The browser interface came into being in the 1990s solely as a way of letting users find information on the Internet. It evolved into a general-purpose interface, used as the front end for applications beyond Internet searching. Its characteristics include a standard bar, in which you insert a Web address, and pull-down menus with additional navigational tools; the applications appear within this framework.

Browser interfaces exist for traditional applications, such as word processing and spreadsheets; for specialized applications used within companies, such as inventory tracking; and for repetitive tasks, such as filling out forms. Because nearly all computer users today have Internet experience—and for some users it is their entire computing experience—making an interface browserlike ensures that it seems familiar to Internet users. Familiarity leads to quicker and more intuitive learning. Windows itself now uses a browser-style interface to navigate files. Business applications accessed through a local network rather than stored on individual computers—company manuals, time cards, and the like—often rely on a browser interface.

The big advantage of the browser interface is that most machines already have some kind of browser installed, so you don't need to have specialized software to run applications stored on a remote server. This is a major appeal of Web-based mail applications. Although they are less efficient than specialized e-mail software that is run locally, they enable a user to check mail from any Internet-connected computer. When you check your e-mail from a friend's computer, you are most likely using a browser interface to an e-mail application, even if you typically use a dedicated e-mail application from your home computer.

On the other hand, the browser interface's purpose is fairly simple and straightforward: to navigate the Internet and view information. It's not really designed to be an interface for running complex software applications like word processors and spreadsheets, and using it as one leads to too many mouse clicks or keystrokes, wasted time, user frustration, and errors. Generally, your active "state,"—that is, what you were doing in an application the last time you were there—can't be preserved. So you need to tell the computer repeatedly many things you think it already knows—like where you were when you left the application or recent changes that have been made to format or content.

A variety of tricks make the browser interface more useful. If you allow it, Internet applications may store cookies, or bits of information, on your computer to remind the applications of what they already know about you. Or you might use a plug-in device that has details about you and your applications and data, such as a Smart Card or some type of memory-containing key that plugs into your computer's USB port.

Lately, interface and application designers have been looking into ways of extending the browser interface to provide a richer graphical user interface, or GUI. The first to go public with such a product is IBM Corp., with Workplace, software it sells to companies that lets them run a variety of traditional business applications remotely over the Web, including customer contact managers, project planners, spreadsheets, word processors, and e-mail programs. This kind of interface is stored on a server and downloaded on demand to a user's desktop. It has application-specific menus for a wide variety of applications that are much more efficient than the general menus of a typical browser.

Navigating large collections of information is difficult and time-consuming. We can create these huge collections easily—from crawling the Web with a search engine to assembling mixed-media collections of documents, databases, audio, and media. But sorting through this information can be a nightmare. A second, more specialized interface addresses this problem.

Let's assume you have an interest in Procter & Gamble Co. The Web contains an enormous amount of information about this company—product information, case studies, newspaper articles, details of product recalls, videos of commercials, press releases, manufacturing process data, podcasts that mention the company, and blogs by employees and consumers. The number of individual items, as well as the sheer size of the database, is astronomical.

Traditional search engines can be flummoxed when attempting to sort through this data. Say, for instance, you are trying to find the source of a quote you recall reading or hearing about the company. Google will lead you to that quote easily only if it has been frequently referenced. A traditional search might work if you remember at least part of the quote exactly—for example, if you recalled the executive saying the word "terrible." But if you misremembered the word as "horrible" or "disgusting," traditional search engines would come up short.

Researchers have long tried to find new metaphors to solve this "needle in a haystack" problem and locate information quickly and easily.

One such product is Star Tree, developed by the Information Interfaces group at Xerox Corp. in Stamford, Conn., and a company it has created to sell its products, Inxight Software Inc., in Sunnyvale, Calif. Star Tree products take vast hierarchies of information and display them in ways that the human eye and mind can readily understand.

Taking a huge collection of information—our "everything on the Internet about P&G" example—Star Tree will arrange it into sets of subtopics, either user-selected or selected automatically: products, customers, external press reports, and so on. These groupings of Web hits are placed into subtopics that look like galaxies [see photo, "Galaxies of Information"]. The user moves the cursor around this universe, clicking on different galaxies or constellations within a particular galaxy.

Whichever area is clicked, Star Tree's universe reorders itself, making the item clicked upon larger and changing the size of the other items, depending on how relevant they are to the selected item. Eventually, you narrow in on one section of one galaxy, moving finally to individual pieces of information. Star Tree can also present this data in list form, for users who find the galaxy metaphor too exotic.

Also tackling this problem, IBM Corp. has developed an infrastructure called WebFountain, for connecting vast amounts of text in a variety of formats collected from the Web or other sources [see "A Fountain of Knowledge," IEEE Spectrum, January 2004]. It's similar to Star Tree in that the software takes huge amounts of information from various sources and automatically categorizes it. WebFountain does not include presentation software—that is, the graphical or text display that the user sees. Rather, it is intended as a foundation for such interfaces.

Factiva, in New York City, a business news and information aggregator service owned by Dow Jones Reuters Co., in Princeton, N.J., is one of the first companies to build upon WebFountain, presenting information to its clients through a multipanel interface of its own design.