HCI and the Web: A report on a CHI 96 Workshop

By Austin Henderson, henderson@apple.com

This is Austin's workshop report for Apple. It has been cleared for public view. It was incorporated into WebHCI by Keith Instone.

At CHI 96 (Vancouver, BC, Canada; April 14-18), a workshop was held on Human-Computer Interaction and the Web. The position papers for this are posted on the web. They cover a wide range of territory and raise many issues. There is also an e-mail discussion list (webhci@ACM.ORG), with an archive.

In the workshop, we reviewed the issues and added some more (total 53) as they occurred. Then we clustered them, and chose 7 topics:

and worked them in detail. Finally, we agreed to continue the discussions on the e-mail listserv (webhci@ACM.ORG; send "subscribe webhci" in the body of an email to mailserv@acm.org) which is open to anyone. A detailed report will be available on the web by July 1, and will be published in the October issue of the SIGCHI Bulletin.

Here I will record issues of particular interest to me.

1. Why is the Web a success?

It is:

global, a single space, dynamically extentable, uniform access, simple access, adequatley rich for most information but not too rich, leverages existing information bases because they are easy to inject into the web space, leverages user models of earlier means of accessing information (e.g. ftp, news, gopher) - reach out and pull it to you, computationaly light on my machine, immedaitely accessable (no login, or registation), cross-platform without commitment to particular software, open, low (or hidden) administrative cost.

It should be noted that many of these properties are being threatened by the advancing technology. We should use this as a checklist when we are proposing new Web (or Internet) technology.

2. Models: What is the user's view of the web?

This latter is the inevitable result of the technolgists continually putting it out there because they can. Excitment, and chaos. And very much a moving target if we are trying to put an Apple stamp on it. Pretty hard to set guidelines for something this fluid, as there is little established practice, and that which there is is threatened by "advances" (e.g., consider the ways "frames" are breaking the history model, changing the meaning of "back").

3. Models: What is the web builder's model of the user?

We should consider all these users when we think about HCI and the web.

As more people create agents which access the web, the design of web information has to consider that reader of the information may well be computational not human.

4. Model: Even the most basic web notions are changing

5. Security

The genius of the web was that it provided a single appropriately simple generic model of a page into which much of the world's existing information could be mapped with simple code (mathematical "injection" of the space of existing information into the new generic information space). The same trick is being attempted with styles: the URL will provide a naming scheme, and styles will become public and shareable across the world. The same has not been done for security, but could be: Can we find an appropriately simple notion of an entity with access rights to pages and responsibilities, and means of reliably registering those rights, and ways of checking them, and accounting ($$) for their use. A single model whose pieces can be made public and sharable (named by URLs). Then map the access models of all the servers in the world into that model piecemeal.

More challenging is the point that security as we have used it in our computer systems to-date has tended to deal in a very limited way with access to information: read, write, change sort of thing. Reasonable security will have to deal withthe purposes to which the information is going to be put. I have a name, but depending on what use, it takes many forms. Or alternatively I have many names used for differnet purposes. Security will in the end have to deal with this. This issue will require much more serious attention than "computer" security has given to the control of access to computer information.

6. The web is a medium

It is, and will be, used for many different genres of communication (personal home pages, organization home site, newspaper, advertisement, web art; flat page, catalaogue, generated page, dynamic page, space, place, chat page; - to consider information pages alone). And most of the claims that can and will be made about the web will be true of only some of those genres. (Compare the early studies of e-mail which did not recognize this differentiation and had at best weak results.) We must recognize that genres are definitional, create/discover them, and make them central in our thinking about the web.

7. Maps of information

Everyone feels like using the web is like viewing the world through a pin hole. Maps of information, and browsers that provide them, or are built around them, are very important in providing some sense of where you are. Maps can be created by others and handed around. However, conventions for their meaning, particularly when you jump into the middle of web site will be very important. Conventions for finding the maps associated with a site, and for additing your own to such a collection , will become important mechanisms for allowing people to help others find their way around, or see the web in their terms.

8. Navigation

A big issue. Probably factorable into two parts - content and context. Content - see Maps (7 - above). Context: a matter of genre: how you navigate differs on what you are doing, how you think of space, place, information , apps. No profound thoughts, but an observation that "frames" are breaking the sequential history model of navigation in Netscape. Much railing against frames.

9. Print and the web

A strong observation that much web content is currently also used in other computer and print context. Many people wanted a single source which can feed all three of these presentations. In the end, again genre suggests that sometimes this will be possible, and often not. A hypertext document and a printed document are really quite different.

10. Applets

With the web being used as a set of applications (form access to servers, and now JAVA), the design space for the web is about to be opened up considerably. In fact, wide open. There is little established practice for how this should be done. And little time to establish it before the web becomes seriously polyglot; interaction chaos. The standard rules about feedback, etc (e.g. Apple's HI Guildelines) should be expressed in generic terms for the web, in hopes that folk will apply them to whatever forms are invented in the next few years while the invention is still wide open (think land-rush). By 2006, it will have settled down to small tweaks as the desktop has settled now. But we're not going to have the luxury of seeing good stuff, and codifying it before most of the world is in motion - it is in motion now.

instone@acm.org / 96-06-18