ETech – From the labs

Microsoft Research up first
Unconventional innovations and cross discipline serendipity. Rick Rashid.

Harnessing human scale storage. We are getting to the point that we can store all information about a person (anything they interact in a given day). Take an image of everything you see during the day for example. They’ve built devices and are using them with memory loss patients for example. Privacy studies–how do people interact with devices like this in the public is another example. These are wearable devices. Another unconventional thing. What happens when you take a conventional things and turn them into surface computing. Showed some video on interacting with virtual objects on a standard table that’s been turned into a input/output surface.

Head of Yahoo Research labsGary Flake pre-beta things. Showed Y!Q. Allows you to make contextual searches in context of what you were doing. They have about 30 scientists that work on things that fall in-between business unites. Work on machine learning, collective intelligence, scientific computing and text mining. They try to hedge their work (short term, mid, long term). Mentioned Yahoo Research Labs page. Introducing the Tech Buzz Game. To tap the collective wisdom of the Web. Get a single opinion, a single expert opinion, aggregating among non-experts, aggregation over population weighted by performance. Use markets as predictors (examples, Hollywood Stock Exchange for Box office , Iowa Electronic Markets for political questions). Allows people to manage a portfolio of things they know. Artificial market on questions related to emerging technology trends. Example. You guess what will drive search engine volume of OS X vs. OS X Tiger. How to time it? The prices of the two stocks reflect the aggregated opinion of which will dominate searches. By timing right, you can make fake money which allows you to invest it in other ways. You participate in collective prediction. Another announcement – Dynamic Pari-mutuel auction. Revolutionize auctions. Good talk.

Google up nowPeter Norvig, Director of Search quality
They treat all engineers as if they’re researchers. Want to improve the user interaction with a search engine–richer. First example is Google Suggest with auto completion. This was done by one engineer in his spare time during his day per week where he could work on anything he wants. Use XMLHTTP requests displayed in a DHTML layer. Also showed Google Maps. They break up the map into little images. When you drag the map, they redraw fast. Searches are done in iFrames. Showed Google personalization. Google Sets. Interestingly, not much new content from Norvig.