CHI 96 Workshop: HCI and the Web, Position Papers
Austin Henderson
Apple Computer, Inc.
henderson@apple.com
I bring along some (but not huge amounts) of knowledge
about the web (I've built a number of web pages, and help design a site), a
good deal of design sense, and a lot of interest in the process of how
SIGCHI might play a role in helping us get beyond the "cool" to the "good".
1. Design
The designer of a web page knows the content. The designer of a
web client (browser) knows the medium of delivery. Both are needed to get
"good" design. HTML gives little control to the page designer, and lots of
control to the web client. PostScript (or RTF or ... which are being
proposed as languages for getting beyond HTML) gives lots of control to the
page designer, and no control to the designer of the web client. HTML is
too high level; PS too low. What is needed is not a blend of these two
languages which are towards the ends of the spectrum, but rather a language
somewhere intermediate between these that permits discussion of looks (and
as things become more interactive, feel too). Through this language, the
resultant look (etc.) can be created as a negotiation between the two
designers who have access to two pieces of information (the content and the
medium) necessary to determine the final outcome. Such a language has to
have a strong theory of "good" interaction underlying it. However, as we
seek this new design separation/collaboration, we should learn from the
experience of the past. Specifically, we've been here before with UIMS's.
What have we learned from the attempts to separate content and medium?
2. Description of interaction
It seems to me that we are, once again,
backing into the design of interaction. So we should learn from languages
which have been good for this in past, and stay away from those that are
bad. C and C++ are hardly good languages. So I hate to see us build them
into the next arena of design as we are doing with Java. The motivation, of
course, is the feeling that everybody understands C. While not even true of
programmers, it certainly is not true of users. Perhaps it is time to think
about who is designing these pages. Perhaps it is time to wrest control
from the programmers and begin to move it to the content experts. The
programmers should address the creation of environments that will make
designing "good" pages (looks and interaction) easy (or at least easier)
for content providers whoare not programmers. Again, an intermeidate
lanaguge rtesting on sound principles of interaction would help alot. C is
simply not such a language, not only because it is a terrible langauge as
programming langauges go, but mostly because it embodies no model of
interaction. What would be better? Not sure, but it would be fun for the
workshop to push on that - not so much to get an answer, but more to
understand the requirements (hopes, dreams) for such a lanaguge.
3. Processes
Suppose the workshop comes up with something that would help
the world. How will we make it have some effect? The big industrial efforts
will struggle to retain control over these descriptions so that they can
force everyone through their turnstyle. They will do this by moving faster
than everyone else. Which also means that the direction they move will not
necessarily be well thought out; indeed it may tend to force things to be
badly, but expediently defined. Deep changes (see above 2 points) are not
liekly to be considered; you can't stay ahead and be thoughtful too. So are
we inevitably going to get something terrible. Can we figure a way to break
the hegemony, and provide for the same kind of open-ended experimenting
that has characterized the internet so far? Or can we figure a way to
inject better ideas into the fast-track creation processes that are driving
the standards now? I would hate to see too much of the workshop spent on
process, but I think we should explore early on what people see as
approaches to having an effect with whatever we discover/invent. Indeed,
the route to influence might interact with the think invented, so possibly
it could be a framing notion.
CHI 96 Workshop: HCI and the Web, Position Papers