Random Organized Thoughts

So here I am again at PDC on day 3. Today has the keynote from MSR followed by a day filled with sessions. The day began with Rick Rashid from MSR talking about the history of MSR — when they were formed, why and what they had achieved. He spent almost 30 mins, in what I felt was, selling MSR. Not sure why esp. since he was talking to a developer audience and it was not clear as to if they were even interested. Looking at the live blogs, it seems like the press folks who are blogging away live are yawning too. Rick spoke about NUMA & MicroKernel – 2 terms he claimed to have coined. Interesting.. I did not know that. He also spoke about himself being the only executive at Microsoft who has had the same job for the longest period of time – 17 years to be precise. Another interesting stat that I was not aware of but was there a point? And also I am not sure if others thought if that was a good or a bad thing. In my personal opinion though (knowing Microsoft), I personally think it is great to have the guy who defines the vision & mission statement stay there as long as possible to realize it as much as possible. He spoke about how the MSR organization structure was flat and resembled a University model. He also called out that 15% MSR budget was invested in funding the work in the various universities. Some other key stats he shared were there were now 850 people in MSR and they had more than 2000 publications during their 17 years. His reasoning for the need for a research team was that it provided agility when the market was down and provided new ideas/direction when we needed them most. It allowed them to foresee the future and prepare for it when Microsoft needed it. Rick now spoke about some of the projects MSR had been working on.

In the field of Software Engineering, he spoke about

• SLAM which is a static driver verifier rule development kit and does software model checking
• TERMINATOR which is a research project focused on general liveness properties and automatic methods for proving program termination
• Dryad LINQ used to simplify distributed computing on large computer clusters using SQL like query language

In the field of Environment, he highlighted

• SensorMap & SenseWeb – this was cool. I’d ask you to check out the website and learn more. This basically involved a sensory device that was very small that they had developed that could be placed anywhere and it would gather key environmental data and send it back to be able to map/plot it to better understand the environment and make environmentally friendly decisions. Foe example he showed us the heap map for the convention center keynote hall for the last 2 days using around 100 sensors placed within this one hall and monitoring it every minute to see how the rooms temperature was maintained and where the inefficiencies were. It was a pretty cool demo to watch as he showed over time where the various hot & cool spots were & which part of the room was not air conditioned properly. This was an example of sensing in the box. They also spoke about sending in the cloud and sensing in the wild (the Swiss experiment project). Some really neat stuff

In the field of Health, he highlighted

• Personalized medicine research &
• Applying machine learning to understanding HIV/AIDs – this I felt was really cool to see how today’s computer science concepts like ‘how spam worked’ were applied to understand & cure diseases like HIV/AIDs. Mind blowing.

In the field of Education, they highlighted

• Conference XP
• Tablet based computing advances
• Robotics
• World Wide Telescopes (WWT) – this is a rich virtual telescope that is a collaboration between MSR and various academic & government agencies. It is available as a free resource to help guide explorations of the universe. They mentioned having 1.5 million active users and also announced the release of the next version Equinox today.
• BOKU – talk about saving the coolest for the last. This demo really rocked! The idea here is how to inspire kids on programming and make programming not a skill that you learn only if you need it but something that every individual possesses or learns as they grow up. What better way to do that than make it a game. BOKU is just that – it introduces the programming paradigm and thinking in kids without having them to even see/write any code. They do all the programming in virtual visible constructs as objects and actions they take on those based on some constructs. You’ve got to check out the demo to understand how it works. I promise you that you will be blown away. Where was this when I was growing up? The entire audience in the hall was clapping/applauding as they all loved it. Rick did later mentioned (I read online) that the code name ‘BOKU’ was going to change before they released this as a google search brought up all sorts of kid-unfriendly sites/articles. Funny.
• Second Light – this was another cool demo that was well received. Here they took the surface computer and extended it to add another layer on top of the surface without much additional hardware. You can take a trace paper or any such object and project a different image only on the trace paper (different from that on the surface). So let’s say you are looking at a map on the surface computer. Now if you were to hold the trace paper on top of a certain location or region on the map, it can provide more details about that location – some text maybe, historical info, etc (whatever you want to show) — on the trace w/o changing the original map on the surface computer. So you are basically seeing 2 images – all through some intelligent science of fooling your eyes using the same equipment to project 2 images – the 2nd image on the trace paper is projected by the same surface computer without altering the image on the surface computer through some fast switching of images that are not visible to the eye. Pretty cool.

All in all the later part of the keynote with the demos was more exciting than the beginning and made it worthwhile attending.

