|
a jaundiced eye: saner heads
for wednesday, april 2, 1997. A Friend of Paradox and Conservatism
I just spent several hours browsing through the ramblings of a very strange individual indeed, Jakob Nielsen. Mr. Nielsen is Distinguished Engineer for Strategic Technology at SunSoft, as well as the author of several books on user interface design. The problem is this - he is a visionary, but he is burdened with a few seriously misguided ideas about hypertext - and he can't seem to keep his ideas consistent from one month to the next, or even utilize the ideas he espouses. It must be strange to be able to see so far into the future but be so incapable of breaking out of old habits, some of which never really applied to the new media in question.
You can't really fault the guy, after all, he has been doing this stuff for much longer than many of us, but it has tainted his perspective. Here, let me give you a few examples.
First of all, hypertext historically has been obsessed with nodes, or lexias. In order to map out a traditional hypertext experience, the author has had to rely on the ability to graph these nodes and the possible interactions and relationships between them. There are many different methods for presenting the hypertext to the user, which traditionally fall into two categories: the card model and the scroller/popup combination. In the card model (e.g., HyperCard) the user navigates through the system by choosing the next step. The choice is presented in the main window, or card, replacing the previous screen. Navigation can be as simple as providing next and previous buttons. The scroller/popup model uses scrolling windows and popup windows to deliver a slightly more flexible experience for the user. A good example of this is Windows Help, with its inline definition popups. Of course, the Web doesn't fit easily into either of these models, being a combination of dozens of different navigation mechanisms. And, of course, popups are noticeably absent, unless you can call JavaScript windows "popups".
This would all be well and good, but the point is that the traditional hypertext author, being oversensitive to the problems of constructing the nodes and providing navigation between them, forgets that the Web is a constantly evolving entity - as opposed to a final product like a HyperCard stack or WinHelp file. People want to be able to find stuff by looking in areas which seem right. They don't want to have to remember the URL, which is a node-bound concept.
To be fair, Mr. Nielsen has recently come around to the idea that the site (as opposed to the page, or node) should be used as a "fundamental structuring unit". The problem is, he then goes on to discuss how such a structure should be hierarchical. My response to Mr. Nielsen: the experience may be structured hierarchically, but don't recommend that the webmaster actually build the filesystem into a hierarchy. Use the capabilities of the server (just as you might the capabilities of the HyperCard or WinHelp authoring/compiling environments) to provide multiple logical views of a single physical filesystem. Our nodes are still bound to the filesystem - let's take advantage of directories to organize our information. You can call this using sites or sub-sites (which I usually just call "areas") or whatever you want - but try to keep your eye on the fundamental characteristics of the authoring and testing as well as the viewing environments.
Another nitpick I have with Nielsen is his insistence on using the hideous blue and purple link colors which have been standard browser issue since Mosaic. It might be fine for Sun, since they seem to go with Sun's corporate colors, but it is a mistake that should have been rectified a very long time ago. If you are incapable of learning how to recognize the hand icon (which signifies a link), then don't turn off underlining in the browser. I personally haven't used underlining for years, and figure that if the author wanted me to be overwhelmed with desire to click on something s/he would have signalled a link's existence to me in other ways. Nielsen's defense, that "users have grown accustomed to scanning pages for blue text that they can click on", is feeble at best - the Web has only been around for a few years, and new users are coming on all the time who may never experience them. Better to nip this bad habit in the bud before even more folks come online. I'm reminded of the Dennis Ritchie tale of why some operators in C have the wrong precedence:
Finally, I get the feeling from reading his articles on web interface design that he sees the user as a pathetic amnesiac - incapable of remembering from one second to the next which interface elements do what. I prefer to think that any interface which is sufficiently used needs to be able to mutate into a less hint-ridden state. No user interface is intuitive - it would be hard to imagine how they could be, since so much of the work done on computers is relatively new to us as animals. How could we have had the time to evolve an inherent understanding of the magnifying glass as search icon? Of course, the other problem with interface design is that nobody gets killed if they click on the wrong thing, and there is no punishment for those who develop mistaken understandings of cause and effect. :^)
All in all, though, I find myself in agreement with him on many issues. He has successfully predicted several high-growth industries, such as push technology, the abandonment of the browser, and enhanced multimedia. But every time I start to get a warm fuzzy feeling, he just goes ahead and infuriates me.
Jakob Nielsen, ladies and gentlemen, ambivalence generator. (applause)
© 1997-2001
Steven Champeon. All rights reserved. |