Microsoft research produces some amazing gems. They are a case study for what good things can happen when you let 2 or 3 really smart geeks works on problems which have yet to be solved.

Unfortunately, MS research is pretty disconnected from the product teams and as a result, they tend to innovate in a vacuum. As a result very rarely are their innovations integrated into the products. They therefore come off as a drain on the company.

Or some of the products are so advanced that the the product teams won't deal with the fruits of the research for many years.

I remember being excited about Microsoft's SenseCam just a little under 10 years ago. 4-5 years later I thought the product would never be available to consumers, and lo and behold, it's finally become technology that is available to the public... but man did that take a long time.

I did when I worked there. The return on investment relative to what the product groups were creating was off the charts. I was all for taking the money and talent to think about long term issues, but it was just distasteful how poorly value was captured from it.

It's no secret that the idea behind Microsoft Research is to keep the talent from working anywhere else. They don't have to ship a darned thing, as long as Google, Apple, or Sun don't either.

I got that impression during an interview with them in Cambridge in 2005. They genuinely couldn't promise that anything that was done was going to end up in a product or would actually be used.

Fortunately I was too crap to be considered a threat if I wasn't hired :)