MOUNTAIN VIEW, Calif. -- Microsoft opened the doors to its Silicon Valley research center here to show off the latest technologies it has under development. The show, called TechFair 2010 Silicon Valley, is similar to the TechFest events held at Microsoft's (NASDAQ: MSFT) main Washington campus.
Microsoft Research is the home to a lot of very smart researchers with PhDs and advanced training in very esoteric technologies who don't worry about selling technology, just how to develop it; the product part comes later.
As Rick Rashid, senior vice president of research at Microsoft noted in his keynote, many key developments have come out of the software giant's research division. He said that "there isn't a product on the market today that hasn't been at least somewhat affected by Microsoft Research." This includes everything from Microsoft's Bing search engine to the Xbox to SQL Server to its new Azure cloud services.
Microsoft's product people are charged with finding a home from a lot of the research undertaken at facilities like the software company's Silicon Valley campus.
One example was Contrail, a mobile-to-mobile service for exchanging information without requiring an intermediary. Today if a person wants to share photos, their location or what they are currently doing, they can upload pictures to Flickr or Facebook, or update their status on Facebook or Twitter. But that means the whole world can see it unless the user spends extra time on privacy settings to restrict access.
With Contrail, a person can take photos with their phone or set their location and those pictures or that information is sent through the Azure cloud service, but not stored, to the approved receivers. This is done through what Microsoft calls "edge filters," that are set up on both the sender and receiver's phones.
The user can choose to share photos with one or many people of their choice and never put them up on Flickr. An employee can make their location known only to their boss, and only during work hours on weekdays. The product is built, according to Microsoft researchers, now they need to productize it.
The highlight of the show, with the biggest crowds around it, was The Translating Telephone. Two researchers spoke to each other via headsets, even though they were a few feet apart. One was a native German speaker. He would speak in German, and the computer would recognize the German, translate it to English, and display the English on the other person's screen. It even read it aloud in a computerized voice. The English speaker's words were translated into German on the other person's screen but, in the demo at least, not spoken aloud. Overall, the demo had a few other hiccups but for the most part worked well.
This required three technologies: speech recognition, language translation, and text-to-speech. Microsoft said it wouldn't be feasible yet for mobile phones because the horsepower required would drain the battery too fast. Then again, it won't be on the market any time soon, either; Microsoft estimates the technology it needs for a finished version is still five to 10 years out.
Another lab experiment is the Privacy Integrated Queries (PINQ) language, a derivative of Microsoft's LINQ data-oriented language. It's used for doing data queries and allowing outside parties access to data while maintaining the privacy of that data.
Say a medical insurance firm wants to know how well a certain type of preventative medical treatment worked. Most people would balk at their insurance company using them for data mining. PINQ is designed to let an outside entity like an insurance firm query a hospital or medical group, find all the patients who had said treatment and see how many avoided the need for further treatment. This information could then be used to help determine how effective the preventative treatment was and whether to approve or deny it in the future. The key is PINQ keeps patient information anonymous so, for example, an insurance company wouldn't glean specific patient information.
The Bing group showed several projects in the works, including one that does real-time searches of Twitter. While that's possible now, Twitter does not filter redundancy. Say you wanted to know who was voted off "American Idol" this week. The same link might be re-tweeted thousands of times. Bing/Twitter will filter the repeats out and show just the most popular link once. It will also intelligently check links for spammers and links to dangerous sites and block those people's tweets.
In his opening keynote, Rashid talked of how there is so much compute power and data available, Microsoft isn't just working with fellow tech firms, it's working outside of computing in areas like HIV and hepatitis research, in ecology and hydrology and environmental research.
"The field of computer science, the ideas, and technologies are becoming an important underpinning to all the things we do. This is being referred to as the fourth underpinning. Increasingly, the availability of large amounts of data and computation power are having a tremendous impact on how science thinks about broadly solving problems," he said.
An example of this new kind of problem-solving involves monitoring conditions along the Russian River, a major water supply in Northern California that's also vital to the local fish supply. California is monitoring water flow and water supply to make sure endangered fish don't get trapped, for example. Working with Microsoft, the State can now tap a computer mashup of everything from rainfall to sediment patterns and they can monitor conditions in the water and control how much they let through the dams.
With the advent of this data-intensive science, it's becoming more about collecting data and understanding it, because sensors have created a "deluge of data" and understanding, said Catherine van Ingen, a partner architect at Microsoft Research.