Lots of people would love to know what fabled software developer Ray Ozzie is up to. More than a decade after he dazzled the world by creating Lotus Notes -- the remarkably durable "groupware" program -- he has formed Groove Networks Inc., a Beverly, Massachusetts-based startup. His company is believed to be working on a new form of collaborative software that might represent a huge leap beyond Notes's capabilities.

But don't expect to hear that from Ozzie. At industry conferences, he greets curious colleagues, smiles, and says, "We're in stealth mode." A visit to his company's Web site (www.groove.net) doesn't yield much more information. Even a section labeled promisingly "All about Groove Networks" consists mostly of whimsical, long-ago photos of a shoe factory in Beverly. A brief note says that Groove Networks is "a jam session" that Ozzie will explain only to people who come to work for him. But don't bother trying to glean tidbits from unsuccessful job candidates. "Of the 100 people we've hired," declares Ozzie, 44, "you can count on one hand the number who knew what they would be working on before they joined us."

A few years ago, Ozzie's brand of my-lips-are-sealed secrecy might have seemed eccentric. Now it's routine. More and more Internet startups are going through their initial year (or years) of development with no publicity, with no clear explanation of their business on their Web site, and, in some cases, with strict instructions to employees not to talk to outsiders about what they do. Silicon Valley venture capitalists say that typically three or four of their portfolio companies are keeping utterly mum -- becoming the business equivalent of pupae sitting still before emerging as butterflies. In March, at the PC Forum conference in Scottsdale, Arizona (the high-profile annual gathering convened by Esther Dyson), nearly half of the new companies making presentations announced that they were emerging from as much as a year of being in "stealth mode."

This spring, in perhaps the weirdest twist of all, the venture-consulting firm Atomic Tangerine bought more than three dozen ads on Silicon Valley radio stations, urging people to go to the Web (www.sri.com/secret) to learn about an amazing career opportunity at a "pre-IPO company." At that time, Atomic Tangerine hadn't yet publicized its name. So, in language that veered close to parody, the consulting firm told job applicants that they couldn't even learn the name of their possible future employer until they signed a nondisclosure agreement.

What's behind this passion for secrecy at Internet startups -- companies that are, after all, planning to do business in a medium that is all about open communication, instant connection, and, increasingly, endless hype? Are we suddenly in a world where people pay lip service to the teachings of Michael Dell and Jack Welch -- but model their actual behavior on the communication style of the late Howard Hughes? Has the new mantra of the dotcom world become "I'd tell you, but then I'd have to kill you"?

It's not quite that bad. Intelligent advocates of corporate stealth see it as a short-term tactic -- one that should be used only briefly, when a startup is trying to sort out initial business decisions and could use a little camouflage. They acknowledge that stealth per se isn't good and that isolation from the rest of the world creates its own hazards. As Ozzie himself puts it, "You should never stay in stealth longer than you need to."

The best way to evaluate the case for stealth is to look at the five main justifications that startup executives and their VC backers offer for it. Each argument turns out to be double-edged. In the right situation, stealth can be a boon. But if it's taken to extremes, or misapplied, it can wreck an otherwise promising new venture.