no similarity between the HTML pages and little or no consistency. For the reader, the boundaries of such a Web site are almost invisible, and the actual definition of the Web site may only be known to the person who maintains the site.
Thus, within this paper, a Web site is a fairly nebulous entity which can be thought of as an idea that draws together a series of HTML pages and their resources under one broad subject.
A Web page is the most humble of all structures on the Web. The Web page is a single HTML source file as received by a client from a server. The Web page may contain extra content, like graphics, or complex content, like a Java applet. The source may have been generated on-the-fly on the server, but if it appears as a single "page" of HTML source when it arrives at the client, then it is considered a "Web page".
The Australian National University (ANU) has a relatively long history with the World Wide Web, hosting one of the first Web servers in the world. A short time later, as other Web-based resources began appearing around campus, the ANU developed a "home page" which was designed to give Web surfers quick access to the handful of information which was available. We could call this phase one of the ANU's Web site.
Phase two, which we could call "the age of the CWIS" (Campus Wide Information Service) began around 1995. By 1995, there was growing support for information such as e-mail addresses to be available on-line. The Web was beginning to grow in popularity, and its usefulness for distributing campus information services was identified as an important role, thus ensuring continued funding and support for Web publishing activities around campus. The concept of a "campus-wide information service" or CWIS was developed to provide structured access to Web-based administrative resources around the University, and the notion of a centralised Web site or "home page" as a front door for the University was formalised.
During this period, the number of Web sites around the campus grew exponentially. Basic courses in HTML were offered by different groups from around campus, and a centrally maintained Web server was developed not only to host the university's home page, but also to provide a publishing platform for authors staff and post-graduates students to put their information on. Some departments in the University took advantage of the server, while others preferred to establish their own Web servers.
The highly distributed, decentralised and autonomous nature of Web publishing at the University hardly reinforced a cohesive image for the University, which was of particular concern to the University administration. Interest in Web publishing standards grew, as it was hoped that by implementing some kind of standard look and feel across all University Web sites, some degree of corporate identity could be maintained.
Unfortunately, under the decentralised autonomous model which had developed, there was little chance of enforcing such standards, and there was some debate about the desirability of implementing a "corporate image" in the university environment. But the problem could not be left alone, either. As the Web became more popular in the community, so the audience that was being reached was growing and changing. The image of the University on the Web became an ever growing concern, even while the amount of information kept growing.
Most organisations are now reaching what might be called the third phase of Web site deployment. Where before it was enough to simply put a new kind of information on-line, user expectations now demand that information be useful and well presented. Of course, for information to be useful, it must necessarily be found, and so navigation has also become a key issue. There are thus three key concerns about an organisation's Web site:
The third phase also ushers in a new kind of audience for the Web site. The audience in 1995 was academics, computer junkies and the odd interloper from organisations which were lucky to have access to the Internet. Now, as Internet access is becoming more mainstream, the audience is becoming more diverse. For organisations such as universities, a very large percentage of the total number of Web surfers may potentially be your audience.
For the ANU, the problems were easy to spell out, but very difficult to solve:
The problems that were manifesting themselves in feedback were similar:
To begin devising solutions to these problems, it was necessary to think more generally about the Web presence of organisations, and specifically, about the Web presence of the ANU.
An organisation's Web presence is made up of one of more Web sites, which in turn, are made up of one or more Web pages. This is true for any organisation's Web presence, whether they are a very large organisation with many thousands of Web pages on-line, or a small organisations with only one Web page on-line.
There are four broad publishing models that all organisations that have developed a Web presence fit into. They are:
These models are arbitrary, but serve to define the different challenges faced by different organisations in making their information accessible.
These four models allow us to discern four Web presence models:
We will examine each of the four defined models separately.
A centralised restricted model is typically adopted by organisations who wish to present a very sleek and professional corporate Web presence. By centralising the maintenance of the Web pages, and implementing page design guidelines, the content and look and feel of the organisation's Web presence can be very strongly controlled.
The features of this model are:
Often, the centralised restricted model is employed by corporations that have the resources to fund a committed Web publishing team, and a strong desire for the site to reinforce the organisation's identity. Examples, therefore, are corporate and include BHP, Microsoft, IBM, Sun, Hewlett-Packard.
A decentralised restricted model has no central publishing agency or team, but instead encourages individuals across the organisation to develop their Web publishing skills and place information on the Web as needed. Under a decentralised restricted model, the kind of information, and its look and feel is well-defined. While there may be many authors across the organisation who are publishing information, they must publish the information in a standardised format. Web authoring training may be developed within the organisation with a bias towards software solutions which encourage or enforce use of a corporate style.
