Microsoft research does the PDC keynote

Wednesdays keynote here at PDC was given by Rick Rashid from Microsoft Research, some interesting highlights of the reams of facts and figures that were given here
- Microsoft Research is modelled on an academic structure like a university faculty
- 15% of the money Microsoft spends on research goes to outside researchers.
- They grow at rate equivalent to creating a Berkley computer science faculty every year
- 20% of US computer science Phds have worked at MSR at some point.

They've also been busy on doing formal proving of software in particular they have partially solved the halting problem, they have a product called terminator that can say if C program will complete, it doesn't work for all programs but it does for a lot of the real world programs.

Dryad and DryadLINQ are two new technologies for cluster computing which allows management of programming across multiple clusters with a sensible query language. There was a little dig at Google with a comment that this gives you much more power than map-reduce algorithm that powers a lot of googles code.

There then followed a segment about using small remote sensors for the environment. Lots of good and worthy stuff but it's not something to really excite a geeky audience.

In the sex and sizzle department, there was a demonstration of Boku which is a programming language for kids. It looks like a perfect system to generate small 3d games with visual actors on screen. Programming is done with the XBox 360 controller and an iconic programming interface.

We were shown in a very quick five minute hello world demo how to generate a little shootem up which promptly destroyed Mat MacLauriens player character. Looks like a great system, worth noting that the kids who have play tested this said that the ability to shoot was mandatory.

Also mentioned were the RoboChamps program and the new version of the World wide telescope project which has 1.5 million active users.

The other fun technology on show was a Microsoft Research in Cambridge project called SecondLight which expands out the surface touch technology. Using an LCD shutter and a projector they sync up the shutter with the projector and project one layer onto the main surface and secondary projection which can then be shown on a diffusing surface (a piece of tracing paper). One example which was interesting was on the main screen a image was being shown and then the presenter took a scroll unrolled it over the screen and text was magically revealed.

The final segment of this demonstration was done with a small panel that had the ability to be tilted and the image displayed was distorted to allow the user who was looking at the panel to see the image correctly. A couple of their suggested applications were the ability to literally scoop a window from the main screen and then work on it on the new surface or wave the surface "through" a 3D image to get a slice of data.

Phew anyway off to catch up with panel on the future of programming languages.