Back when Microsoft first split Windows 8 in twain, it initially called the two divisions Windows 8 and Windows on ARM (WoA). Windows 8 would be the full-bodied x86 version of the operating system with every bell and whistle, and Windows on ARM would be a cut-down version designed for low-power ARM SoCs.
Most notably, WoA lacked the capability to run any Desktop apps that hadn’t been pre-approved by Microsoft (i.e. Office, Paintbrush, and that’s about it). Outwardly, Microsoft’s justification was that it wanted to ensure a smooth and reliable experience for people with WoA tablets; it didn’t want tablet users to be bogged down by toolbars and malware and other third-party Desktop software. Internally, though, I think that Microsoft was taking the first steps towards killing off the Desktop entirely, on ARM and x86.
The clues have been there right from the beginning. A few months after the Windows 8/Windows on ARM schism, WoA was renamed Windows RT. The RT designation has absolutely nothing to do with ARM; instead, it refers to the Windows Runtime (WinRT) that powers Windows 8’s Metro (new-style) interface and Windows Store Metro apps. So why did WoA get renamed? Because Windows RT is the trial run for a version of Windows that has no Desktop at all – a version of Windows that does away with the old Win32 runtime, leaving just the new WinRT-powered interface and its new Metro-style apps.
The below image shows Metro apps on the left, and Desktop apps on the right. Soon, the right-hand side will disappear.
Ultimately, if my suspicions are correct, ARM has nothing to do with this grand vision. Windows RT currently only supports ARM, but that’s more of a side effect than an end goal. Microsoft has a lot of experience with ARM devices (Windows Mobile, Windows CE), it saw how well the Android OEMs and Apple were doing with their cheap, low-power ARM tablets, and decided to jump on the bandwagon.
Microsoft has all but confirmed that Windows RT is derived from the Windows 8 codebase; the engineers have made some tweaks to optimise the low-level code for ARM, and cut out a few features to improve performance and reduce power consumption, but Windows RT is essentially Windows 8 compiled for the ARM architecture. In short, it would be very, very easy to build a version of Windows RT for x86 – a version of Windows that runs on Intel and AMD-powered tablets, but which doesn’t support Desktop apps.
This might sound utterly crazy, but hear me out. One of the biggest complaints about Windows RT (and its devices, such as the Surface) is the lack of functionality. Compared to a Windows 8 tablet with an x86 processor, Windows RT devices are useless. With no access to the billions of Desktop (Win32) apps, a sparse Windows Store, and half-baked stock apps from Microsoft, Windows RT pales in comparison to other tablets, such as the iPad. The other big grievance, ironically enough, is the confusing nature of the Desktop – as you stand in the aisle at Best Buy it looks like the real Desktop, but once you get it home you can only run Microsoft’s pre-installed Desktop programs.
In an ideal world, for both consumers and Microsoft alike, there wouldn’t be a Desktop at all on Windows tablets. This isn’t feasible until the Windows Store is fully populated and Microsoft finishes baking the stock Metro apps (News, Music, etc). If you weren’t aware, Windows 8.1 has a much better selection of stock apps, and a significantly updated Windows Store.
There’s still only around 100,000 apps (as opposed to 750,000+ for iOS), and lots of them are low quality, but the situation is improving quickly. Furthermore, let’s not forget that the only real reason that Windows RT actually has a Desktop in the first place is for Office RT – and Microsoft has confirmed that the Metro version of Office, codenamed Gemini, will be released in 2014. In a year or two, then, it’s not totally unreasonable to think that someone will pick up a Windows RT tablet and be satisfied enough with the Metro interface that the Desktop is no longer required.
The final hurdle: Performance
Beyond the lack of functionality and apps, the final problem that plagues the first generation of Windows RT devices, which are powered by Nvidia’s Tegra 3 SoC, is performance. Despite the tweaks that Microsoft made, the Surface RT and other Tegra-powered tablets are very sluggish, experiencing major slowdowns when launching, switching between, or multitasking apps.
Fortunately, this problem has a very simple solution: Use an x86 SoC, such as Intel’s upcoming Bay Trail. Intel’s internal metrics, backed up by some leaked benchmarks, suggest that Bay Trail will outstrip every ARM SoC on the market, while consuming comparable amounts of power.
This winter, Microsoft and its OEMs could easily produce Windows RT tablets powered by an x86 chip. Heck, even the updated Surface RT 2, which is due around the same time as Windows 8.1, could use an x86 processor. If x86 is faster and has the same battery life, why would Microsoft or its OEMs not use it? You might point to ARM’s lower price, but let’s not forget that a year ago, Intel’s then-mobile chief Mike Bell said that the company “would not lose a design win on price.”
But why would I ever want a Windows RT tablet?
At this point you’re probably wondering one thing: Why would you ever opt for Windows RT with its crippled feature set, over Windows 8.1 and its ability to run every Desktop program and game created in the last 10 years or so?
The answer is simple: On a touchscreen tablet, trying to prod your way around the Desktop and legacy apps that haven’t been updated for touch controls (i.e. most of them) is a lesson in frustrating futility. Tablet users really would prefer it if they could just stick to the Metro interface – which is why Microsoft focused heavily on the Windows 8.1 (and thus Windows RT 8.1) touchscreen experience. With the improved PC Settings (Metro Control Panel), better stock apps, a growing library of third-party apps, massively improved search, and better split-screening, the usefulness of the Desktop is rapidly declining.
Finally, here’s a chilling thought: If Windows RT is ported to x86, there’s a strong chance that the Desktop will be killed off completely. It might not happen with Windows 9 (or whatever Microsoft calls its future OS), but the writing is well and truly on the wall. Desktop and laptop PC sales have declined for five straight quarters, and they show no signs of improving. For now, this decline comes from prospective netbook and laptop buyers opting for a tablet instead – but as the technology, input methods, and range of apps improve, the bulk of the PC market will also make the switch.
The mouse-and-keyboard PC won’t just roll over and die, of course. There will still be some power users – gamers, sysadmins – who will try to hold onto the Windows Desktop, but it will be in vain. With the majority of the market using tablets and smartphones, the software industry will have no choice but to develop touch-oriented Metro-style apps. The Desktop will persist for as long as people install Windows XP, 7, and 8, but the lack of support from Microsoft and third-party developers will eventually force it into a marginalised, unhappy existence, probably assuming a position and role that’s comparable to modern-day Linux.