Does Windows RT have a place in the sun?

There have been a lot of articles/posts lately questioning the future of Microsoft’s Windows RT.  My biggest problem with the anti-Windows RT thinkers is that they take such a short-term focus.  Windows RT, just as the Windows RunTime (WinRT) it is named for, are long-term initiatives.  We are currently looking at V1 of those and people are trying to project that Windows RT (aka Windows on ARM for its most obvious origin) will die, apparently before even reaching its first birthday.  While Microsoft could have a change of direction, I’d bet that the path they started on is very much intact.

Later this year we’ll have V2 of  both WinRT and Windows RT.  Next year we’ll likely see V3.  And as we all know, it typically takes Microsoft (and most others actually) three releases to get something right (or more precisely, to cover the landscape sufficiently for most people to feel comfortable it covers their needs).  I think Microsoft will give WinRT and Windows RT the full three releases before considering if they’ve gone in the wrong direction.

I’ve written about this topic a number of times over almost a year and a half.  Back in December of 2011 I wrote about why Microsoft was doing ARM support and in January of this year I pointed out that Windows RT was not specifically about ARM.  Windows RT is about producing a legacy-free version of Windows for the future.  And in between these two I gave some history of the Windows 8 effort that further explains the origins of the WinRT and Windows RT.  Numerous other posts offer additional insights.

Let me summarize all this.  The Win32 application model is broken and unfixable.  Microsoft has been looking to replace it for over a decade.  The first version of that replacement, the Windows RunTime or WinRT, was released as part of Windows 8.  To scope the effort the version in the first release focuses on what was necessary to address the tablet market, though it is not tablet specific.  The name Windows 8 is used to mean versions of Windows capable of running both legacy Win32 apps and new WinRT apps.  The name Windows RT is used to mean a version of Windows only capable of running WinRT apps.

In the long run WinRT expands to cover far more of the application space currently addressed by Win32, and Windows RT becomes the mainstream version of Windows.  Windows RT and Windows “8” eventually switch roles, with the non-RT version of Windows becoming the niche offering for those who need legacy Win32 capabilities.  Does this happen in Windows “9”, Windows “10”, Windows “11”, or Windows “18”?  Who knows  But it will happen.

Critics of Windows RT, particularly those who claim it has no role in life, are too busy looking in the rear-view mirror.  They are commenting on V1, not thinking about where it will be by V3.  But it is the latter that really matters.

Could Microsoft change direction and run to some other alternative to WinRT and Windows RT?  Sure.  Will they?  I doubt it, not at least without seeing the path they are on through another 2-4 years.  If at that point the “re-imagination” of Windows has failed we’ll have to see a re-imagination of Microsoft as a whole, probably away from the broad client-computing realm.  They don’t want to go there, and the key to avoiding that fate is in making Windows RT succeed.



This entry was posted in Computer and Internet, Microsoft, Windows and tagged , , . Bookmark the permalink.

35 Responses to Does Windows RT have a place in the sun?

  1. Ed Ferron says:

    Well written…

  2. The short sited-ness of some supposed “big names” in technology writing is astounding actually. It is clear to me what MS is trying to accomplish here, and I think you are right on target. Not a day goes by that I am not ridding a non-tech savvy consumer of rows of toolbars and self installed Malware. I am a little uncomfortable with the app store model being in charge of all the apps, but security does matter. Let the blame for insecure apps fall on the app stores shoulders, not the consumer. I’m all for it, for now.

  3. dafowler says:

    I think part of the issue for RT, and I would say the issue many tech writers have with it is the lack of support for x86. It has been surprising to see the tech press’s reaction especially when early on many complained about having desktop applications on tablets to begin with. Another issue is price; since Windows RT isn’t “real” Windows the consensus is it needs to be priced the same as the Nexus 7 or Kindle Fire even though both are smaller tablets. I don’t think any of these death of RT articles would’ve been written if RT devices were sold at a loss. It all just goes to show the catch-22 nature of the consumer market; you either have the number of apps required to meet the minimum standards or you have the number but they aren’t designed for the form factor like Android tablet, or you don’t have the right ones in enough quantities like Windows 8. The criteria by which they are judged is arbitrary and always shifting.

