Date: Tue, 10 Dec 1996 21:39:28 GMT Server: NCSA/1.4.2 Content-type: text/html Last-modified: Thu, 08 Jun 1995 17:02:11 GMT Content-length: 11288
The World Wide Web provides addressing and resource naming conventions that permit several kinds of computer resources to be shared globally. These resources include:
Access to these resources is available through application software known as Web viewers or browsers. There is no inherent, centralized organization to the Web. As a result, Web indexes and search mechanisms have been developed. Users are free to build their own hypertext index or save pointers to Web pages in lists (e.g. Mosaic's Hot List or Netscape's Bookmark feature). These lists are controlled by menus included in the browser's interface.
The NSF has designated the integration of Web browsers into the client workstation desktop as a research priority for the World Wide Web. This integration would eliminate the lack of symmetry between a user's local files, documents and services, and those available on the World Wide Web. If done right, we believe, it would assist the user in the organization and access of this content so that items found once could be easily found again.
The NSF's research priorities don't explain in detail the phrase, "integrate into the desktop". For this project we will assume that "desktop" refers to user interface software such as:
These user interfaces provide the following services, to varying degress:
In today's "desktop" interfaces there are different ways of organizing and displaying local and Web resources:
Before discussing what exactly is meant by "integration of a browser into a desktop," let's list some of the problems casual users seem to have using the Web. In each case, let's also ask if this problem might appear with any distributed, decentralized collection of resources.
We first noticed several of these problems watching a user navigate the Web while trying to build a set of Web pages to serve as a topic tutorial.
To focus on the ways in which a user retains and organizes access to
Web documents, we studied a novice Web user in a
In the next section we'll try to explain how this would work.
Comments and suggestions are welcome. Later on we'll have more to say regarding the desktop metaphor.
The following ideas guided our design:
We've included a description and drawings to tour our design.
We invited out test subject to play with our paper prototype. Our observations and her suggestions are described in this section.
In one sense, integrating Web access into the computer desktop really means integrating Web "objects" into other applications on the desktop. The desktop serves to organize and launch the computer's resources in support of some activity. Many common task can be initiated by direct manipulation techniques that move a document onto icons representing available services (like "Printer" or "Mailbox").
Newer platforms are capable of supporting 3D interfaces (see VRML and Hot Java). This work could be implemented in a 3D interface, but would not take advantage of perspective, eye position and viewing direction. Several 3D hierachy views could be used to visualize the user's organization of Web documents. 3D visualization may be more useful for navigation and exploration of the Web than organization.
We haven't tried to apply a novel, reality-based metaphor. That may be either a curse or a blessing. Instead we've decided to extend the popular "desktop" metaphor. We feel that this metaphor, in several of its implementations, has assumed a "form of life" of its own almost separate from physical desktops. Computer users today for the most part are familiar with some desktop GUI. The use of folders to organize directory trees has proven useful in desktops, even though physical file folders are not often hierarchies.
We have envisioned an extension to these metaphors that attempts to make network resources appear similar to local resources. This has already been done to some extent with prporietary networks and research prototypes, but not specifically with the Web.
What we've envisoned is consistent with standalone computer desktops. We need however to address network delay and reliability in accessing network resources. In what way would a user accept asynchronous activities (for example, would it make sense to delegate something to an "agent" or dispatch it, and ask for an interruption upon completion?).
This approach retains the key problem interfacing with a large or deep hierarchy. Items deep in the hierarchy are hard to see. A "Find File" or "Search" tool is usually provided to help users find what they have buried.
In class we mentioned several metaphors for Web access that differ greatly from the desktop metaphor: The Web as an ocean, the Web as a highway, for example. We discussed these metaphors and decided that they have more use as metaphors for navigation and browsing (e.g. "fishing", "trawling", "wandering", "road trip"). When applied to organization problem, they don't appear to offer assistance. In real life we cannot arrange the ocean or the interstate highway system to suit our own needs.
We have discussed several extensions to this approach:
Our approach could suggest modifications to the Web information structure and HTML. Authors and users often collect links into a page that acts as a directory or folder. Often these pages have no content of their own, or may contain personal annotations. What if HTML were extended to incorporate directories or folders (instead of logical types like ordered and unordered lists)? Browsing software and viewers could then support presentation metaphors suited more for organiztion than for reading. As an extension it would neither force authors to use it nor viewing software to present it differently from a document.
We feel that Web viewing software can be easily enhanced to support dynamic, personalized organization. Leveraging existing desktop software is a natural way to do this.