Hypertext. It’s a term that causes eyes to glaze over and heads to nod dumbly. Most people have heard the term, coined in 1965 by Ted Nelson, but few who haven’t used it could define it. And there’s the question if the term is at all accurate any more, or if hypertext applications that are not, shall I say, data-format challenged (that is, they work with multiple data types), should be termed hypermedia.
Academic and semantic quibbling aside, the basic idea behind hypertext is non-linear text, or more commonly, chunks of text linked in numerous ways (feel free, as I said, to substitute graphics or sounds or video for "text"). The widening range of the hypertext field came clear at last week’s fifth annual Hypertext ’93 conference, put on by the Association for Computing Machinery. Topics ranged from hypertext help systems to hypertext fiction to massive corporate infobases and all the way up to something termed a "massively-parallel, immense-scale, widely-distributed, international digital library." These people don’t think small.
Perhaps the most interesting meta-conversation of the conference came when I was hanging out in the lobby area waiting for a session. Several of the conference organizers were talking, and mentioned that the courses, for which participants paid fairly big bucks, were crammed. In contrast, the "pure" research parts of the conference had poorer registration levels. Because of this, future conferences will probably better mix the courses and the research presentations to encourage users to stick around after the courses (this year the courses were all held Sunday and Monday, and a significant number of attendees left after Monday’s sessions). So it goes.
Designing Electronic Publications — The first interesting course was "Designing Electronic Publications: How We Do It," given by Paul Kahn and Krzysztof Lenk of Dynamic Designs. Kahn and Lenk discussed the evolution of visual methods of presenting information in context to how these methods are used in today’s graphical environments, concentrating of course on hypertext systems. They posited that everything is communicated visually by some combination of symbol and representation; that is, you must have a symbol, and for the communication to be successful, that symbol must represent something in the real world, or at least an abstraction the user can mentally grasp. With that basis, they went on to discuss methods of dealing with limited space – despite 21" monitors, the computer screen is always a window to a larger world. Think of it as the tyranny of the desktop. Like so many things, however, it turns out that the methods we use to present information in that limited space reflect traditional methods handling the same problems in art. Such methods include using multiple points of view (SimEarth, or any multiple window environment with updates in each window), raising the physical point of view to look down on a larger area (SimCity), using relative size of element to indicate relative importance (think of the different sizes of icons available in FirstClass), and merely the brute force method of cramming information together (compare less-spacious art museums of old with a thumbnail view in Aldus Fetch).
Interestingly, techniques like the vanishing point perspective common in the Western tradition are often ineffective in the electronic environment. In contrast, Chinese perspective is generally flat, and although perhaps less realistic, better conveys the same amount of information in limited space. Compare SimCity with A-Train for an example of this.
One of the most intriguing points Kahn made is that the concept of multiple windows is by no means new, and has been used in the art of various cultures for hundreds of years. With that in mind, the current legal wrangling over who owns what sort of graphical look and feel seems even more stupid than normal. The parallel with art goes a long way, even as far as using the image of a hand as a spatial indicator. The hand cursor in HyperCard was probably not coincidental.
Kahn also focused in on the use of text on screen. Although many paper-based designs and concepts translate badly to the electronic environment, typographical rules about white space, line length, and text color (the overall blackness of a text chunk) still apply. Kahn found that relatively short (60 characters or so, no more than 80, just as on paper) lines work best, with plenty of white space on either side of the text and with slightly larger than standard leading, say 12 point text on 17 point leading. Being unable to control those variables can hurt the visual display of on-screen text, and this might contribute to the impression that electronic text is somehow less professional than printed text.
After looking at text, Kahn discussed the role of color and icons in interface. Colors are often overdone, and he said that he usually stuck to no more than two colors, or four when dealing with the inherent colors in Windows, for instance. Colors have different psychological weights, so using lots of different colors constantly distracts the eye. That’s undoubtedly one reason Apple chose to use subtle spot color in the Macintosh interface, rather than the garish full-color look of Windows. Speaking of Windows, Kahn went on to condemn the overuse of 3-D in icons and controls. Careful use can enhance the interface by visually distinguishing elements, but after a point, which Microsoft hit about a year ago, 3-D controls merely confuse even further. These points apply especially to icons, which are also often overused in today’s graphical interfaces since it’s difficult to create icons that novices will understand quickly and that experts will be able to use effectively every day. A friend once proposed writing an article (that he’s never written) entitled "The Icon as Haiku." Given the proliferation of utterly incomprehensible toolbars (what do these people think a menu is for?), I’d love to see more thought on when icons are appropriate and ways of creating more useful ones.
Turning finally to hypertext presentation, Kahn looked at various different ways of portraying the space – global maps, local maps, or hierarchical trees. None of these methods are entirely satisfactory, and most current hypertext systems use a combination of them. The manner of specifying links, either by color, style, or font also seems not entirely satisfactory, and although he wouldn’t commit to it, Kahn seemed to prefer the use of background color to indicate links. When looking at the overall presentation of the onscreen information, he reiterated the point I made at the beginning of this issue – that paper presentation and electronic presentation are two completely different beasts, and must be treated as such.
I obviously cannot hope to completely represent the three-plus hours of the talk, but for those who design electronic interfaces, and for those who, like me, merely use them, these points are well worth considering at length before foisting an ugly and useless interface on the world.