In high technology, like comedy, timing is everything. If you aim too far ahead, your product might appear before the market is ready for it or before it can be produced and sold at a profit. If you don't aim far enough, it is hard to develop anything that can make money before someone else reverse engineers it. In many ways, then, the technological sweet spot lies somewhere in between, yet few outfits aim for that spot. One such outfit is a brand new company called Packet Design, just announced this week. I see Packet Design as one of those new companies that will make a real difference in both technology and in our lives in the years to come.

The CEO of Packet Design is Judy Estrin, who until less than a month ago was the Chief Technology Officer of Cisco Systems. Another founder is Van Jacobsen, lately Cisco's Chief Scientist. Yet Packet Design isn't a Cisco spin-off, has no Cisco money, and really has no special relationship with the networking giant. That's because Packet Design is a whole new type of business aimed at developing, then spinning-off a variety of mid-term technologies. At its heart, Packet Design is a perpetual startup.

Why would anyone want a perpetual startup? And who would put up $24 million to create one? These were questions I wanted answered when I recently met with Judy Estrin and Bill Carrico, the principals behind Packet Design.

Carrico and Estrin are wife and husband, CEO and Chairman. This is certainly not their first startup. That was Bridge Communications, which they founded with Eric Benhamou in the mid-1980s. Bridge, which made network bridges in the pre-router era, had a successful IPO, and then was purchased by 3Com in 1987. Bill Carrico was, for a time, president of 3Com, one of the many titles that Benhamou carries at 3Com today. After Bridge and 3Com, Carrico and Estrin (who first met when they both worked for Hewlett-Packard) started Network Computing Devices, a company that built X-terminals and became the largest in its market segment. NCD, too, had a successful IPO, and the pair left a couple years later. Then they started Precept Software, a company that tried, maybe a little bit too early, to bring television to the Internet. Precept was bought by Cisco, and its IP-TV product line is doing well on LANs and corporate intranets. That was two years ago.

This is a couple that has shown a consistent knack for starting companies and building them to a certain size. Then they get bored or annoyed, or maybe the pressure from VCs and other investors who want to take their profits and run becomes too strong, so Bill and Judy move on to a short vacation, and then to starting yet another company. It's compulsive.

"I wasn't any good at being retired," said Bill Carrico when I asked why he felt the need to start yet another company. With millions in Cisco stock, the pair certainly doesn't need the money. But beyond giving them something to do, Carrico and Estrin believe Packet Design is addressing problems that are overlooked by most Internet businesses with their focus on the current product cycle.

"Here's the problem," explained Estrin. "Traditional companies are stuck in first generation technology. They have existing businesses to run that keep them from moving forward. Startups, on the other hand, concentrate on solving immediate technical problems. And universities do basic research, but have abandoned to the startups anything that's short or mid-term. So the traditional companies — by these we mean the General Electrics, Fords, and CitiCorps — are forced to paint over the cracks in their technical infrastructures with startup solutions that don't really solve the underlying problems. That's where Packet Design comes in. We will be running three to five parallel projects focused on different time horizons, but all intended to create evolving, backward compatible solutions to structural problems in the IP world."

In many ways, this sounds like Interval Research, the company founded several years ago by David Liddle and Paul Allen as a kind of mini-Xerox PARC to do basic research and license new technologies. Interval recently closed its doors, a failure. The difference between Interval and Packet Design, according to Carrico and Estrin, is the difference between basic research and mid-term solutions. Interval was aimed so far into the future that unless it guessed exactly right, there was little likelihood that its work would result in commercial products or services. Packet Design, on the other hand, is aiming a bit closer and intending to solve problems that already exist if you know where to look. This was a point where it might have been very useful for the pair to lay out for me a number of project descriptions, but they didn't. That's all a secret, though apparently the plan is to work on scalability problems that might also be the target of companies like Inktomi and Akamai. What isn't a secret and what might be even more valuable over time is Packet Design's business model, which is unique.

Packet Design is intended to develop technologies, build companies around those technologies, then spin-off the companies through sales or initial public offerings. While this sounds like a technology incubator, it isn't. An incubator takes in entrepreneurs and their companies and nurtures them leading to a sale or IPO, but all the ideas come from outside. Packet Design thinks it has enough ideas of its own, thank you, so no entrepreneurs need apply. The difference between the two concepts is that Packet Design doesn't have to share equity with any flaky and inexperienced founders, and it doesn't have to, itself, go public. The result is occasional and delightful bursts of liquidity for the original investors without the annoying aspects of taking Packet Design public, like having to share power, or in the case of Carrico and Estrin, having to go start yet another company.

This idea of taking bits of the company public without taking the whole thing is interesting. That's because going public as a whole is a draining and taxing experience in so many ways. Founders get gouged by investment bankers, retail investors get gouged by retail brokers, and founders who thrive with 100 employees consistently fail with 500. If only there was an easier way to take a little money out of the company without having to pretend that having shareholder meetings was part of the original plan. That's why Packet Design likes to see itself as a perpetual startup.

"We LIKE this life," said Carrico, sitting at a rented table in a rented office that will be too small for the company three months from now. And that's the point. For this high-powered couple, excitement is starting a company and building new products, not dealing with financial analysts and investment bankers. They want to stick to the fun parts of doing a startup. And they'll be thrilled if Packet Design never goes public, but instead spins off an Akamai or Inktomi every year or two. Their investors, who include former Netscape CEO Jim Barksdale, and Sun Microsystems co-founders Andy Bechtolscheim and Bill Joy, are counting on Estrin and Carrico to do again what they have already done three times before.

I would have invested, too, but they turned me down.