John Robb asks:

Has anyone seen anything that came out of Microsoft R&D that was worth a darn?  Could anybody (Rob) provide some pointers into what $5 b a year in R&D is creating?? 

Just thinking through the math. $5 billion / year buys and provisions 33,000 engineers at $150K/year. If you go through the MS product family, there are at a few dozen projects (e.g. SQL Server) that could consume 1000 bodies each (80 ten-man teams plus overhead).

In most business, Research is long term (2+ product life cycles) investigation, and Development is current and next life cycle design and construction.

Software could sure use more pure research; experimentation on very high risk (high failure rates) and very high payback projects. Most grad schools have been coopted by industry to the point that most CS grad students and professors work on applied research. Does that mean the bulk of pure research will move to China and India?

Should government subsidize more pure computer science research?