Slow Down Windows 10 Development, You Say? (Premium)

I’ve argued for years that “Windows as a Service”—(WaaS), or what I think of as “Windows masquerading as a service” or, better still, “Windows as a disservice”—is untenable. And thanks to the passage of time, I now have the evidence to back it up: After failing to delight users with big new features in 2016-2017, Microsoft moved to shore up quality and fit and finish in 2018 and still fell flat on its face: Windows 10 versions 1803 and 1809 were the buggiest-ever releases of this system, leading to major public embarrassments for the software giant.

Surely, then, Microsoft would learn its lesson. And maybe, just maybe, we would move to a more logical and manageable system in which Windows 10 was only updated once a year. And in doing so, Microsoft could finally respond to the needs of its customers, none of whom—business and individuals alike—are interested in installing major new Windows versions twice every single year.

Hilarious.

Instead, Microsoft, still reeling from its fumbling of the 1809 release, has decided to speed things up and further complicate Windows 10 development. As of this week, there are now three Windows 10 versions in active development: Windows 10 19H1, which will arrive as Windows 10 version 1903 sometime in the next few months, Windows 10 19H2, which is set for the second half of 2019, and now Windows 10 20H1, which, as the name implies, is scheduled for an early 2020 release over one year from now.

There’s no Windows team today per se. But the communications we’ve received recently from what I’ll call the Windows team troublingly echo those we used to receive from the Sinofsky-era Windows team, which never felt the need to justify its actions let alone explain what it was doing and why.

We’ve seen this behavior twice in the past week alone.

Last Friday, Microsoft released a new build of 19H1 that includes undefined and unexplained “technologies [that are] tailor-made for gaming.” Ignoring the fact that Windows has included “technologies tailor-made for gaming” since 1995 and WinG, the predecessor of DirectX, Microsoft offered no explanation at all for what those technologies are. They just offered a free game to play, temporarily and only to the lucky few who signed-up quickly, andit  will apparently use telemetry data and overt Insider feedback to determine how well it works. Whatever it is.

And then there’s this week. Yesterday, Microsoft released another new build of Windows 10 to the Insider Preview. But this time, the build targets Windows 10 20H1, a release Microsoft has never once mentioned publicly, either to Insiders or otherwise. That it is the third active Windows 10 version being actively developed at the time of this writing is both unprecedented and unwelcome. But Microsoft’s “explanation” for this release, its rationale for needing to do such a thing, is completely missing in action.

“Some things we are working on in 20H1 require a longer lead time,” is all Microsoft says about this situation in its announcement. That’s it. That’s what we got.

Remind you of anyone? Anyone with the initials S.S.?

I thought we had put this kind of secrecy behind us when Mr. Sinofsky was shown the door, unceremoniously, in 2012, with Windows 8 and RT blowing up in Microsoft’s face and his secret plans to usurp the CEO position from Steve Ballmer exposed. In the wake of Sinofsky’s departure, Windows was handed to what Sinofsky had previously labeled “the B team,” the group running Windows Phone that was led by Terry Myerson. Myerson’s first course of action? Tear down the walls of secrecy and develop Windows 10 in the open.

Myerson didn’t get everything right, and he was certainly tasked with some terrible goals, including, yes, that WaaS nightmare that we’re still dealing with. But the one thing he got very right was to reverse all of the terrible decisions that Sinofsky had made about developing Windows in secret. Windows 10 would always be a known quantity. This is comforting to individuals, for sure, and desirable for enthusiasts. But it is most important for Microsoft’s biggest customers, the enterprises that deploy and manage Windows at scale. The world was right again.

But this week has shaken my confidence in the Windows team—which, again, doesn’t really exist—to new lows. Coming as it does in the wake of the 1908 debacle, this kind of overt secrecy, combined with an actual acceleration of Windows 10 development, is ludicrous. It’s the type of thing no one would ever predict or even imagine, and if it had come up in some alcohol-fueled debate about how things might go, it would be laughed at and quickly forgotten.

And yet here we are.

I know. Some will argue that Microsoft has pretty much always juggled multiple future Windows 10 versions at the same time, and that releasing code to the public this way doesn’t represent a change to the underlying process. But that’s not true. Actively developing three future Windows versions with the public is, as noted, unprecedented. And it exacerbates a problem that Windows 10 was supposed to fix: It simply had too many Windows versions to support, and if the world would just adopt one Windows version, en masse, we’d all be safer and more secure as a result. And this would be easier on Microsoft because it would have fewer Windows versions to target with security and quality updates.

This simplistic view of the WaaS future ignores a simple fact: There are now more supported Windows versions out in the world than at any time in history. And while two of them—Windows 7 and Windows 8.1, both of which will be supported for years to come in some capacity—predate Windows 10, several Windows 10 versions—which, again, are all different versions of Windows—are being supported as well. Microsoft showed itself to be incapable of reliably deploying a single Windows 10 version in 2018. Now it’s working on three new versions, all of which will land in the next year or so.

If you don’t find this troubling, you’re not paying attention. A year ago, before the 1803 and 1809 release debacles, I argued that Microsoft should pay attention to what Apple was doing, and slow down the addition of new features in Windows 10 and focus on quality. And while Microsoft’s 2018 releases did, in effect, slow down the addition of new features, they conversely arrived with major quality problems. That’s not a success, it’s a failure.

So the reaction this year should be to finish the job and to improve fit and finish while delivering reliable, high-quality updates. Actually, updates is the wrong word: Microsoft should simply release one new version of Windows 10 this year, perhaps mid-year, before moving on to whatever nonsense it plans for 2020.

And that’s the most troubling bit, really. 2020. In its one note of explanation of 20H1, Microsoft noted that “some things [it is] working on in 20H1 require a longer lead time.” This tells me that 20H1 will be a major, major update, with fundamental changes to Windows 10. Which is exactly what Windows 10 does not need.

Perhaps we’ll get clearer communications from Windows at some point. Perhaps we’ll learn more at Build. And perhaps I’m making a mountain out of a molehill and I should learn to just trust those at Microsoft who have been entrusted with improving and releasing Windows. Perhaps. But I have too many years behind me and too much experience with this kind of thing to see anything positive in these developments. This is not exciting. It’s ridiculous.