Microsoft Research: Beam Me Up Scotty?

That's the yin and yang of technological advancement: On one hand, the Worldwide Web and e-mail have combined to create the "Global Village" that Marshall McLuhan described in the early '70s. On the other, that village keeps sending us pleas from the sons of deposed Nigerian presidents. Today, most people who track these sorts of things estimate spam accounts for more than 80 percent of all e-mail traffic.

As it turns out, Microsoft has more than 700 researchers scattered among labs around the world who think Big Thoughts about these and other technological ramifications -- and how to deal with them before they get completely out of control. At its Mountain View, Calif., campus on Wednesday, Microsoft demonstrated several projects now under way at its Mountain View, San Francisco and Redmond, Wash., labs.

An overflow crowd of more than 300 -- comprising a few suited-up venture capitalists and investment bankers, and lots of technology wonks -- swarmed around the demonstrations.

That spam problem, for example? The Silicon Valley lab (aka Mountain View) discovered that some spam Web pages can be identified through statistical analysis. So, the nearly 700,000 hits to a site purporting to show nubile teens is, obviously, porn. And the nearly 8 million hits to an odd little site in Italy? Well, that's the Web page whose developers hijack nearly 80 percent of all Italian queries before sending them on to the rightful target. The result: One heck of a winner in the world's search-engine popularity contest. Microsoft's scientists will soon publish a paper on their findings, enabling any vendor to make use of their statistical insight.

In fact Michael Schroeder, assistant director of the Silicon Valley lab, explained that all Microsoft scientists same the face "publish or perish" imperative as their academic counterparts. Which brings up an important note about the technology being tested inside the Microsoft crucible: Don't expect these lab-based explorations to translate immediately -- if ever -- into tangible products. The software giant's great wealth gives it the luxury to fund studies more akin to basic science than applied research.

"The moment you tie what you're working on to a product, it limits how you think about the research," said Microsoft researcher Patrick Baudisch, in front of his display of Visualization and Interaction for Business and Entertainment (VIBE). "It's better for us to look further into the future. We are the scouts of technology, and sometimes the product teams pick up what we've working on."

Baudisch is working on the problem of sorting files across giant computer screens. Accent on that word "giant," as in spanning a 20-foot wall. Once screens become this large -- and there's every reason to expect they will -- the question arises: How will we move files to a folder that's 12 feet over and 3 feet above our heads? Jumping does not seem the most attractive technique.

Baudisch calls one approach "drag-and-pop." Think Star Trek transporter meets computer science. That's because the underlying system looks at the kind of file being grabbed and the general direction the user moves her hand -- then creates beam-like paths to the likeliest folders. The user simply drags the folder to the correct path, and the system pops the file into its new home.

VIBE appeared to be a hit with attendees, who were stacked 10 deep around the booth.

Other popular demos included shielding against computer worms, preventing application conflicts, managing thousands of photographs, shortest-route calculations and mapping the United States, foot by foot.

"The demos have been fascinating, since they're looking at many of the same issues I am," said Jim Meehan, a principal scientist with Adobe, of San Jose, Calif. "The photo triage project has to do with managing huge amounts of image data, which is something we have to think about. And that one over there," he said pointing to the booth labeled Race Detection in Managed Code, "shows some clever ways of making sure applications don't clobber each other when they run into conflicts."

One project not on display is Boxwood, a new way of thinking about virtualizing storage across hundreds of components. "The innovation we figured out is a better way to build a distributed B-tree that scales well," Schroeder told CRN. "We think we've provided a kit to write file systems, Web search engines -- whatever can make use of the storage farms."

According to Schroeder, who said he's been working on the concept for nearly 10 years, the lab will soon write design papers on Boxwood. "We're trying to build a convincing enough demo that it becomes obvious to make these into products. The burden of proof is on me and people like me to get it to the point where others can really see the benefit."