What if Windows 10 isn’t the Future of Windows? (Premium)

It’s been a tumultuous year for Windows. But what if Windows 10 has little to do with the future of Windows? What if Windows 10 failing was, if not a plan, then at least something that Microsoft feels is acceptable so that it could move more quickly to the next thing?

My worry is that this next thing won’t really be Windows. And the evidence certainly supports this worry, since Microsoft has been openly testing and shipping various platforms that are not Windows for decades.

Looked at in the context of history, Microsoft began developing something called Windows—which at the time was a graphical shell that ran on top of MS-DOS—in the early 1980s. The first (terrible) version shipped in 1985, and the first truly usable version shipped in 1990. But when Windows surged in popularity in the early 1990s, it became the dominant, if not monopoly, force in the PC industry.

Despite this, Microsoft knew it needed a more sophisticated computing architecture to succeed in a rapidly evolving market. After a few fits and starts with UNIX and OS/2, it brought in Dave Cutler and a small team from Digital to adapt a new system into a multi-platform, network-capable UNIX killer called NT, for New Technology. The first version, released in 1993, looked and worked like the DOS-based Windows versions that were just becoming popular at the time. But NT had a modular architecture that allowed it to support multiple OS “personalities” that could run side-by-side. Originally, Microsoft expected OS/2 and POSIX to contend with Windows, though that plan quickly went by the wayside. (That said, the resulting architecture is what later led to x86 app emulation in Windows 10 on ARM.)

NT replaced the DOS-based versions of Windows, and became Windows, starting with Windows XP in 2001. Something that was not Windows … became Windows. It wasn’t just called Windows. It was Windows.

It took NT only 8 years to make that transition. And yet here we are, almost 18 years later, still using the same NT architecture in Windows today.

Yes, Windows (NT) has improved dramatically over those years. Microsoft has done a lot of work to componentize the system, for example, to ensure that it can scale from very small devices up to servers and datacenters and special-use hardware like HoloLens and Surface Hub. It has been enhanced almost endlessly with new functionality. And after Microsoft let its cross-platform capabilities fall by the wayside, Windows (NT) is once again being actively developed across multiple hardware architectures, including ARM.

But it’s still NT. It’s still a relatively ancient codebase, parts of which date back to the late 1980s. All of the improvements since then have added complexity. And as our usage of Windows has evolved, so too have the demands on this mountain of legacy code. One can almost hear it creaking under the demands of Windows as a Service (WaaS), an aggressive servicing plan that its original architects could never have imagined or wanted.

That Windows 10 works at all today is a miracle of software engineering for which Microsoft gets little credit. But alongside this work have been separate but related efforts—like Windows RT, Windows 10 S/S Mode, and Windows 10 on ARM—to pull another NT. To make a thing called Windows that is not really Windows.

There other side-trips along the way, of course. Cairo, an attempt to adapt Windows to the then-trendy world of object-oriented computing. Longhorn, Microsoft’s first attempt at fully componentizing the NT codebase and adapting it to a broader range of system types. And of course Windows 8, which correctly predicted a future of touch interfaces and non-traditional PC form factors but sacrificed the past and the current in a mad race to knee-cap Apple and take control of the future.

The problem for Microsoft, as is so often the case, is cultural. Windows has withstood so many attacks over the years, from the Mac to IBM and OS/2 to the Internet tidal wave to Linux, that its durability has become a key part of the Microsoft myth. Bill Gates has referred to Windows repeatedly as the most versatile software that’s ever been made. And while that claim is long past its fresh date—witness the Linux-based Azure Sphere as the latest evidence—that belief still pervades at Microsoft.

When the software giant told me about its plans for what was then called Windows 10 Cloud in early 2017, I pointed out several flaws and asked what the point was. Just to make some alternative to the Chromebook? No, I was told. This wasn’t a Chromebook replacement. Instead, Windows 10 Cloud—which became Windows 10 S—was a new attempt by the by the Windows team to create a new future for the platform.

But Windows 10 S, like Windows RT before it, suffered from a key mistake that the original NT team had addressed over 20 years earlier. Windows 10 S was a version of Windows … that could not run Windows applications. And I think this is the key thing that the Windows team has missed again and again in its recent weird attempts to push Windows past Win32 and the other desktop technologies that make Windows what it is. It’s been trying to make something that is not Windows. And never will be.

