Post-defenestration Microsoft: It's the APIs, stupid. And Metro
Sinofsky, Microsoft's Caligula
Analysis "Tiles to the right of them, Tiles to left of them, Tiles in front of them"
- Alfred Tennyson, The Charge of the Metro Brigade  (1854)
The sudden departure of Steve Sinofsky from Microsoft leaves Redmond with its biggest crisis for years - and it needs to assure investors as a matter of urgency. He's achieved a huge amount of change, but he's also left a real mess, the full extent of which isn't appreciated by financial or technology sector analysts.
Sinofsky was an energetic and capable executive, delivering real results, and a breath of fresh air after a "wasted decade". During the Noughties Microsoft's feuding factions and spiralling bureaucracy meant years were spent just trying to ship Vista. Sinofsky's biggest achievement was putting the development processes back on track. Then, he directed them on a long, strange journey.
A lucky general - at least until his luck ran out - Sinofsky benefitted from the WinMin project, a years-long mission to cut accumulated bloat from Windows NT. The idea was that mobile devices, desktops and servers could eventually run on the same kernel. MinWin started in 2003, the benefits could be felt in Windows 7, and now Microsoft has mobiles running on the NT kernel. That's a great achievement.
But Sinofsky also had a meglomaniacal qualities - as can be measured by the number of senior executives  who clashed with him and left. Microsoft lost its Xbox veterans and Ray Ozzie in 2010, giving Sinofsky huge control over product development. Metro unbelievers were shunted aside, in fact Metro became a dogma. It had begun to resemble a Maoist cell - and once Sinofsky had the power to kill products in rival divisions there were few to stand up to him. Not that many people knew what Sinofsky was actually up to: he survived as long as Ballmer believed in his radical strategy.
The logic behind it appeared to be coherent. If Microsoft couldn't establish itself in tablets and smartphones, went the argument, it would lose the consumer market entirely. Apple has notched up 100m iPad sales and people spend less time with their PCs. Relegated to the enterprise space, Microsoft feared it would become another Wang. Microsoft needed an ecosystem and Sinofsky would brute-force one into existence. A new unified API would be the means to do it: developers would write a Metro app and it would run across a range of Microsoft devices. So although Microsoft was non-existent in tablets and barely visible in smartphones, developers would be obliged to make it their third choice.
What is not widely understood today is how much the Metro strategy has failed in its primary goal - to create a unified API for applications across desktops, smartphones and tablets. Instead of 'WORA' (Write Once Run Anywhere) or even WORAM (Write Once Run Anywhere Microsoftish), there are fragmented APIs requiring separate code bases for Metro widgets on desktops, smartphones and tablets. They should be compatible. They're not.
In addition, the Surface tablet adventure has burned Microsoft's relationships with OEMs and channel partners. It has created a unique and much praised piece of hardware - but Microsoft still heavily depends on Dell, Asus, Acer and HP to shift boxes and these are its bread and butter. Surface looks like the ultimate vanity project - like Caligula planning to make his horse a consul: "a combination of all the gods and to be worshipped as one". Meanwhile, the OEMs were left with a new version of Windows that they're extremely reluctant to sell - entirely a consequence of Sinofsky's dogmatism.
This is so much the case, in fact, that the Microsoft salesforce - desperate to sell something, anything - is recommending Windows 7 in an attempt to get customers off XP .
Windows 8 Phone's move to the new kernel is the great achievement of this, but it still doesn't run Windows RT, the ARM-tablet flavour of Windows, and so Sinofsky has created a new incompatibility for Microsoft to bridge. A year ago Microsoft said, "Wait till you see what we'll deliver in a year". But Microsoft's phone partners will have been astonished, and disappointed, to find that Windows 8 Phone's userland looks almost exactly like Windows 7 Phone's userland. While the pace of development on iOS and Android is furious, Microsoft has wasted a year's development.
Now Ballmer has to pick up the pieces - and it's hard to see how he can do so without driving a stake through the Metro strategy. Making Windows 8 sellable again, and dodging what we predicted back in March  would be a missed enterprise upgrade cycle, must be Ballmer's first priority. The second is giving Metro what everyone thinks it has, but it doesn't, a unified API. And third is persuading Microsoft's OEMs and channel that they're partners - rather than collateral damage in a war to make Metro ubiquitous. ®