Report on the "HCI and the Web" Workshop at CHI 96

To appear in the SIGCHI Bulletin, October, 1996.

The "HCI and the Web" workshop was held on April 14, 1996 at CHI 96 in Vancouver, British Columbia, Canada. The organizers, Steven Pemberton and Keith Instone, had several goals in mind when setting up this workshop, including:

identifying areas of the World Wide Web where the HCI community can contribute to the development of the Web from a user's perspective,
initiating a plan for the HCI community to become more involved in the technical aspects of the Web, and
promoting "good practice" on the Web, such as usable designs.

Although one day was hardly enough time to pursue all of the topics in as much detail as we wanted, we did accomplish a lot and set forth very reasonable plans of action. One outcome was a forum on the Web to continue the momentum of the workshop: see http://www.acm.org/sigchi/webhci/ for additional notes from this workshop, including the position papers, as well as information about other aspects of the Web that are of interest to HCI professionals. Be assured that anything mentioned in this report (such as the Missing Link symposium mentioned below) will have information and links about it on the WebHCI page.

Position Papers

Before the conference, all potential participants had to submit a position paper. The organizers ended up accepting all submissions (18), which gave us a large crowd, but fortunately not too large.

The position papers were submitted in various formats, but were all formatted in HTML and made available to participants on the Web to read before the workshop. Of course, everyone simply printed out the papers and read them on the flight in, but this gave us a common Web experience to discuss during the workshop.

The contents of the position papers were the basis for many aspects of the discussions, with several people citing other papers during the workshop (so, yes, the participants really did read the other position papers). The "HCI and the Web" position papers were also made available to the participants of "The Missing Link" Symposium (held at The Open University, UK, in early May) and several of the Missing Link position papers cite the workshop papers. So, the position papers have provided a good basis of "common ground" that we and others have been able to build upon.

Here are some quotes from the position papers to whet your appetite:

"If the entire world switches to Web pages and Java for its user interfaces tomorrow, we will set user interface design back at least five years." -- Robin Jeffries
"The World Wide Web is an interesting bundle of contradictions when it comes to questions of usability, and maybe we should regard this as a clue to the fact that, what usability there is, is more accidental than deliberate." -- David Gilmore
"1996: I think that web-surfing is dead." -- Jakob Nielsen
"It seems to me that we are, once again, backing into the design of interaction." -- Austin Henderson

Identifying the Issues

The workshop started with some quick introductions by each person. Since we had a large number of people, and since all had read the position papers already, we did not give each person time to make their own presentations. Instead, we jumped right into going around the room and listing HCI issues of the World Wide Web. Each person mentioned one issue, it was written on a yellow sticky-note, and then the next person got a turn. We continued around and around until nobody had any more issues. During the following coffee break, the organizers grouped the ideas into rough categories to help synthesize all of the issues into some semi-coherent structure.

Ten groups of issues evolved:

High level: These issues all dealt with the Web at a very high level and included: the user's model of the Web, genres, defining cool vs. good, and understanding why the Web has been so successful. (As a person rooted in the technical side of the Web, the author found many of these topics enlightening, forcing him to think differently, and making his head hurt.)
Organizational: These issues dealt with what SIGCHI as an organization and the HCI community in general can do to improve the usability of the Web. One issue was having a "SIGCHI seal of quality" for sites that achieve the highest standards of interaction and usability. Another issue was how to get SIGCHI involved in the standards processes, set mainly by the World Wide Web Consortium.
Accessibility: These issues dealt with ensuring that the Web is accessible to all types of users: physically handicapped, "technologically disadvantaged", non-English-speakers, and so on.
Information: This category contained issues pertaining the information that makes up the Web, with issues including information maps to help users visualize the structure of the information, meta-data, and how to promote efficient resource discovery.
Authoring: Many users are also authors on the Web, so issues dealing with the usability of authoring environments and collaborative authoring were common. Two important issues were: the cross-disciplinary nature of education and tools that are needed, since Web development involves many different fields; and, the need for the tools to embody guidelines for quality design, making the "path of least resistance" lead to good designs and making it hard (but possible) for authors to create poor designs.
Design: The design of Web pages and sites was a separate category by itself. Web design involves graphic design, interaction design and information design, all activities HCI professionals have been involved with for years. One issue that stood out as being especially important for the Web is the problem of having the print and online worlds existing together in harmony, with the problem of designing for both simultaneously being particularly thorny.
Applets: With the arrival of applets to provide interaction to Web pages comes some of the standard programming problems of code reuse and user interface design. Also, with this added functionality, will the Web become the entire computing experience for users?
Browser: With the browser acting as most users' gateway to the Web, there is no wonder many issues were concerned with the usability of this software. History, bookmarks and printing were just some of the browser features that were mentioned. Also, three aspects of navigation were included here: improving tools based on existing navigation research; future research into navigation that needs to be done to understand how people are using the Web; and how to separate the presentation aspects from the navigation aspects.
Technical: A lot of technical issues were lumped together: HTML, VRML, URLs, frames, external links, client side image maps, cookie technology, for example. Although these are low-level concerns, any decisions made at these levels will manifest themselves somehow in the user interface, so people with an understanding of HCI issues should be a part of the group working on the specifications of these technical areas.
Miscellaneous: There is always a catch-all category. Regulation, commerce, copyright and volatility were some of the issues that did not get placed elsewhere.