Now for the sessions that I attended today

1. Windows 7: Sensor & Location Platform: Building Context-Aware Applications (by Dan Polivy)

This was a session focused at a pretty narrow audience – developers who were targeting sensors, GPS & other location based devices. The speaker spoke about the various API’s they were providing in Win7 for application developers to leverage these hardware in a standardized way without having to program separately for each individual device. It was not too relevant/exciting for me.

Video & Slides of Talk: http://channel9.msdn.com/pdc2008/PC25/

2. Windows 7: Design Principles for Windows 7 (by Samuel Moreau)

I am not sure the speaker achieved the goal he had for this session. Based on the title, I believe the goal was to share the design principles and maybe explain them. But I found that Samuel spent too much time in the introductions and later rushed through some of the design principles and even then seemed to share just a few incomplete set of design principles. The talk felt incomplete. It would have been nice if he had shared more stories & used more examples like the one he did about how the original idea for accessing the jump list (using a drop up list next to the taskbar icon – essential making the taskbar icon a split icon) was dropped since it was good for feature discoverability but bad after user understood the concept. It would be in your way & annoying and accidently accessing the jump list was seen as common in usability studies. As someone who has worked on every build of Windows, I always wondered why they changed it and now I know. Personally I felt the ‘Welcome to the Windows 7 Desktop’ session on  the 1st day was better than this one and shared more design principles than what Samuel managed to convey.

Video & Slides of Talk: http://channel9.msdn.com/pdc2008/PC22/ 

3. Optimizing Applications for Remote File Access over WAN (by Mathew George)

Mathew is a good friend of mine (we went to school together – UIUC) and I knew he was presenting. But the problem was I did not know when. The reason I found out later (the hard way) was that his session was not in the PDC session pamphlet they had handed out to all attendees on day1 as part of their PDC badges. So at 3pm, I was trying to find Mathew’s session and could not find any title that closely resembled what he was working on. I went to a session on Offline access and found out that was a talk by Pablo (not Mathew). Note: At any point in time there are 10 or more sessions happening at the same time. Mathew’s name was not listed on any of those sessions. So I gave up trying to find his talk and was loitering around the convention center floor checking out the booths. I accidently passed by the Information Center in the middle and picked up a sheet of paper which said ‘Updates to PDC sessions’ and in that sheet was listed 2 new sessions that had added to the PDC session list along with several others which were dropped or whose time had changed – one of them was Mathew’ s session. Talk about discoverability! Very few people (if any) would even know about the session to attend it. And I was right. When I walked in there were just 8 or 10 people attending the session. Among the sessions I attended, this was the session with the lowest attendance I saw at PDC – and this was no fault of Mathew. I felt really sad for him. He had put a lot of effort and time into preparing his talk and his content was just right. It was a really good talk. His talk focused on do’s & don’ts for application developers on how to use RDR APIs for SMB shares and other remote file access. He was talking about which APIs to avoid using to avoid the extra roundtrip and how it was better to use the Handle based APIs and not path based APIs. It was a pretty darn good core OS level talk. It was good to see a talk that was not about just API’s or just about UX or just about the cloud/services but how to better program against Windows to make it more performant. As I was listening, I really felt that we in the Windows Security Team (team that I work in) could have had more such sessions targeting the same/similar audience that Mathew was targeting. Hats off to Mathew for trying. The fact that the few who attended, wanted to know more and had a lot of Q&A itself speaks to the content. They infact chased Mathew all the way to the Microsoft pavillion after the talk. A plea to organizers next time around is to ensure all sessions are added to the list in time and if we don’t, we should not bother adding them later – esp. with discoverability being close to impossible. They could have solved the discoverability by handing out to everyone (who walked into the keynote room on day1), the sheet with the updated PDC sessions.

Video & Slides of Talk: http://channel9.msdn.com/pdc2008/ES23/

I spent some time during the day playing with a one of the several  Surface Computers & Win7 Touch interface on the HP Touchsmart computers. Boy oh boy! If you wanted to see/play with the surface computer or TouchPC – this was the place to be. There were so many that many were just sitting there. I work at Microsoft and I have never seen one other than the one they showed in the Company meeting on stage. So it felt good to touch and play with it. I felt that the Win7 touch interface still had a lot of rough edges and needed refinement prior to shipping. It was not responsive enough and I had to do certain things twice/thrice for it to recognize my finger input. I could never get it to go back in IE. Forward worked. It was something that I was doing – but even so it needs to be intuitive. If I (a software engineer) cannot get to use it, my mom definitely cannot.