    • Great point, but it probably has more to do with the delicate nature of the OEM partnerships than Microsoft’s ability to price them cheaper. I think we can all agree that this device would have sold a LOT more if it were $200 cheaper. I don’t think MS felt they had that option, nor do I think they felt it had to be done cheaper or they would lose the game. I think they had a lot of bugs to work out in the beginning, and doing it this way allowed them to run a sort of “beta test” among early adopters, they were thinking much more long term to next fall. IMO. To their credit, a LOT of bugs have been worked out, and from the looks of Windows Blue screenshots leaked this morning, their will be some really nice interface adjustments as well.

  4. BennS says:

    This discussion would take a very different path if the library of “Modern” apps for Windows RT were significantly larger, and with higher quality. At the point when all the “important” apps are on Windows RT, with good quality, Windows RT will achieve its place in the market.

    Unlike the iPad, which could run all, or nearly all, iPhone apps, Windows RT has a fairly small app library, consisting of whatever apps Microsoft shipped with the system, and the first versions of Modern apps. All software takes a while to mature, and that’s exactly what has to happen for Modern apps and Windows RT, as Hal pointed out. By the time V3 rolls around, the Modern app library will have matured, and the OS itself will have a much broader API set to support even more innovative new apps.

    Until then, predictions of Windows RT’s early death will continue. It will be interesting to see what initiatives Microsoft undertakes to try to accelerate the maturation process as well as improving the Modern app library quality and breadth.

  5. Shaun says:

    I own and develop software for Windows RT. Hal, I’ll be completely honest with you — I laughed when I read the comment “Windows RT is about producing a legacy-free version of Windows for the future.” While I don’t question that this was Microsoft’s goal for Windows RT, my experience is that they are not executing on this goal. If you navigate through the \Windows directory you’ll find hundreds of executables and Dlls that were introduced in previous Windows versions. It’s an archaeological dig of files that no longer serve a purpose and it raises the question — why was so much of this legacy brought forward into the RT world if they were attempting a clean break from the past?

    With regard to usability. the Windows RT hardware still cannot keep up with my typing in OneNote (yes, even after patches are applied!). I own a more powerful Samsung Slate with full Windows 8 (the developer slate). Two first party Metro apps (Mail and IE) now crash every time on load, and I can no longer access them. It’s an otherwise vanilla machine and I wonder if an update broke these two critically important applications.

    And now that gets us to the topic of Marketplace updates and usable disk space. I discovered (and shared here on this blog) the revelation that each Marketplace App update is saved as a *complete directory copy* on the file system. For example, The MS Travel Metro app is 400MB. It has been upgraded three times now and so there are four folders on the disk which occupy 1.6GB of space. Wow! If you’ve been following along with the public outcry over available disk space, this could be a problem that gets worse.

    • halberenson says:

      In order to start removing legacy you first have to eliminate all the dependencies on it. Today that isn’t possible in any real sense. First, you still have key apps (Office) that are Win32 on Windows RT. They need to move completely to WinRT. Then you have lots of utilities and tools that Microsoft did not target to move yet. Once those are eliminated they can start looking to remove legacy code that nothing is relying on.

      Your other points have nothing to do with strategy. They may or may not be applicable to most users. I can’t out-type either the Metro OneNote or OnenOte 2013 on a Surface RT. I run Windows 8 on 4 PCs besides my Surface RT system and don’t have problems with Mail or IE. But in context of the strategic discussion any issues are minor and short-term. I take no issue with someone writing a blog post saying that Windows 8/RT shows the immaturities of a V1 system as I agree with them (if not necessarily each specific criticism). I take lots of issue with turning that into clueless strategic claims.

      • stonstad says:

        Hal, I greatly appreciate your perspective and I’ve learned a great deal by following your blog. I also agree with you that stability issues don’t necessarily figure into high level strategy. Nonetheless, I’ll say this at the expense of sounding like a troll — offering a rock solid product without perceived stability issues is an important strategy when engaging the consumer market.