Some Issues in Detail

After the issues were grouped, participants voted on the topics that they wanted to discuss further. A summary of each of these topics is included in this section. The discussion of the final topic, Organizational Issues, actually served as the wrap-up for the workshop and has the next section devoted to it.

Why has the Web been so successful?

In order to understand the HCI issues of the Web, one has to understand the nature of the Web itself. Certainly the key to understanding the Web is to see why it has been so tremendously successful. Some of the reasons for its success are:

The URL provides geographic universality and a single conceptual space
It has a relatively simple user interface, hypertext
It leverages existing documents on the Internet
The protocol is "communicationally light" and easy to implement
It is open on many fronts
The standards make it fairly interoperable and it is very cross-platform
It is built upon free software
It provides immediate gratification, access and retraction
It is relatively easy to become an author
Human intermediaries can provide "intelligent information filtering"
There is plenty of content
The Internet and corporate intranets are mature enough to handle it

We tried to keep in mind the nature of the Web when discussing the other issues in detail.

What are the models of the Web?

What mental models are users forming of the World Wide Web? The simple, one-static-page-per-screen provides a simple model that people can understand fairly easily. But with forms for input, dynamic pages, frames and increased interaction, how do users see the Web? Currently, the language used on the Web (such as "Welcome to our site") implies a model of visiting other places, when actually what happens is that users are downloading information to their own computer.

User are also taking on a variety of roles on the Web:

Information consumer
Application user
Information provider
Information curator
Information manager
Entertainee

And, to reverse the picture, the Web is taking on many roles itself:

Entertainer
Application data locator
Information delivery
Collaboration tool
Document delivery

All of these have a profound impact on the mental picture users build of the Web.

We also discussed the genres that are emerging on the Web and what impact that has on users models. The most visible genre that has emerged is the "Personal Web page," which users are utilizing to do "social navigation". To find out about a certain topic, people are going to the personal page of someone who is also interested in that topic, expecting to find links they themselves will enjoy. To support this type of navigation, aspects of the Web could identify personal pages as special objects ("genre" meta-information, perhaps), allowing people to restrict a search to only them, for example.

How can information maps help on the Web?

The basic problem of helping users identify "where they are" and "what information is here" was discussed in detail. Information maps/overview diagrams are one solution.

Human-constructed maps were considered better than automatically generated maps. But the key is giving content providers the tools to make the maps. It is simply too difficult to make good information maps today, however: more structure needs to be added to the Web so that the generation of these maps can be more automated.

What navigation research needs to be done in light of the Web?

Navigation on the Web is somewhat different than the navigation studied in other hypermedia systems. As a starting point for identifying aspects of the Web that need further investigation, we pointed out five types of navigation which exist in the Web:

Browser navigation: back, history, open URL, etc.
Content navigation: hypertext links, buttons inserted in the page, etc.
Within-page navigation: scrolling, mainly, but with VRML and applets, this will begin to vary more. Especially interesting is the switch from 2D to 3D with VRML.
Multi-browser navigation: using several browsers simultaneously.
Application navigation: going from a word processor to a browser, and vice versa, for example. This could be a standard "desktop" application or one downloaded from the Web.

