You can have taste without being able to program well, but you can't program well without good taste.

The original point was that MS had smart people so a lack of smarts couldn't be the problem. The salient points are 1) smarts don't guarantee programming skill, and 2) smarts don't guarantee taste.

> You can have taste without being able to program well, but you can't program well without good taste.

Fair enough. In that sense, then, I wouldn't say that programmers at Microsoft lack taste by definition.

There are things at Microsoft that have demonstrated solid taste: the original NT kernel comes to mind, as do early iterations of Word and Excel (starting back when they were primarily developed for Mac). LINQ. WCF. The DLR. There are also smaller pieces of tools that seem elegant, even when there are issues with the larger design. Active Patterns in F#. Attributes in C#. The ADO.Net CommandBuilder and DataAdapter.

I'm not convinced that it's that everyone at Microsoft is individually defective as much as that their methods of collaboration, like those of most large corporations, usually yield results less effective than good individuals or small teams could have produced with sufficient autonomy.

"I'm not convinced that it's that everyone at Microsoft is individually defective as much as that their methods of collaboration, like those of most large corporations, usually yield results less effective than good individuals or small teams could have produced with sufficient autonomy."

Which brings this thread from the religious fervour of "what it means to be smart" right back to the subject at hand: Why MSFT would pay M$20+ for Xobni rather than write a kock-off in-house.