So what is RT doing to solve the problem of Win32?
Short answer: WinRT replaces Win32. Totally, wholly and completely. It cannot happen overnight in one release, for many obvious and not-so-obvious reasons.
Medium answer: WinRT cannot be compatible with Win32, for insurmountable but very complex reasons ultimately rooted in CPU architectures outside of MS control. To bring out a new but totally incompatible version of Windows would not only be daft, it would be suicide - and not just for Microsoft but for millions of users, developers, companies, and industry professionals. So the transition has to be careful, meticulous and gradual, without losing everyone along the way. So, in the transition from Win32 to WinRT, Windows has to do both, or it will die - and that unavoidably makes it a 'dual' system, because Win32 and WinRT cannot possibly call the same GUI. The only question is: will the transitional 'dual' system have one or two windowing GUIs - one for Win32 and one for WinRT? The two can never meet in one, and to know
why you need to grasp arcane technical details about historical and current CPU and memory architectures. So, the introductory interim transitional choice is to keep Desktop/Win32 and expose WinRT initially only in a "full-screen Start screen replacement" - it's easier to explain than to justify two separate windowing GUIs were a windowing WinRT GUI to also ship (Help Desk would be tied up with "why can't I run my new (WinRT) program on the same screen as my (old Win32) desktop, huh Microsoft???!!" - the explanation is just too complex and so ends up sounding like bulldust. Of course it isn't, as system people know, but how do you explain it to 1.3 billion users?
Cerebus, the whole discussion is not helped by Microsoft's equivocal use of "RT".
Excuse another long post, but there's just so much one has to explain by way of background ... it's hard to know where to begin. In the final analysis there is no way of understanding
why Windows 8 is like it is without understanding the whole history of why we are where we are. So apols if I say too much - it's hard to know what people already know.
First it needs to be said that Microsoft's strategy for many years has been to position the name "Windows" as a franchise in its own right and not tied to any particular technology. Look under the covers and Windows 3.x, Windows 95/98/ME, Windows NT/2000/XP, Windows CE/Mobile, etc are very different. The particular GUI visual design and even the interaction elements are pretty superficial and easily changed ... it'll take a junior programmer only an afternoon to make anything
look like a particular Windows. The "engines" are very different under the hood even though they're all called "Windows".
At its highest level (the visual bits), the GUI is just the "dashboard" and "steering wheel" that the operator uses to interact with the system, and changing the appearance and interaction method it only a design challenge not a technical one.
Basically, the word "RT" refers to two totally distinct things:
WinRT and
Windows RT. They are not the same, though related in certain ways.
a) WinRT is the new API set and framework for "future" applications on Windows. All Windows, on every platform, from server through desktop to mobile. This is the bit that's strategically significant. It is architecturally elegant, modularised and layered in a way that can serve not just current but also future and yet-undeveloped hardware architectures, comms systems, file systems, structured and unstructured data, etc, etc. And because it's not tied to specific hardware architectures, and carries none of the baggage of the past, it can be implemented across systems. This is what supersedes Win32.
b) Windows RT is also the marketing name for the first version of Windows on ARM, using a Metro full-screen GUI specifically designed for tablet/mobile because those platforms are suited to fullscreen "immersive" apps.
Important points:
1) WinRT is utterly GUI agnostic. It so happens that the initial way it's exposed is in full-screen Metro, but that's just a marketing decision, mainly to take advantage of the tablet/mobile opportunity to get the framework to developers and apps kickstarted.
2) As I said before, Win32 is hangover from antique hardware architectures and is instantiated in successive Window releases for one reason only: app compatibility. It cannot be dispensed with overnight. This necessarily means our mainstream desktop/office/productivity systems must support both Wi32 for legacy, and also WinRT is WinRT is to ever get to enable the new capabilities required in the future (where Win32 cannot go).
3) Because of the elegance and flexibility of the new WinRT API set, it can support any GUI type, even 3D holographic GUIs (in the lab already). But part of the problem with Win32 is the way it's tied to the current Desktop. To make WinRT call the same Desktop as current Win32 apps do (ie all our Windows apps today) would cripple or hobble WinRT from the get-go, and bring forward many of the problems of Win32 and defeat thus defeat the whole point and purpose of the new architecture.
4) Because any new version of Windows today must also run Win32 apps (anything else would be unthinkable and certain suicide for Microsoft), there is only once possible choice: the new Windows must support both Win32 in its own environment (the Desktop metaphor/GUI as we know it till now), and also WinRT apps. There is no way on earth to make a common UI for both - this limitation lies in Win32, old processor architectures, and basic CPU physics. Whichever way you look at it, the new release of Windows has to have two utterly discrete and separate 'desktops/GUIs/interfaces' - that is a given unalterable fact. The next choice is: so what do these two 'desktops' look like, and how does the user interact with two separate 'worlds', for separate they must unavoidably be? In this first release of WinRT, Microsoft decided to not expose the windowing/3D/holographic GUI calls but to make it full-screen, because its first large-scale use would be on tablet and mobiles, so you can bring WinRT to market quickly, educate developers in WinRT dev, and drive app development initially for tablet and mobile. But even though the first version's target platform is tablets and mobiles (which are by nature fullscreen for obvious reasons), the REAL and MUCH MORE STRATEGIC OBJECTIVE is to inject WinRT APIs into the ecosystem, and then leverage it backwards onto real power desktops and servers. When/if it goes mainstream on those platforms, and legacy Win32 apps are no longer so important, then you can make the WinRT GUI anything you want it to be, including multi-windowing, multi-monitor, 3D, holographic, whatever.
All of this is too complex to explain to the market in any coherent way. There's just too much scope for confusion.
Does this give you an inkling?