• Sidebar: Examples of Technology Transfer

Despite periodic press showcases, Microsoft Research (MSR) remains one of the quietest and most unusual divisions at the company, with operational practices found nowhere else at Microsoft. Although it plays an important part in product development, MSR also functions as an open member of the global research community, helping Microsoft gather intelligence and appeal to academics. Understanding MSR's practices, goals, and areas of concentration can give IT planners and partners insight into how Microsoft sees its own product lines and technologies evolving over the next decade and help businesses plan their own research and development strategies.

Focus on the Next 10 Years

Founded in 1991, MSR now has about 600 researchers, five locations (Beijing, Cambridge, Redmond, San Francisco, and Silicon Valley), and an annual budget of approximately US$250 million. This is about 5% of Microsoft's planned research and development (R&D) budget for this fiscal year, which ends June 30. The other 95% of the R&D budget goes to the product teams, which spend it on development, testing, and program management.

According to MSR Vice President Dan Ling, the group's charter is to "investigate technologies that are likely to have an impact on Microsoft products in the next decade." By looking at MSR's work today, one can make educated guesses about what Microsoft believes will be important in the next 10 years, and how it might design its products to meet these trends. Researchers are now working in the following areas:

Internal tools. MSR has designed many proprietary tools that Microsoft uses in its own software development process. Now, it's focusing on tools for creating secure code (e.g., making sure the debugging tool checks for potential buffer overflows). This will help the company reach the goals recently laid out by the Trustworthy Computing initiative. (See "Trustworthy Computing a New Priority" on page 25 of the May 2002 Update.)

Database technology. MSR has already contributed research on data mining and automatic performance tuning to SQL Server and will continue to work in these areas. Researcher Jim Gray, who oversaw the TerraServer project, in which SQL 7.0 databases were used to store one terabyte of aerial and satellite photos, is now working on similar technology for the "scientific domain," such as enabling astronomers to use SQL to store and arrange the massive amount of observational data they collect.

Digital media. MSR continues to be instrumental in improving the audio and video codecs for Windows Media Technologies, and it is also playing a role in digital rights management (DRM) technologies for securing digital content. In addition, researchers are working with the Windows Messenger team on video protocols for the next generation of videoconferencing.

User interface. Researchers specializing in interface design work with many product teams, particularly the Windows client group. A specific concern is designing interfaces that make it easy for users and administrators to set security at the appropriate level (e.g., ensuring that default settings do not sacrifice security for limited gains in productivity).

Languages. MSR is working to improve C# by introducing generics, which are similar to templates in C++ and would allow developers to write code that works uniformly on different types of data (e.g., allowing them to create a list object that can hold integers, characters, strings). This would reduce the likelihood of programming errors and ensure C# stays on par with Java, to which Sun Microsystems is also considering adding generics.

Probabilistic and contextual thinking. Today, computer logic works by calculating whether specific events are true or false. MSR is trying to get computers to account for probabilities and to operate contextually. This could be particularly useful for notifications—for example, a system could look at a weather report predicting an 80% chance of rain, determine that a user is probably driving rather than walking to work, and guide personal messages to her auto PC console rather than her cell phone.

Nanotechnology. Although Microsoft does not research chip architecture, MSR is looking at microelectromechanical systems (MEMS)—"systems on a chip" that integrate digital processing power with extremely small mechanical systems. This suggests that Microsoft is interested in hardware as well as software. Most likely, the company will use this research in developing its vision of the "connected home." (For background, see "Broadband Crucial to Consumer Strategy" on page 17 of the May 2002 Update.)

Relationship with Product Teams Crucial

All this work will be for naught if MSR is perceived as being out of touch with Microsoft's business needs. This is a risk with any corporate research lab: the most famous example is Xerox, which created a graphical user interface at its Palo Alto Research Center (PARC) in the 1970s but was unable to commercialize it.

To minimize this risk, Microsoft has taken concrete steps to ensure that MSR is not isolated from product teams. Unlike many research labs, MSR started on the same campus as the company headquarters in Redmond, WA, and its satellite research labs still report to the main office there.