In a centralised autonomous model, an individual or small group of publishers are the only ones in an organisation who are able to publish information, but they work autonomously with few restrictions on the content or look of information which they are publishing. This model is typical for organisations (particularly IT organisations) who have decided to develop a Web presence but have not formalised tasks and responsibilities, and for organisations which have a Web presence due to a group of enthusiasts who have taken the job of Web publishing on themselves.
In a decentralised autonomous model, people from across the organisation can publish information on the Web without needing to conform to any organisation style. This leads to a highly fragmentary Web presence, often made up of multiple Web servers. This, in turn, makes it very difficult for people to find information, because there is no central place to start looking from.
Organisations which have independent branches or units will tend to develop a decentralised autonomous model of Web publishing, particularly if there is some kind of formal information technology expertise within the units. This is particularly the case for universities, which are naturally broken into sub-units (schools, faculties, centres) each of which often have some internal computing expertise. It is also the case with many government departments.
Most universities follow a decentralised autonomous model, partially because this model reflects the organisational structure of most universities, but also because universities have developed their Web sites in tandem with the growth of the Internet, and so have never started from a "clean slate".
Back in 1994, Dr. Christine Quinn of Stanford University identified one of the central issues in her paper From Grass Roots to Corporate Image The Maturation of the Web. In it she asked "How does a large, diversified, distributed organisation such as a University begin to collect [information] and provide a mechanism by which Web offerings could be presented in a similar way?" (Quinn, 1994). Once the four Web presence models had been developed, it was easy to see where the ANU fit, and it became possible to approach the question that Dr Quinn had asked.
The ANU has a decentralised autonomous Web presence. With over 160 Web servers, and probably twice as many Web sites, each maintained by one or more separate groups which had little or no contact and certainly no formal coordination, the ANU could almost be seen as the quintessential representative of the decentralised autonomous model.
Many of the problems which had been identified can be seen to stem directly from the decentralised and autonomous nature of the University's Web presence. The lack of central coordination of publishers meant that wheels were being reinvented everywhere. Navigation was difficult because everyone was providing their own list of links to internal resources, and naming them differently. The home page for the ANU is http://www.anu.edu.au/. It was sometimes incorrectly linked to as http://elisa.anu.edu.au/ which is the library's home page. This incorrect attribution of the ANU's home page was of concern because it shifted the focus from the general to the specific: the library would provide navigation that was biased towards library information, which would mean that some resources were very visible (the online library catalogue, for example) while others were lost completely (many administrative areas, for example).
Even though there was a central home page, there was some confusion by the people maintaining that page about the relationship between other sites and the organisations who maintained them. What is the difference, for example, between a faculty's official home page, and the home page of a faculty member who wants to provide an alternative?
By identifying the model that the ANU fit into, it was possible to start thinking about the options available to us.
The first possibility was the simplest but least satisfactory. Everything could be centralised, that is, we could move to a centralised restricted model. All sites could be moved onto one Web server, maintained by one central group, and all Web servers except that one turned off. A standard look and feel could be developed and enforced on all Web sites. Of course, this method would have not only been draconian, it would also have been politically impossible to implement. Also, it would not have been desirable, because it would have destroyed the very nature of the university Web site. Needless to say, this option was never entertained.
The second option was to distribute and enforce some kind of standard look and feel around campus, thus implementing a decentralised restricted model. Attempts to do this were met by a number of obstacles. First, there were many, many pre-existing pages, which people were unprepared to modify. Secondly, there were a variety of HTML skill levels involved. Some people were extremely good at hand-coding HTML, while other relied entirely on editors, and had very little idea how HTML worked. Some scholars simply didn't want to learn HTML, particularly if it meant they would have to increase work-loads by preparing on-line course notes. This often defeated the purpose of standardisation because the look and feel from one page to the next would change as skill levels and interest in learning HTML varied.
The third option has been much more successful. Rather than attempting to change the University's Web presence model, the idea was to develop the concept of the "home page" into something a little more flexible. Rather than attempting to design a single Web page which would provide Web surfers with access to the massive amount of information on the ANU's many Web sites, the trick was to develop a Web site which had one purpose only: to help people discover resources. This Web site, which incorporated a home page, we call a "home site" or "navigation site".
The basic principle of the home site is similar to that of the "home page" it is designed to help readers discover the information that the organisation has put on the Web. Where the home page acts as a central point of resource discovery for a Web site (i.e., a collection of Web pages), the home site acts as a central point of resource discovery for an organisation's Web presence (i.e., a collection of Web sites).
The home site not only provides a navigational and identity focus for the organisation, it also provides an effective locus for discussion, which tends to highlight information or services which could be centrally provided, but aren't. Good examples of central services include staff indexes, campus-Web site indexes, organisational indexes, and full text indexes (often supported by a Web crawler and search engine).