      • Bob - former DECie says:

        Hal, according to , VC++ developers get access to some of the win32 API and the new super-COM. ( I was going to call it COM+, but we know that name was used years ago). This gives VC++ developers the ability to do some things that .Net developers can’t, but more importantly, by exposing pieces of the win32 API and super-COM, will Microsoft ever be able to eliminate the dependencies upon “legacy windows”?

        • halberenson says:

          First let me say that Peter’s article is awesome! I don’t think I caught it when he first published it, or more likely (since it seems familiar) I got interrupted while reading it and never got back to finish it.

          There are three separate points I think need to be addressed:

          – WinRT is layered on Win32 in many cases. I don’t think this necessarily restricts Microsoft’s actions as it is more of a development expediency (Win8 was planned to have only 18 weeks of coding, but I think they added a milestone mid-project so that they had 24) than an architectural statement. However, to the extent they let Win32isms through to the new API it does limit their future action.

          – Some things that are disallowed are technically possible and apps will use them. “So sad, too bad” as a friend would say. Microsoft will be completely within its rights to break apps that violate its guidelines. And I predict they will.

          – Some use of Win32 is allowed, particularly DirectX. Ok, this is an area that will put limits on Microsoft’s ability to remove legacy. I don’t know how widespread the impact will be. That’s both because even a narrow exposure of classic Win32 can lead to a tremendous number of dependencies and because it is unclear how many apps will use those features. If the latter is small then Microsoft can eventually deprecate the use of Win32 completely (after providing an alternative and giving apps a migration window). More importantly, one wonders if Microsoft thought carefully enough about what they exposed to new apps so that they know it is things they are willing to preserve going forward. I’m hoping that they had clear thinking about how to evolve even with that limited Win32 usage, but they certainly could have messed it up.

          • waldtaube says:

            think it’s best to distinguish the application model from the APIs here – they primarily seem to have cared about isolation, sandboxing, process lifetime management, and deterministic install/uninstall. since all of that, along with the new UI, meant that developers would have to rework their apps anyway they treated it as a chance to opportunistically clean up the APIs a bit as well (mostly by deprecating GDI and USER – the new windowing system currently depends on parts of USER but I don’t think the USER semantics leak into the APIs) – but that wasn’t really the focus of the overall effort. the WinRT APIs themselves were less about getting rid of cruft or creating some kind of architecturally elegant clean break, and more about creating an API surface that would be more approachable for developers new to Windows, work well with their preferred language/runtime, and steer them towards doing the right thing by default (but not really force it – for example the famous async-only APIs can still be trivially wrapped into sync APIs).

      • A runtime is an API. The tried and tested way of moving platforms is by deprecating old APIs and introducing new/alternate ones. Microsoft is a master at this. This is why we old timers have hung around for so long. The social contract between a platform and its inhabitants lives and breathes around this.

        The problem is not engineering. It is business model migration. How does Microsoft charge for an incremental alteration to the software? How does Microsoft move from open to closed? The only solution is to create a brand new platform.

        However, there is simply little appeal when good alternatives exist. Little regard is paid to Microsoft’s strategy of coming from behind like Windows, Word. Instead, Windows RT reminds me of Lotus Improv. Competent but nada traction.

    • Shaun, it sounds like you’ve created other user accounts on the system which haven’t installed the app update. Thus they still require the old version to stick around until that happens.

  6. Do you think WinRT will ever expand to support “non-modern UI” apps (with separate, overlapping windows, etc.)? That was the first thing I thought of when I read, “In the long run WinRT expands to cover far more of the application space currently addressed by Win32”.

    From the point of view of a user/consumer though, buying into Windows RT today just doesn’t make sense. Mainly this is because Intel’s Clover Trail platform is very competitive, with better performance than the ARM platforms, comparable battery life and price, a wider range of devices, and of course, compatibility with Windows apps.

    Just because Microsoft wants a WinRT-only OS to “force the issue” for developers, doesn’t mean that consumers should have to suffer the consequences

    • halberenson says:

      I don’t know if we’ll ever see a return of overlapping windows specifically just as we never brought every feature of an IBM MVS mainframe into DEC’s VMS or everything the two of them did into Windows. At some point you provide alternative solutions, not identical solutions. There are a class of workstation apps and usage patterns that may not move to “Modern” apps for a decade or more because the WIMP UI paradigm is so intertwined with them. But that is a much smaller set of situations than most critics accept.