More important, MSR and other parts of the company exchange personnel and communicate regularly. New researchers are strongly encouraged to forge relationships with product teams in their area of expertise. When a product team decides to incorporate a researcher's work, the researcher plays an active advisory role and, in some cases, actually moves to the same location as the product team and writes code. For example, MSR today has a member working with the team responsible for the Common Language Runtime (CLR) and another contributing graphics expertise to the Xbox. MSR has even spawned an entire product division: the Digital Media Division, which creates technologies for recording, encoding, distributing, and playing digital media on the Windows platform, grew out of an MSR project to improve video compression for video-on-demand applications in the TV industry.

Ling emphasizes there is no formal process for contributing to product development. Rather, it's a two-way street: sometimes an individual researcher convinces a product team to incorporate specific advances; other times, a product team comes to MSR for assistance in solving a particular problem. (For examples of both scenarios, see the sidebar "Examples of Technology Transfer".)

Academic Model Helps Achieve Other Goals

MSR is also designed to be an active participant in the global academic community in several ways.

First, the charter of MSR is purposely broad. Although some boundaries are observed (for example, MSR does not work on chip architecture), Microsoft encourages researchers to investigate popular topics in the field of computer science that might have no obvious or near-term application to Microsoft products.

Second, researchers are judged not only on technology transfer but also on the quality of scientific papers they publish, their influence in the academic community outside Microsoft (measured by awards, invitations to speak at conferences, and so forth), and the quality of patents they hold (20% to 25% of Microsoft's patent filings come from MSR). Researchers are responsible for deciding whether their work needs to be kept quiet for competitive reasons or can be published immediately, and, if published, how it must be protected (e.g., by filing a patent application before disclosing the research). Although the product teams contribute to this process, each researcher has the final say over his or her work.

Third, MSR has a more horizontal reporting structure than the rest of Microsoft, with fewer managerial layers between researchers and the group's leader, Rick Rashid (who reports directly to company founder and technical leader Bill Gates).

These characteristics are in stark contrast to the rest of Microsoft—the Windows group seldom works on features that are not slated for the next version of the operating system, for example. But they are necessary to help MSR accomplish goals beyond technology transfer:

Intelligence gathering. Participating in the academic community helps ensure that Microsoft does not get blindsided by the next major shift in technology. Academic and government research labs, rather than private-sector companies, have been the source of many of these technology shifts, including the World Wide Web; e-mail; the Transmission Control Protocol/Internet Protocol (TCP/IP) used for transferring data over networks, which formed the technological basis for the Internet; and various operating system kernels, such as Linux and BSD. Having a group of people that is in contact with researchers outside the company allows Microsoft to stay abreast of new developments.

Public relations, particularly in academia. MSR serves a general public-relations purpose, as evidenced by the steady stream of press articles about its more forward-looking projects. But equally important, MSR helps Microsoft make inroads in the academic computer science community, much of which used to view Microsoft as a manufacturer of a relatively inelegant operating system (DOS) and desktop software. This is important because universities train tomorrow's programmers, and many are now teaching Java rather than Microsoft-oriented languages, such as Visual Basic and C#. To help meet this goal, MSR has a University Relations Group, which conducts faculty workshops and seminars on Microsoft technologies and contributes funds to university research projects, among other things.

Recruiting renowned scientists. The academic evaluation model and lack of strict structure in MSR helps Microsoft recruit well-known researchers who might otherwise be more comfortable in academic settings. Since establishing MSR, Microsoft has attracted a number of luminaries, including three winners of the Association for Computing Machinery's Turing Award (Jim Gray, C. Antony R. Hoare, and Butler Lampson)—an award given in the field of computer science that is considered on par with the Nobel Prize. According to Ling, recruitment was the main reason MSR opened offices beyond Redmond. For example, Cambridge is known as a center for programming language and systems research, and Microsoft realized that valuable researchers there might not be willing to move to the Seattle area.

Resources

For more information about MSR, see research.microsoft.com.