MSR has about 850 PhD researchers world-wide, working in a variety of areas. That’s a larger faculty than all of Carnegie-Mellon, or the equivalent of building a Berkley faculty every year for the last 17 years.
The fundamental concepts in computer science date back to Turing and Church, and are the results of heavy, long-term investments by corporations and governments.
1990: MSR was created in an environment where other companies were scaling back on R&D funding… even though Microsoft was a “small” company at the time ($1B in sales, 5000 employees)
MSR mission statement
- expand the state of the art in each of the areas in which we do research
- rapidly transfer innovative technologies into Microsoft products
- ensure that Microsoft products have a future
Organizing for innovation
- university organizational years
- open research environment
- strong ties to university research
Huge impact in the field: over 4000 published peer-reviewed papers; Microsoft tends to have 10-30% of the papers at major CS conferences. Over 1000 PhD interns per year, strong ties with universities, heavy participation in the broader research community. Global presence.
There’s an entire PM team whose goal is technology transfer. There are joint product/research teams. Incubation products also exist, such as Microsoft Robotics (Concurrency Runtime/CCR and Decentralized Software Systems/DSS).
Value of MSR
- source of 25% of the company’s patents, and tend to be more “fundamental”
- solves hard problems
- early warning system: ears to the ground over a broad range of areas
Basic Research -> Agility
- basic research allows a company like Microsoft to respond more rapidly to change
- it also provides a reservoir of technology and expertise that can be brought to bear
Looking Forward to 2020
- Provable systems: SLAM/Static Driver Verifier; software model checking; input is source code and API rules; an abstraction is created; the model’s state space is systematically explored; “Terminator” can now prove termination/liveness properties
Current areas of research
- Church’s thesis proven
- CHESS
- Code contracts
- Pex
- Z3
- DyradLINQ and Dryad: cluster computing
- Small sensor packets with wireless connectivity
- Healthcare: personalized medicine; X-prize for mapping an individual’s genome for <$1000; how do we get the information we need? There are mathematics and CS theories around information that help with understanding how the human body works; CodePlex has the Microsoft Computational Biology tools
- Education: Center for Collaborative Technologies; robotics work (Center for Personal Robotics in Education); World Wide Telescope (new version released today – Autumnal Equinox version); project Boku (system to allow kids to program and learn on their own)
If you only see one thing from MSR, the WWT is it… .absolutely incredible. 21 giga-parsecs of imagery.
SecondLight
- next gen Surface unit; the goal is to move off the surface and into the area above the Surface.
- demo actually projected different information on to the screen and a piece of tracing paper held above the screen
- the innovation is in the surface material – the display is a special crystal that is opaque normally, but transparent when a voltage is applied
- two different images are projected, coupled with the switching of the surface transparency
- the infrared camera is used to detect both objects over and on the surface
- demo: showing a lens above the surface… with the image projected onto it being pre-distorted as the lens is tilted so that the image on the lens is always correct
- the second, independent surface, which has *no* instrumentation, can also be touch-enabled