      • dafowler says:

        I think a better thing to think about is something along like WebOS’s card view, which provides a multi-app view; it allows for multiple applications to be seen and run but not in the traditional overlapping way

  7. Jeroen Frijters says:

    The Windows RT naysayers apparently don’t remember the early days of Windows. Microsoft does.

    I think their heads will explode when Microsoft releases Windows RT for Intel. Which is quite likely to happen with Blue already given Intels Bay Trail platform coming at the end of the year.

  8. Philip Schmitt says:

    It’s not shocking people don’t recommend getting a Windows RT product now, since most of the devices are half-assed and overpriced, but this is more due to the current hardware and OEMs rather than about the future of Windows RT.

    I think you understate the importance of ARM. Right now there are a lot more Android devices than Windows ones and OEMs seemingly put a lot more effort (and marketing) into their Android stuff. This is where I see a lot of potential for Windows RT. If Windows RT easily ran on all Android devices, OEMs could simply offer both an Android and a Windows version of the same devices (with the Windows version maybe being $50 more expensive). With attractive, similarly priced Windows tablets, consumers could really choose the OS and not just the hardware (i.e. cheap devices). So which one to choose? “this version comes with Office and supports printers, usb sticks, cameras….”. Shouldn’t be that hard to sell.

    I also disagree with the notion that Windows RT should have come without the desktop all along, which various articles bring up again and again. Even without x86 applications, the desktop sets Windows apart from the competition. There are lots of schools and other institutions that have adopted iPads. I’d think especially schools would value the inclusion of the desktop. Students could take the tablets home, plug in an external display, mouse, keyboard, and then use the full Office next to a web browser to write papers and stuff (and then maybe even print it).

  9. Eric Hill says:

    If it is really Microsoft’s intention that Windows RT will become Windows, the glaring flaw in that theory is: Who is going to spend the trillions of dollars to re-write all the Win32 software that currently depend on for their business processes? Even if people deem it worthwhile, it’s a 5 to 10 year process for that to happen. In the meantime, Win32 seems to be in mothballs with no meaningful enhancements being made, which makes it very difficult for us luddites still building and selling Win32 software to keep it modern.

    Microsoft has kept its market share despite often not having technically superior products because every version of Windows was backward-compatible with the previous version, so folks could upgrade and keep all their existing software. That kept businesses from seriously considering moving to a different platform. Windows RT moves away from that model. With Windows 8 for Intel allowing folks to put their toes in the Windows RT water without losing anything, that reduces the backward-compatibility issue, but it also keeps people attached to Win32.

    I was around for Windows 1.0, and Windows RT is far better than Windows 1.0, but Microsoft does not have as much time with Windows RT as they had with Windows get it to maturity, because we are in the midst of a paradigm shift and the competition is several orders of magnitude better now.

    • halberenson says:

      Look to the history of both Mainframes and Minicomputers for clues. Look to the switch to Java and .NET for the last 10 years of server apps. Look at the plethora of technologies being used to write Cloud and Web apps, few of which have any relationship to Win32. Look to the last 10 years of client app investment, which is all about the web, and more recently smartphones and tablets, for confirmation. Investment in Win32 apps flatlined over a decade ago. Lots of people are living on that legacy, but nearly all new development has been elsewhere.

      • Frans Bouma says:

        > Investment in Win32 apps flatlined over a decade ago. Lots of people are living on that legacy, but nearly all new development has been elsewhere.

        I find that hard to believe, considering e.g. .NET apps also also not being WinRT compatible without a rewrite. For LoB apps in .NET, win32 / desktop is very much a reality, WinRT will require a rewrite.

        For apps which have been moved to a webclient + server-side service/webapp, what ties that to WinRT exactly? One could use any device, heck, even a chrome book, to work with those. If that’s the model of future apps, there’s nothing stopping people to move away to other OS-es.

