HOW TO INVENT GREAT PRODUCTS.

Dan Bricklin has spent most of the past two decades inventing whole new product categories. In 1979, he and Bob Frankston created the first electronic spreadsheet, VisiCalc. Bricklin's next big concept was a DOS-based prototyping and simulation tool called Dan Bricklin's Demo Program. He then helped launch Slate, a developer of pen-based applications, and recently co-founded Trellix Corp., where he's been working on a leading-edge authoring environment for large Web documents.

Like most inventors, Bricklin doesn't always hit financial home runs. (He lost control of the spreadsheet concept through a combination of bad advice about patent protection and an unwieldy publishing contract.) But his software inventions usually have a profound downstream influence on the products his successors create; in fact, Windows spreadsheet developers still rely on design conventions that Bricklin first devised for a character-based 64K Apple II platform.

Where do Bricklin's product ideas come from? "I picked up a lot of my early training working on commercial typesetting and word processing systems," he says, "and most of the products I've developed since then have been tools for creating and controlling output, for helping people use computers more naturally to express themselves." Bricklin's products are also known for intuitive interfaces and close attention to details; again, Bricklin says he gained valuable insights about usability from typesetting systems, "where every unnecessary keystroke and operator error literally costs money."

We asked Bricklin to sum up a few of the principles he thinks are most important for creating new products:

*Watch what people really do: "Computer science labs often produce interesting technology, but a technology isn't a product," Bricklin points out. Bricklin says many of his best ideas came from simply observing the kinds of tasks that people find annoying or frustrating. "When I was in business school," he says, "we were always recalculating numbers on a blackboard. I wrote a little Basic program to speed up the calculations, but then it dawned on me that calculation speed wasn't the problem--guys with programmable calculators were much faster than my Basic program. The real problem was the amount of work it took to set up a series of related calculations and display the results. Solving that problem was what made VisiCalc a much better solution."

*Create real code, not just spec sheets: "Programming is like any other medium--there are constraints and solutions you discover only when you are forced to create something real," he says. "In Demo, the whole product depended on a little loop that switched from one compressed page to another. There was only one way to find out if the loop was fast enough to keep up with keystroke commands--I had to write the code and see what happened."

*Drill down to the essence of the product: One of the toughest disciplines for software developers, says Bricklin, is to identify a new product's core concept. "Many products are overcomplicated and almost impossible to explain," he notes. Visual metaphors--the VisiCalc grid, the "ink" of Slate's pen-based environment, the Trellix navigational map--are especially powerful ways to communicate core concepts, Bricklin adds, "because they combine the product's function with its user interface."

*Fine tune the controls: In addition to a core concept, says Bricklin, "a good product lets the user control the computer almost intuitively, without awkwardness or excess keystrokes." The Back button on a browser, tabbed dialog boxes, rulers and automatic word wrapping in a word processor, and Visio's "intelligent shapes" are examples of well-crafted controls that Bricklin admires. "They seem completely natural, but they are all very powerful interpreters for an underlying process that really isn't simple at all."

*Get lots of feedback: Bricklin is a strong advocate of constant and early usability testing; in fact, he created Demo to make it easier for developers to prototype user interfaces. "It's amazing how much you can learn from watching people fumble around," he says. It's also important to encourage honest criticism from co-workers and friends. "When I was designing VisiCalc, people kept saying, 'you can do better,' and it really helped."

Dan Bricklin, chief technology officer, Trellix Corp., 51 Sawyer Rd., Waltham, Mass. 02154; 781/788-9401. E-mail: [e-mail].