The next big thing in user interfaces is unlikely to be a single, all-purpose interface, used for a host of different tasks. Instead, three interface categories are emerging. First is the browser interface, perhaps not the most efficient tool for a particular task, but with it users can move easily from one computer to another. Second is a special-purpose interface for navigating large collections of information such as the Web. And finally, there are a variety of interfaces that computer users are beginning to acquire from both established and new companies for managing their own collections of information.

Designing new user interfaces requires a tradeoff. You can either exploit the newest interface theories—using radically new metaphors, software, or devices—and risk alienating hordes of experienced users, or you can exploit the familiarity of your users with a huge, installed base of existing products. Doing both is difficult.

Microsoft Corp., in Redmond, Wash., faces this dilemma every time it releases a new version of Word, its popular word-processing program. Many Word users have already spent a lot of time with the product, memorizing the locations of tools on the menu bars and the shortcuts to various functions. When Microsoft rearranges the menus to make them more efficient, the company is greeted by howls of protest from those who intuitively click where "Bold" used to be, for example, and find they've indented a paragraph instead. Many people simply continue to use their old versions of the software, refusing to be inconvenienced in the short run by a new interface, even though it may ultimately be more efficient.

This interface familiarity also protects products from being usurped by competitors. OpenOffice.org, a free suite of office tools available for Microsoft Windows and Linux, does 75 percent of what Microsoft Office does—more than enough for the typical user. But the interface is organized completely differently; nothing is in the same place. And it turns out that many users would rather continue paying Microsoft for its familiar version than retrain on the free one.

What happens, therefore, isn't surprising. The owners of software with many entrenched users are reluctant to make big changes, and their interfaces tend to be refined incrementally and retain their familiar look and feel.

Microsoft's desktop interface will evolve somewhat with its next operating system, Microsoft Windows Vista (formerly Longhorn), and will also be offered on Windows XP, both in 2006. The company is making enhancements, such as using transparency to guide the user through the navigation process. For example, an application or process not used for a while would become more transparent, seeming to fade away, although still visible. This lets the user focus on the task at hand while maintaining instant access to other software and data. Shading and rendering will be improved, and a 3-D look will help guide the user through complex processes.

Sidebars, or panes, that appear in Windows and Office applications will also be used more, being made accessible to other developers for use in their applications. Nevertheless, a Microsoft spokesman assured IEEE Spectrum that the intention is to "continue to provide the Windows user interface look and feel."

Many smaller companies that build products for the mainstream market are also caught up in this familiarity problem: they want to retain their customers as well as their relationship to mainstream products. So they're not going to opt for rapid or startling changes, either.

As a result, most new interfaces are not coming from established companies. Rather, they're sprouting from new ventures, research laboratories, and independent inventors. This limits their immediate impact on the market, but if the design is obviously valuable, there's a good chance that hearts and minds may eventually be won over.