        So MS needs people to stay on Windows specifically, because Windows provides something all the other platforms don’t. Today that’s Win32 (directly, or indirectly through .NET). To make people become tied to the future windows, they’ll have to be tied to WinRT, nothing else, so the apps have to run on WinRT. Current apps, all those gazillion pieces of software out there currently requiring Win32 to run, don’t do that. This thus requires rewrites. While that makes sense in many cases, it’s not a given that people will automatically rewrite their stuff for WinRT: why would they? Because MS says it’s the future? Today there’s no need, and only if MS simply kicks out the desktop completely, there never will be. (as there’s a desktop, so win32, so no incentive to rewrite).

        Sidenote: about the naysayers: Microsoft has said so many times in the past that a given tech was the future. Case in point: ‘.NET’, or more recently, Silverlight, WPF. .NET is ‘still alive’, considering you can use it to make WinRT apps, but is there really much energy put into making .NET a first class citizen on WinRT from WinDiv? I don’t really see that energy, to be honest. WinRT is a new runtime, like .NET was too. As .NET isn’t the runtime of the future (seriously, why on earth not? It’s a VM… ), why should people believe Microsoft now all of a sudden when MS says ‘WinRT is the future’?

      • Eric Hill says:

        Yes, a lot of classes of software have moved to the web, but I work for a company that develops Win32 software that is growing sales 15-20% a year. My neighborhood bought software for swim team automation this year, the same software that virtually all of the 50+ neighborhoods in the league use for their summer swim teams. It’s Win32 software. Anybody want to do their Visual Studio or Eclipse development on WinRT? Is AutoCAD moving to a tablet OS?
        Yes, look at mainframes. Much has moved off of mainframes, but mainframes are still used for tasks that couldn’t be moved off of them, and mainframe operating systems continue to be improved and upgraded.
        I understand the trends, but the notion that it’s time to completely mothball the desktop, which Microsoft seems to be doing, is just bizarre. If there is a company that should be interested in the desktop living on as long as possible, it is Microsoft. Yet they seem to be eager to kill it off long before an adequate alternative exists.

        • halberenson says:

          As I’ve said multiple times this is a 5-10 year evolution, and some things will linger on beyond 10 years. But that doesn’t mean they’ll be in the forefront of activity.

          Oh, and I suggest you visit

          • Eric the Red says:

            Okay, AutoCAD has a browser interface. Is it the primary interface that designers/drafters use? Is AutoCAD phasing out their destop/workstation applications and going whole-hog to browser/iOS/Android?
            The real point, though, is that if we can agree it is a 5-10 year evolution, can we also agree that Microsoft needs to continue investing in the desktop platform for at least another few releases, if for no other reason than to add development frameworks that make development for the desktop more similar to development for Windows RT/Windows Phone? Like, for example, a XAML+C++ native code UI framework? I know Windows 8 was a special case, but I got the impression that Sinofsky felt like Microsoft’s work on the desktop was over. That makes me not sad he’s gone, but it’s anybody’s guess what Julie Larson-Green things, and Lord knows, Microsoft is absolutely not going to tell us what they have in mind going forward, making it impossible for us to plan. Why should I invest in Windows RT when I invested in WPF and WPF is now in mothballs? Fool me once, shame on you, . . . .

            • halberenson says:

              Many AutoCAD users are still using Windows XP and most of the rest only recently migrated to Windows 7. This is true for virtually the entire class of users who have specific need of the desktop paradigm. No matter what Microsoft did it was not going to see a significant migration of those users to a post Windows 7 OS until at least 2015. So there really has been no point in focusing Windows 8 or Windows Blue on that market. Even Windows 9 may arrive before organizations are over the fatigue from the migration from Windows XP to Windows 7. So it was far more important to focus on areas where a paradigm shift is, or will soon be, underway.

              Will Microsoft come up with a smoother migration environment or just let the existing desktop environment stagnate? I don’t really know. If they went to Autodesk and asked them if they’d rewrite it using a desktop XAML+C++ native environment I think they’d get laughed at. Ditto for all the big ISVs, especially if those ISVs produce multi-platform products. They’d get told no one is interested in focusing on a rewrite and any new product efforts are mostly focused on expansion of the potential customer base (read: mobile and tablets). And they’d be told if they ever do a major rewrite of the core product it would be to better support a major change in paradigm. Autodesk doesn’t need anything to keep designers/drafters happy. To keep their entire ecosystem happy what is important is supporting scenarios such as

