PUTTING DATA IN PALM OF YOUR HAND.

Byline: Evan Ramstad Associated Press 

In his speeches and book, Microsoft Corp. Chairman Bill Gates paints a future filled with gee-whiz things like computerized wallets and watches and homes.

After several fitful starts, the company is at last closing in on a piece of software that represents its first major step toward that kind of future.

The program runs the basic functions of hand-held computers, similar to personal digital assistants like Apple Computer Inc.'s Newton or the Palm Pilot of U.S. Robotics Inc.

And its development illustrates the difficult balance Microsoft must strike - rejecting part of its past and repeating another - to stay relevant as new computerized devices come along.

Its hallmark MS-DOS and Windows programs are too big for small computers. And so Microsoft must restart from a smaller base.

At the same time, Microsoft also wants to re-create in new computers a pattern of program upgrades and repeat buying that was profitable for it in personal computers.

``We aren't very interested in the businesses that represent just writing the code once and burying it in the machine, like a toaster oven,'' said Craig Mundie, the senior vice president in charge of non-PC software development at Microsoft.

The hand-held program, known by the code name Pegasus, is something that both Microsoft and other software makers can build upon. It is due to be available late this year or early next, appearing on new machines made by other companies.

Although project leaders decline to describe it in detail, they expect Pegasus to have some basic database and communications functions, including connecting to the Internet, and fit tightly with personal computers that use Microsoft's Windows operating system.

``What we're trying to do is bring value to these other platforms (new devices) by making them have some synergy with the established Windows computing environment,'' Mundie said.

Windows has become too big for hand-held computers because its latest version, Windows 95, requires 8 megabytes of main memory to run, 60 megabytes of hard drive space to store and works best with a 32-bit microprocessor, such as the Pentium chip of Intel Corp.

Pegasus will be a subset of Windows, able to run on an 8-bit microprocessor, the kind that ran video game systems that were new in 1989.

``We've thought about how to take the most critical functionality and scale it,'' said Jon Magill, product manager for Pegasus.

Microsoft has relied on customer surveys and interviews to help decide what functions it should have an what should be left out. But competitive forces also are shaping decisions.

U.S. Robotics' $300 Palm Pilot has been a critical and sales success since rolling out last fall. Palm Computing, which U.S. Robotics acquired last year, created the Pilot's operating software after going through much of the struggle Microsoft is now in with Pegasus.

``The approach we were all taking before was to say, `We're going to take the PC and shrink it into this little box,' '' said Donna Dubinsky, chief executive of Palm. ``There was this implication that it had to do everything the big box did. What we realized after having some failures was there was a different approach, which was to assume the hand-held is a component of the bigger system.''

Palm designed Pilot to synchronize itself with a desktop PC at the push of one button, a feature Microsoft's Pegasus designers acknowledge they also are trying to achieve.

And, as Microsoft's programmers are now doing, Palm went through some tough design decisions to keep the Pilot's operating program small. One was that it cannot print. Instead, data must be sent to a PC and printed from there.

The rise of computer networks, most prominently seen in the growing popularity of the Internet, helps software designers make such compromises by providing new places for data to reside or be processed.

Just last week, Sun Microsystems Inc.'s software unit said it had created a compact operating system based on the Java programming language that has excited Internet developers for a year.

And last month, Lucent Technologies Inc., the AT&T Corp. spinoff that includes the illustrious Bell Labs, announced an operating system called Inferno for hand-held and other specialized computers that will rely on networks.

``Maximizing all of the assets in the network are some of the key design principles in Inferno,'' said Mike Skarzynski, a Lucent vice president and general manager of the product.

The first devices to run the software will have less than 1 megabyte of main memory and also use less powerful microprocessors. Skarzynski hopes future versions will use even less computing power, even appearing in ordinary telephones at little or no added cost.

Microsoft has the opposite idea with its futuristic software like Pegasus. As with MS-DOS and Windows, the company wants to build a ``platform,'' on which it and other companies can layer on more features that will keep customers hungry for updated versions of software.

Though it hasn't even been publicly displayed, Pegasus is far enough along that Microsoft held a conference earlier this year to share its basic design with companies that build hand-held devices and software that runs on them. In one sign of its possible direction, Microsoft also bought Aha! Inc., a creator of handwriting recognition software.

The failure of several earlier projects, stretching back as far as a ``modular Windows'' concept for Tandy computers nearly a decade ago, to reach the market has left some observers wondering whether Microsoft can pull off Pegasus.

Its most recent similar project, called WinPad, rose and fell during the frenzy in 1993 and 1994 over products like Apple's Newton and AT&T's failed clipboard-style EO computer. Microsoft had several large partners involved in WinPad, most notably Compaq Computer Corp., the biggest PC maker.

``We recognized the products were not going to be great products. So we hit the reset button,'' Magill said.

For now, Microsoft executives say Pegasus has become a bigger short-term opportunity than its other experiments in new kinds of computers, including a trial of interactive TV.

``This is not magic,'' said Harel Kodesh, general manager of personal electronics products. ``We know ultimately we'll get to Bill's vision.''