Spreitzer himself is the kind of scientist one expects to find at Xerox PARC. He's friendly and open, a 41-year-old bespectacled man with thinning hair who goes back and forth between gifted communicator and opaque computer scientist. He talks a lot about granularity, modularity, new paradigms, and the "divide and conquer" approach to engineering. Spreitzer even looks the part, dressed in tennis shoes, beat-up jeans, and a pullover polo of a kind that can be found in the sale bin at K-Mart. His book shelves are stacked high with computer journals and monographs; in a corner sits his backpack, a freebie stamped with the IBM logo. Spreitzer has been associated with PARC since the age of 23, when he started working there first as a summer intern and then as a Ph.D. student. He has worked as a full-fledged PARC researcher since 1989, and he's been working on http-ng since late 1996, but his office bears the markings of the distracted. It has no decorations unless one counts a wipe board dense with scribbles or the foam-board visual aide he created to explain http-ng at a World Wide Web Consortium (W3C) meeting held last year.

Yet this is hardly the Xerox PARC of old, as Spreitzer and his colleagues well know. At the new PARC, new projects aren't funded simply because they sound interesting; gone are the days when researchers were free to follow a good idea wherever it took them, its money-making potential be damned. "It's been a challenge keeping this project going inside of PARC," Spreitzer admits. "In fact, it was a challenge just getting it started. We had some difficulty with http-ng, telling the story on how it benefits Xerox." Because it's about technical improvements that are very generic, it would benefit Xerox's competitors as much as it would benefit Xerox. "It used to be we were pretty independent," he continues. "We did our own thing as we saw fit and that was the end of the story. Now the business people have a say in what we do with the money."

When I ask if occasionally the powers that be pull the plug on research midcourse, Spreitzer doesn't hesitate: "It happens all the time." That's his worry on the http-ng project. Helping to create a refined set of protocols that serves as the backbone of the web offers the mothership a core competence worth a few years of his time, but Spreitzer knows it's just a matter of time before PARC pulls the plug on his project. Then it will be up to Microsoft, Sun Microsystems, and other interested parties who are part of the W3C to duke out the final implementation.

YOU CAN STILL catch glimpses of the future when visiting the PARC campus. But whereas at the old PARC a visitor would swallow the future in overwhelming gulps, now it's more like drinking in little sips. "Is anything they're up to at PARC the equivalent of inventing the PC?" asks Novell CEO Eric Schmidt, a PARC alum. "I'd like to say there are some very smart people doing some really amazing things there but, to be honest, I think not." No one doubts that PARC circa the 1970s and the early 1980s was a magical place. The first general purpose personal computer, the first networked PCs, point-and-click computing, the laser printer, a mouse that could serve as more than a pointer -- these are among the technological breakthroughs attributed to the PARC of old.

"To me PARC is the most productive research lab ever... the lab against which every other lab is measured against," says Michael Hiltzik, a long-time Los Angeles Times technology reporter and author of Dealers of Lightening: Xerox PARC and the Dawn of the Computer Age (published earlier this year). Yet there are two prevailing, diametrically opposed views of the current PARC. The conventional wisdom holds that it's just another industrial R&D outfit, distinguishable from other facilities of its kind only because of its storied past.

Schmidt and the Institute for the Future's Paul Saffo are among those who see that view as unduly cynical. "There are still people doing pretty amazing work at PARC," says Saffo, who keeps close tabs on the goings-on at PARC, both as a friend of facility director John Seely Brown and as someone whose job it is to keep an eye on what's around the next digital bend. "Xerox PARC is working on some pretty out-there stuff that makes you wonder how it's ever going to be tied to a product. It's still a world-class place."

The truth seems to lie somewhere between these two views. PARC researchers are working on their share of provocative projects. Electronic Paper, an invention eight years in the making and slated to hit the market in the year 2001, is "electrified" paper (a special pen electrifies microscopic beads suspended in oil to show text and pictures) that can be reused thousands of times and stuffed in a pocket. A touted researcher named Bernardo Huberman is studying internet "ecology." He and his small team are delving deep into existing disciplines created to impose an understanding of complex systems (sociology, economics, politics), to help improve the workings of the web. Smalltalk, the first object-oriented computer language, was developed at PARC; now researchers are working on what they hope will be the next big advance in modular programming -- "aspect-oriented programming" How cutting edge is aspect-oriented programming? When I asked PARC PR manager Lois Wong for a description, she said she had listened to the main architect's presentation "three times and I still don't understand it."

THE TRUTH IS that many, if not most, of the "really, really exciting" projects that Wong told me about were more modest efforts -- narrowly proscribed, routine-sounding R&D projects that represent incremental advances in technology. They seemed aimed at either improving existing products and services offered by Xerox, or developing new software applications that are all about Xerox joining the race to cash in on the internet. While at PARC, at Wong's urging, I met with Sanjay Swamy, the marketing director for a Xerox business unit selling a product called ContentGuard. In the early 1990s, some prescient folks at PARC saw that soon the world would be distributing important documents over the internet, so they devised a way of embedding copyright protections within an electronic document. The resulting product, the recently released ContentGuard, allows publishers to protect their intellectual property over the internet, whether text, music, art, or video. Using ContentGuard, a music-oriented web site could allow users a free listen to a song but prevent them from saving it to disk unless they pay a fee. An interesting product, sure, and one that fits nicely with Xerox's view of itself as the "digital document company," but hardly a technological breakthrough. "ContentGuard might not be based on any exciting new technologies," Swamy tells me, "but it opens new markets that will have significant revenue impact within the publishing industry."