As Web functionality finds its way into more and more applications, and eventually, operating systems, then inter-application navigation will need to be researched.

How can the print and online worlds exist in harmony?

The print and online worlds need to co-exist on several levels:

Online information gets printed, so useful paper must be able to be generated from Web pages. Additional features, such as the grouping of pages into "books", could be added to HTML and browsers to produce better paper output.
Information mangers need to be able to maintain their content for both worlds simultaneously. The tools needs to be improved to make it easier to manage documents destined for both paper and online at the same time.
The differences in designing for paper and designing for the Web need to be clarified. Authors trained in paper design are now placing the majority of the information on the Web. But as the Web becomes more interactive, the skills of user interface design will become more important.

How will increased interaction change the Web?

Everyone felt Java, applets and the increased interaction that is coming in the future will mark significant changes in the Web user interface, with the pessimistic participants predicting a shattering of the user interface, making the Web much harder to use.

One of the major stumbling blocks we saw was the need for an interaction language that operates at a much higher level than Java. Before that can be accomplished, however, we need a model of this interaction.

A second important usability problem with applets is defining where they "exist" with respect to the rest of the user interface: are they part of a page in a browser, a separate application, or more invisible to the user? Also, do we really want "Internet applications" to look just like current "desktop applications"? Perhaps Internet applications should look different so that users build different models for them.

Results and Plan of Action

The final topic of discussion ended the workshop and was about the things workshop participants could do (often through SIGCHI) to improve the usability of the Web. We were conscious to keep the goals very specific and attainable, instead of aiming for the sky, sure to fall short.

The first "action item" was to gather feedback from other CHI 96 attendees. The technical program was used in two ways to let attendees know what went on at the workshop and to get their reaction. One was a poster and the other was a 5-minute presentation during the "Pre-Conference Event Session." Both were very useful in two ways: the author was forced to summarize his notes from the workshop to prepare for the presentations, and he received feedback from other attendees who were interested in the topic but could not attend the workshop. This was the author's first workshop, but he cannot imagine not doing a poster and presentation in order to wrap-up the workshop and integrate it into the rest of the conference.

The second action item was to create a forum to continue the discussion and gathering of HCI issues of the Web. During the conference, the author set up a SIGCHI Web directory ("webhci"), created an initial page about the workshop, and had an ACM mailing list created. Thus, before people went home from CHI, there was a place on the Web to go for more information, and each workshop participant had already received a note from the mailing list.

The third initiative from the workshop was to identify two ways to promote good Web designs. The first way is to encourage Web sites to apply for interactions design awards, with perhaps even the formation of a special category of awards for the Web. Austin Henderson (henderson@apple.com) is taking charge of this action item, so he should be contacted for more information. The second way to promote good designs will be to collect good design examples and publish this very selective list on the Web. A committee is being formed to judge the sites. Diana Gromala (gromala@u.washington.edu) is the contact for this action item.

The fourth and final outcome of the workshop were several ideas on how to promote the HCI community's involvement in the development of the World Wide Web. The Web is a unique system where individuals can get involved and make a difference. The most significant suggestion from the participants to achieve this was to have SIGCHI join the World Wide Web Consortium (W3C), so that SIGCHI members could participate (along side the vendors) in the various W3C working groups and standards activities.

The HCI community is already significantly more involved in the Web today than it was before the workshop. Certainly, the "HCI and the Web" workshop has only played a small role in generating this increased involvement. From the discussions at CHI 96, and the discussions hence about CHI 97, the author is very confident that "the Web" will be on many people's tongues in Atlanta at the next conference. Be sure to check up on the WebHCI page in the meanwhile to keep in touch with what is happening with regard to human-computer interaction and the World Wide Web.

About the author

Keith Instone is a research associate in the Computer Science Department at Bowling Green State University.

His days are now filled with making the Web a more usable place.

instone@acm.org / 96-06-05