(FORTUNE Magazine) – Arguing with successful innovation is a dirty job, but somebody has to do it. Sure, innovation can be glamorous and even profitable, but that shouldn't obscure the costs of mismanaging it. Yes, the U-2 and SR-71 Blackbird high-altitude spy planes are masterpieces of aeronautical engineering--and products of the very first skunkworks, the one Kelly Johnson created at Lockheed. IBM's first--and very profitable--PC came out of a Big Blue skunkworks initiative in Boca Raton, Fla. Steve Jobs ran Apple's breakthrough Macintosh computer as a renegade skunkworks that ultimately annihilated its Apple II internal rival. Even management guru Tom Peters has run "skunk camps" for aspiring innovators.

Designed for leanness, meanness, and agility, skunkworks are the organizational innovation du jour for FORTUNE 1,000 organizations intent on becoming more innovative right now! Set up "teams" of passionate intrapreneurs insulated from the imperatives of the business mainstream, and you'll be amazed at what they come up with. Scores of companies are raising their tails and spraying their very own Internet and e-commerce skunkworks in ongoing bids to get themselves up to Net speed. An e-skunkworks becomes a Gen X company within a company. Tattoos are optional; stock options aren't.

There's no denying that skunkworks can work. Maybe they represent the last, best innovation hope for organizations whose costs of coordination have outstripped their economies of scale. But companies that bet on skunks should have no illusions about the management message they're sending to people: We've anointed an innovation elite--and then there's the rest of you.

This kind of "innovation apartheid" may occasionally give birth to great new ideas, but it almost always breeds even greater resentment. Smart, capable people hate being marginalized. They loathe the arbitrary way such opportunities are distributed throughout the firm. Skunkworks become the exclusive suburbs in contrast with the grubby slums of the mainstream business. Where would you rather live?

A brilliant and geeky longtime friend who's had great fun building up the digital skunkworks at a Huge Company was complaining over drinks about how slow his mainstream counterparts were in grasping what he was trying to do. "They try to kill every good idea we have," he moaned. I reminded him that back in grade school, he was almost always the last boy picked for teams in gym class. He blushed at the memory. "You now make 50-year-old males with families feel as if they've been picked last in gym class," I went on, "plus you have financial equity they don't. Why are you surprised they don't like you and your group? They don't think you're their savior; they think you're their enemy."

My friend has now gotten into an argument with his CEO over putting members of the mainstream business on his skunkworks' advisory board to encourage healthier interaction between the businesses. The CEO is resisting because he doesn't want his pet skunkworks contaminated by the legacy culture of his company.

A skunkworks is not a sign of great management championing innovation; skunkworks are instead a signal of management that has given up on innovation. When an enterprise goes skunk, what's the real message? Top management effectively acknowledges that their corporation is incapable of internal organic innovation and must set up a different organization with different people, different values, and different incentives.

In other words, management launches a skunkworks because it has failed to create an organization that can innovate without skunks. Top management gets celebrated for being "innovative" and "visionary" as a result of its own inability to create an innovative culture in the first place. Who takes the hit? The schmoes in the mainstream business who are told they aren't innovative enough to compete in this dynamic, ever-changing environment, blah blah blah.

Setting up a skunkworks is the easy thing to do. Just how hard is it to find talented people who want to work hard to do new things that have an impact? No--what's truly hard is creating an environment that encourages innovation without managerial shortcuts like skunkworks. Or--harder yet--figuring out how to integrate skunkworks-like ventures into the mainstream business. How do you create innovation cultures that aren't based on apartheid? That's the real "innovators' dilemma." Managing innovation is easy; managing expectations is hard. Any enterprise that wants to do the skunk owes it to its people to be upfront about what level of elitism and separatism is acceptable and what level of integration and collaboration is desirable. Organizations have to decide whether their innovation cultures should be built more on inclusiveness than exclusivity. The tradeoffs are undeniably painful, but skunkworks that succeed at the expense of the organization ultimately prove to be Pyrrhic victories.

MICHAEL SCHRAGE is a research associate with the MIT Media Lab and the author of Serious Play. He can be reached at michael_schrage@fortunemail.com.