e need UIs that unite, instead of separate, all those modes. These interfaces are just arriving, as our Cover Story documents. At first glance, they look like today's GUIs, but under the covers, much is different.
We dub this new generation of interfaces network user interfaces (NUIs). However, the shift is also about what travels through the network: Computing is becoming more information-centric.
My contention is that until now, GUIs have not been primarily about information, but about applications. For the last decade, our approach to computing has largely been, "Open a certain application to enter or manipulate some information and then produce the appropriate output." Was the information text? Use a word processor. Numbers? Use a spreadsheet. GUIs have been the way we accessed those applications.
This was certainly an advance over the pr
ior UI, the command line, which in turn was an advance over its predecessor, a combination of punch cards and cables and toggle switches. Both those interfaces had a limited but important objective: Control the hardware. Think about the commands that DOS used: PRINT this, COPY that. The actual information being printed or copied was secondary.
Within the application-centric paradigm, software engineers began to craft a document-centric model. The application-control aspects of the interface would change as the user touched different kinds of content within one document. Some NUIs build on this notion, but the document-centric concept still misses the point. New interfaces need to accommodate information that isn't in documents and may be transient, such as live news or videoconferencing.
Sadly, the first generation of NUIs stops short of that goal. Some, for example, still open separate windows for separate functions, until your screen is cluttered with more windows than is useful.
The NUI w
on't end the proprietary data formats we use, but it creates pressure to use standard open ones such as HTML. So the first NUIs still won't enable a search that finds information whether it's in a proprietary .doc format, HTML, or e-mail-but they will create a platform on which to build that capability.
Nor will NUIs mean the end of discrete applications. To be really useful, NUIs will have to maintain links to legacy data. The radical net-computerists may want to start with a clean Web, but that's not likely to fly, anymore than halfhearted attempts to put a face-lift on today's cross-platform squirrel's nest. The NUI will succeed or fail based on its ability to make it easy to navigate our increasingly complex information landscape. That's a tall order, judging by the lengthy development cycles for both Microsoft Internet Explorer 4.0 and Netscape Communicator. But they and their ilk are an important start.
The ability to combine live content with active content that integrates data and interact
ive logic with our entire storehouse of data is probably the next step in the information age. It won't be the end of that path-someday we'll get the kind of machine intelligence and human-interface computing that places such as MIT's Media Lab are already developing-but it's a huge improvement over what we have now, even if the icons still look the same.
Mark Schlack, Editor in Chief,
mschlack@bix.com
Why Interfaces Matter