              Now why should you invest in the Windows RunTime? Customers. Have you read

              • Bob - Former Decie says:

                Hal, at this point in time I see two huge issues for Windows RT:
                1) As others have mentioned, if one wants to cover Windows RT, Windows 8, and Windows Phone 8, you almost, but not quite,have to write 3 completely different versions of an application. This will change over time, but who knows how long that will take, 2 years, 4 years, ???
                2) The cost of deploying a Windows RT application. Unless you are a large Microsoft customer with all the tools for remote management of your devices, the cost to side load an application can exceed $100/device and that’s not a one time fee. If an employee leaves and the company reassigns the device to another employee, it’s another $100 fee. No small to medium company is going to incur those costs for a tablet. Who knows when or if Microsoft will fix this issue?

                • halberenson says:

                  From an app perspective, why would you need to write separate apps for Windows 8 and Windows RFT? As long as you are sticking with WinRT apps they should be identical. WP remains an issue for now.

                  Windows RT’s initial target was purely consumer. They gave a nod to enterprises, but just a nod. That will change with time.

  10. The reality, Hal, is your description of Microsoft’s plans for WinRT strikes me as competition-agnostic. It is like single-player chess.

    RT doesn’t enjoy reach. Windows XP systems still abound. The tablet market is dominated by iOS and to a lesser extent Android. The only reason people have been buying Windows isn’t just because they can run their older software on it. MS has had effective monopoly of mindshare on what people commonly think of as operating systems. Every non-MS product sold in the shifts that perception. Don’t look at raw market share alone. What matters is the velocity of adoption of these non-MS products.

    Not many enterprises can afford to tool up and develop expertise in a platform that they don’t believe will enjoy longetivity. Looking around, iOS and HTML is the best, while Android doesn’t have a business case, and Microsoft and Blackberry are on the watch list. While Windows remains a strong franchise, and Microsoft is hoping RT hitches a ride on Windows coat tails before it is too late, it is evident that RT is not getting traction despite it being technically competent.

    • Tim says:

      Chui – I assume you’re referring strictly to mobile devices/OS’s when you lump MS with Blackberry.

      As far as enterprises tooling up, I don’t think WinRT requires that big of a leap for existing .NET developers. But the issues I see playing out for the company in which I work are (A) slow adoption in general and (B) a trend toward mobile platforms that enable write-once, run anywhere (e.g. IBM Worklight, SAP Unwired, OpenText Wave, etc) without the need for specialization in any one native platform. So even if I delivered a great POC for WinRT, I’m afraid management wouldn’t see the value because they view WinRT as being ONLY about mobile…which goes back to point B above.

  11. dregourd says:

    Today Microsoft makes the buzz and generates trillions of comments in the blogs because people just can not read the Microsoft strategy: it is an enigma. We are all imagining scenarios, strategies, secret plans, just to explain the nonsense that comes from our (formerly) favorite Sorftware Editor.
    And if… and if these systems brought to light in panic were simply accidental?
    RT? A subproject of Windows based on a bet on ARM that began few years ago, remained secret, then had to be monetized and used just to issue the first Surface hardware in time.
    METRO (modern)? A set of graphical tools and researchs started at the Zune epoca, also tenths of millions expensive, and which also had to be monetized.
    SURFACE? An old project name, very good but not used, that had to be monetized.
    WINDOWS? A war treasure, an iconic word that is used to identify the company, and to glue together all these diverse items.
    As a bottom line, my theory is that Microsoft, strangled by bad figures and by a feroceous competition throws all its old secret projects in the battle without any strategy. It is only reaction, defense, and it produces a choas that could be fatal. Inovation has desappeared. MS run after the competition.
    Like Napoleon, MS once had an Austerlitz, but is now experiencing a Waterloo.

  12. Pingback: stratēchery | The Week in Review — March 25-29, 2013

  13. Bob - former DECie says:

    For some reason, once the replies get to a certain depth, I can’t reply to specific replies, so this is in response to your “From an app perspective, why would you need to write separate apps for Windows 8 and Windows RFT? As long as you are sticking with WinRT apps they should be identical. WP remains an issue for now.”

    Check out:


Comments are closed.