And that was dumb
It may have been dumb, but it was intentional. The clear strategy was to move away from the "legacy" Windows API towards something that, if not universal, was, supposedly, portable across a much wider range of technology.
And, as with all established platforms that run up against new technology, it turns out that if you insist on users moving away from what they already have, they are at least as likely to move away to a competitor.
And suddenly all you have left is legacy business. Away from the consumer sector, there's probably a fair bit of life left in the desktop market - it is indeed dumb to choke that off in the hope of competing with tablets and phones, but it presumably looked like a spectacularly good idea in the Powerpoint presentation to the board.