Microsoft is famous for spending vast amounts of money on research and development (R&D), and journalists tend to conclude that it gets little or nothing in return. I usually take the view that if Microsoft can't be bothered to correct them, it's their problem not mine….
However, in a weak moment, I tweeted that the correct answer to "What does Microsoft get from its $9bn spending on R&D?" is: Windows, Windows Server, Microsoft Office (Word, Excel, PowerPoint, Access, Outlook, OneNote etc), Azure, SharePoint, Exchange, SQL Server, Lync, Microsoft Dynamics, Visual Studio, Expression, Xbox, Kinect, Windows Live, Xbox Live, MSN, Bing, Zune, WP7, Silverlight, Surface, Mediaroom, Windows Embedded, Windows Automotive, and a whole bunch of other things. (OK, I've extended the original beyond the 140-character limit.)
As you may know if you follow me on Twitter, I was trying (but failing) to prod the BBC's technology correspondent Rory Cellan-Jones into clarifying the introduction to his post from (I assume) Microsoft's annual TechFest in Seattle: Inside Microsoft: Innovation still on the menu. This begins:
"If you spent $9bn a year on research and development and employed 900 of the world's top computer scientists to come up with new ideas, what would you expect in return? More than a new way of playing video games, a cynic might say."
"I know, I know - I said R&D not just R," protested Rory, which is true. But how many non-techie BBC readers would know enough to say how much was R and how much D? It matters because the difference is significant.
As it happens, Microsoft's Steve Clayton posted something yesterday on the Official Microsoft Blog (which he assures me wasn't a response to my tweets): Small r, Big D: Microsoft Research TechFest Demonstrates the Future. Steve doesn't actually provide the figures I requested more than five years ago -- how much is R and how much is D -- but we can work out some ball-park numbers. He says:
"With 850 PhD-level researchers in Microsoft Research and around 40,000 developers in our product teams, that should give an indication of how we balance that $9 billion between research and the development of shipping products. I call it small r and big D."
If we assume all R&D staff get the same salary, on average, then each one costs $220,000 per year ($9bn divided by 40,850). That would put the cost of 850 researchers at $187 million a year, which is just 2% of the $9bn total.
In other words, roughly 98% of Microsoft's R&D spending pays for people to program things like Windows, Word, Excel and other products (see above). Only 2% is spent on research. This is not actually a high figure.
Whether Microsoft's researchers should produce more clever stuff is a different issue, but it's not what they are there for. Microsoft Research was set up by Rick Rashid, a professor of computer science from Carnegie Mellon University, and it's run more like a university research department. Staff are rated according to how many papers they publish.
I am familiar with this aspect. An ingenious and innovative British researcher -- a pioneer of PC sound cards, inventor of a SmartQuill pen that captured handwriting in 1997, of the wearable SenseCam, and with an important patent used in the iPhone -- was let go because she kept inventing things instead of writing papers. (See my interview with Lyndsay Williams, Ex-Microsoft star whose paper trail was too short, in The Guardian in 2007.)
It's perfectly reasonable to criticise Microsoft for its less-than-agile software development processes and very odd management systems: I think it could do more good work with fewer people. Feel free to criticise Microsoft Research for not creating more innovative products, though by its own founding criterion (peer-reviewed research papers published per year) it’s stunningly successful.
My simple point is that believing $9bn worth of Microsoft R&D spending only results in things like Surface and Kinect is completely and utterly wrong. It doesn't. It results in products that brought in $62bn last year.