There are lots of ways in which we could define what Windows is. But I think it’s simple: Windows is the thing that can run all Windows applications. Something that just runs new Store apps is not Windows. It’s something else. Something lesser.

When Windows NT debuted in 1993, it could run the applications that had been developed for the DOS-based versions of Windows that preceded it. And by the time that Windows NT took over as Windows with Windows XP, Microsoft had evolved its application platform into Win32, ensuring that it ran on both Windows 95 (which was DOS-based) and its successors as well as NT. This eased the transition for users: Their applications worked and they could move forward.

Yes, the applications platform has evolved in recent years along with Windows. There are Store apps that run on both real Windows and Windows in S mode (as Windows 10 S is now called). These Store apps include Universal Windows Platform (UWP) applications and games, but also Progressive Web Apps (PWAs) and Desktop Bridge apps, which are really specially-packaged desktop applications that are sandboxed and containerized.

I’ve always found that last bit to be particularly interesting. Microsoft’s Desktop Bridge solution is in many ways the latest in a long line of app virtualization and container technologies that it has developed over the years. And it points to a possible future for a new Windows that really isWindows. A Windows that has a leaner and more modern architecture and yet can still run desktop applications. And do so securely and reliably.

I’m curious why Microsoft has not pursued this in any meaningful way. Windows Weekly listeners may be interested in an impromptu discussion along these lines that occurs in the first hour of the most recent episode. But the short version is that I talked myself into the notion that since Windows is no longer the core concern for Microsoft, it should adapt its Satya Nadella era strategy of embracing open systems to include Windows. Where would it make sense for Microsoft to use open technologies in Windows?

On Windows Weekly, I foisted Linux as an off-the-cuff possibility. The idea there is that Linux is developed out in the open and Microsoft could simply build a distribution that looked and worked like Windows 10. The trick, of course, would be adapting Linux to run the broad range of Windows applications out there. Not just Win32, but also .NET and all its permuations, and Store apps and all its permutations.

That might be impossible. I’ve used virtualization and emulation solutions on the Mac and Linux many times and they can be inefficient and feature incomplete compared to running native code.

But what if it were possible? After all, if you accept that Windows is simply the ability to run all Windows applications, then a Linux that could run Windows applications would meet the definition. It would meet the need in a way that, say, Windows RT or S mode never did.

No, there’s no evidence that Microsoft is working on anything like this. That’s the cold water moment. I’m just thinking out loud here.

But there is a ton of evidence that the best and brightest at Microsoft are no longer working on Windows 10: The many reliability problems we’ve seen over the past year demonstrate that very nicely. And it’s likely that, from an OS/platform perspective, it’s because they’ve moved onto something else.

For some, that something else includes further componentizing Windows and adapting it for new use cases, with the old One Core strategy having now evolved into something called Windows Core.

There are also rumors of something called Windows Lean, which to me looks like Windows 10 S version 3.0, or yet another way to make a version of Windows that does not run Windows desktop applications.

It’s also possible that Windows 10 Core and Windows 10 Lean are really the same thing, too. Yet another lighter-weight version of Windows which, again, cannot run Windows desktop applications.

This stuff will excite some enthusiasts and there will be much discussion about how it could enable nonsense new hardware form factors like Andromeda. All it really is, is a way for Microsoft to make something like Chrome OS while forfeiting everything that’s important about Windows. In other words, Windows RT Windows 10 Cloud all over again.

This is depressing. And I don’t have any answers.

What I do have is an understanding that Microsoft has been flitting around the edge of what really needs to happen with Windows while somehow managing to avoid it all together. I don’t think that Windows as we know it is sustainable for the long-term anymore. And that Microsoft needs to figure out a better way to run all Windows applications on a modern platform. Not a different way to not run them.

And if Microsoft is going to ship Windows 10 Core or Lean or whatever it’s called, maybe it needs to finally stop using the Windows name and let this thing float or sink on its own. But for the love of God, Microsoft, figure out some way to a future in which there is a real Windows too. Not just yet another hobbled thing that is not really Windows.