Similarly, FlowPort is another PARC-born product that seems more an emblem of the new PARC than on par with the old. "That one's really, really hot," Wong had said with an excited gasp when I asked her about FlowPort. David Smith, who heads the Xerox unit selling FlowPort, gave me a demo of this product released a couple of months back. He inserts a document into a Xerox photocopier connected to a network. But rather than spitting out a copy, it has performed the range of tasks that Smith has asked it to do in a checklist cover sheet: it's emailed copies to Bill and Bob in Marketing, faxed a copy to a customer named Joe, generated hard copies printed by photocopy machines in the Omaha and Toledo sales offices, and saved the document on a server -- all, as Smith likes to say, "with a touch of a single green button."

"What I'm doing is very pragmatic, very practical. I'd like to help businesses leverage what they've already got in place, to make work processes go faster and to move things more simply and more efficiently," says Smith. Both FlowPort and ConentGuard are interesting products that sound like they just might fare well on the open market, but it's the kind of stuff that pales when set alongside breakthroughs such as Bob Metcalfe's invention of the Ethernet while a PARC researcher. Nowadays, when they talk about "breaking the mold" inside Xerox, they're as likely to mean commercializing what PARC has produced as creating a new generation of technology.

"The PARC of today is always going to be measured against the PARC of old, and in that regard it's going to fall short," says Michael Hiltzik. "But none of today's industrial labs will measure up, except maybe at AT&T and IBM because of all their Nobel laureates. It's for the same reason that nobody can ever measure up against Shakespeare, who set the standard." Pity the folks at Xerox PARC. For years, the pundits mocked Xerox for failing to make money on PARC's many technological breakthroughs (a myth, actually: the company eventually earned billions on the laser printer). And now that they've gotten more in synch with commercial reality, they disappoint those like myself who arrive at its facility with soaring expectations.

"Keep in mind that R&D is an extremely difficult thing to do," Paul Saffo says. "Even when you do it well it's extraordinarily difficult to connect that R&D to the company. But Xerox is still struggling with ways of connecting research to the corporate mission. And, in fairness to the current-day PARC, the 1970s or early 1980s were very different from today." Adds Eric Schmidt, "PARC in the '70s and '80s was able to assemble the 'dream team' of top names in computer science. They had something of a monopoly on the smartest people. It's no longer possible to aggregate that level of talent. What's happened now is people have of course studied this model and replicated it. So its success has created all these competitors."

IN 1992, PAUL ALLEN ponied up $100 million to create Interval Research, modeled on PARC. The idea, as David Liddle, the PARC alum Allen chose to create Interval, described it in early 1998, was to create a place where researchers were free to "follow a good idea wherever it goes." Since funding Interval, he's made a slew of investments in cable companies and other concerns, anticipating the maturation of the digital infrastructure. In September, Interval posted a press release on its web site announcing that Interval was changing its focus. "When Paul and I first started Interval, it was intended as an unbounded research effort," Liddle said in a prepared statement. "Now that Paul's 'wired world' strategy and effort is taking flight, it makes sense for Interval to be able to support this strategy, find synergies with our sister companies, and seize broadband opportunities." The press release also revealed that Liddle had been kicked upstairs to take over as the organization's chairman. There was a time Interval was touted as the new PARC, but the same reality that caught up to PARC seems to have caught up to Interval. "To me it looks like Interval is being aggressively redirected as an in-house resource for Paul Allen's companies," Paul Saffo says.

To Saffo, the only fair way to judge the current-day PARC is to compare it to other modern-day industrial research centers. Microsoft has set aside $3 billion a year -- or roughly one-seventh of its annual revenues -- for R&D. PARC, in contrast, has an annual budget equal to one-third of one percent of Xerox's annual revenues. "Compare PARC to something like Microsoft research," Saffo says. "Microsoft research is an intellectual roach motel: All the big brains check in but nothing ever checks out." He defies me to name a single big breakthrough that's come out of Microsoft's R&D efforts and of course I'm stumped.

Eric Schmidt goes one step further: He challenges me to name a single earth-shaking breakthrough that's come out of any industrial research center over the last 10 years. Even the Java programming language, the big breakthrough at Schmidt's old company, Sun Microsystems, was born not in Sun Labs but as a small rogue operation started within the company. "What Xerox did in the '70s is really hard to produce, and we've all tried. I've tried this at Sun and I'm trying it here at Novell. That's why I'm real sensitive about these criticisms of others," Schmidt says.

"The problem in computer science right now is the eating-the-seed-corn problem," Schmidt says. "Instead of doing what I did, which is basically laboring over a Ph.D. and working in a research lab or university, people at age 22 who are arguably better than I was at that age aren't getting advanced degrees but going off and being internet gazillionaires. They're able to do that because they're damned smart, but the community as a whole is losing a trained community. It's not that I have a solution for this, just the observation that a lot of the basic research isn't being done."

In that context, PARC is more a solution than the problem. "The big thing here is attaining a balance," says PARC's Mike Spreitzer. "Sure, there's this recognition that we can't spend all our time on stuff where there's no clear benefit for the company. But there's a lot of recognition from the highest levels that a big part of what makes a research lab valuable is serendipity. You can't be sure what pays off."

Gary Rivlin is author of The Plot to Get Bill Gates : An Irreverent Investigation of the World's Richest Man... and the People Who Hate Him