Also, if you're referring to Anders Hejlsberg, being the lead architect of C# doesn't qualify as "sit on their thumbs". Considering that .NET is one of the 2 competitive advantages that Windows still has in the corporate environment (the other being Exchange), I think it was a million dollars well spent.
http://news.cnet.com/2100-1023-279561.html
"Yocam maintains that Microsoft is luring personnel away with huge signing bonuses, some in excess of $1 million."
On a brighter note, Borland may have lost but silicon valley ultimately won when Google's incredibly fast growth prevented Microsoft from pulling a Borland on Google's search engine team.
Microsoft Labs has been among the more phenomenally unproductive corporate R&D operations of all times. The amount of talent that was sucked in there ... never to be heard from again, boggles the mind.
I'm not convince that this was a deliberate strategy, but it seems reasonably plausible.
Why not pay them to build better development tools rather than to sit on their thumbs?