Rick Rashid, head of research at Microsoft, was at the Future in Review conference talking about the precipitous drop in students majoring in science and engineering. He attributes the drop to misperceptions that engineering disciplines are not cool and that technology creates more problems than it solves. "We need to get the message out to kids that one of the ways to solve the world's problems and make better future is through engineering, science and computer science. We need to bring the romance back into the field," Rashid said. Nerds are cool...
On a different note, he predicted that, within the next five years, the field of software development would be revolutionized by mathematical modeling that can prove properties of very large bodies of code. He said the Windows team is hiring PhD. mathematical modelers to test code, hoping to replace the dance of software programmers coding and testers trying to show how dumb the programmers are.
He also said that we are at the threshold of "human scale storage." All of the relevant information about a person generated in a lifetime can be carried around everywhere they go. A terabyte of disk costs $600, Rashid said, and it could store every conversation you have in lifetime or you could take a picture every minute of life and keep it on terabyte disk.
Microsoft researchers are also focusing on datamining. An answer program takes questions and then looks on the Web for articles that might have the answer and statistically mines them to get a more accurate answer. "You can ask it some crazy questions and get some interesting answers. During a demo I asked 'What is the meaning of life' and it give series of possible answers," Rashid said. "One of the answers was 'questions.' " That's a rather profound statement for a computer.
Related to datamining and statistical analysis, Rashid said that doing statistical analysis on data without revealing private data is a significant mathematical problem. For example, he said that 83 percent of Americans could be uniquely identified with just three data points-- birth date, sex and zip code.