Much of the structure of the home site is well explained by the Yale C/AIM Style Guide (Lynch and Horton, 1997). The home site follows a basic paradigm that seems to apply equally well to all Web sites. All pages in the site are either content pages or navigation pages. A navigation page provides mainly "menus" of links short lists of links to other resources. A content page, on the other hand, primarily provides new information and does not contain lists of links (although it may well have hypertext links interspersed in the text of the page).
Because the point of the home site is to provide people with access to information around the campus, many of the links are designed to lead people away from the home site, and out into other sites which actually provide the information the Web surfer is looking for.
There are some parts of the home site, however, which provide information in its own right. This information is, as was noted above, centrally supported information such as:
As a site in its own right, it is vital that the home site maintain a common appearance within itself. This common appearance is designed to reinforce the identity of the organisation. At the ANU, this is achieved through the use of a standard design grid upon which all pages within the home site are laid out. The background is a crisp white on all Web pages, and the logo and identifying text (The Australian National University) are repeated in the same graphic format again and again, like a stamp adding identity to the site (see Fig. 1).
Each page in the site always provides a link back to the home page, a link to a comments form, a date last modified notification and a brief copyright notice. Server-side scripting is also used on all pages to support a text-only version of the Web pages, to aid in accessibility and to speed browsing.
Fig. 1. A typical navigation page.
The home page for the ANU is a single HTML file which uses server-side scripting to determine whether the Web surfer is looking at the page from outside the University or within. Once the reader's location has been determined, two different home pages are displayed.
Fig. 2. The ANU's external home page.
The first home page variation is for external clients (Fig. 2), which is designed to be attractive and visually appealing, and which provides instant access to the kind of information that external clients are most likely to want.
Fig. 3. The ANU's internal home pages
The second home page is for internal clients (Fig. 3), and provides a greater number of links up-front, to minimise the number of clicks that information is from the home page. The decision to make the home page for internal clients so link-rich was based on the feedback obtained through Webmaster e-mails over nearly two-years. Essentially, people who use the Web site a lot get very used to the links they access frequently, even in a large jumble of links. These people do not like having to click through many pages to get to a site they go to frequently.
Externally, however, it's a different matter. People want to be guided to the resource they are trying to discover. They don't mind clicking a few more links if they feel they are getting somewhere. Fewer links on the home page and greater link redundancy (that is, if you provide three buttons on your home page, the same link may appear under two of them) on navigation pages is the theme for the external home page. For the external home page, graphics are more heavily used to promote an image of the university. A great deal of work went into optimising the images on the home page to ensure they were as small and fast-loading as possible.
The whole home site uses server-side scripting to wrap the content of each page with the HTML code that provides the formatting information. Thus, when the look of the pages is due for an upgrade in the future, only the include files need be updated, allowing a complete change in design to be accomplished very fast and easily.
We need to stop thinking about large decentralised, distributed organisations like universities and government organisations as collections of static HTML pages. They are collections of Web sites, some of which are well executed, some of which are perpetually experimental, and many of which are both well executed and in a perpetual state of change.
Under such a decentralised and autonomous model, we also need to be aware of the limitations that are imposed by the model: we cannot implement a standard look and feel across all Web sites unless we change the organisation's publishing model, a step which may not, in the long-run, be a wise move.
We need to better understand the beast, and remember that the goal of the organisation's Web presence is multi-faceted. We need to identify our organisation's goals and taking the structure of the organisation's Web presence, develop solutions which allow Web surfers to see that we are meeting these goals.
For the ANU's Web presence, the goal is to make the resources and expertise of the University accessible to other universities and research institutions and to Australian governments, industry and the wider community. The home site helps us achieve these goals and makes it clearer to Web surfers how we are meeting these goals.
Lynch, P. and Horton, S., Yale C/AIM Web Style Guide, 1997,
Quinn, C., From Grass Roots to Corporate Image The Maturation of
the Web, 1994,
ANU Home Page (external) - http://www.anu.edu.au/external.html
ANU Home Page (internal) - http://www.anu.edu.au/internal.html
ANU Library - http://elisa.anu.edu.au/
BHP's Web Site - http://www.bhp.com.au/
Microsoft's Web Site - http://www.microsoft.com/
IBM's Web Site - http://www.ibm.com/
Sun Microsystems' Web Site - http://www.sun.com/
Hewlett-Packard's Web Site - http://www.hp.com/
Samuel M. Hinton (BA Professional Writing) came from a commercial information technology background to the Australian National University in 1995 to provide direction and support for the University's information technology services publishing activities. Most of his work involves providing direction and advice for the University's central Web publishing activities, and support of campus Web publishers through an on-campus group called "Spiders".
Sam is currently working towards a Masters in Communication at the University of Canberra. His research interests include Internet power relationships, organisational regulation and use of the Internet, and the potential of the Internet as a democratising force. His interest in writing this paper stems from his interest in organisational use of the Internet.
Further information can be found on his Internet home page at http://www.anu.edu.au/~e951611/.