Straight talking

Tim Anderson discusses the implications of Microsoft’s bold move towards ‘cloud plus device’.

Published: 01/11/2011 | Last Revision: 15/12/2011

Microsoft BUILD was the company’s most interesting conference for years. Windows 8 is not only a new version of Windows, but brings a radically different user interface and embraces a different model of computing based on cloud plus touch-controlled devices. It also promises to be a safe environment in which applications are sandboxed and communicate with each other only through limited and controllable contracts, while getting their data from cloud services. If Microsoft manages to shift its customers to this new form of Windows it will be an extraordinary achievement. The question though is whether this revolutionary approach will work. Metro looks nice; but we have work to do, and for most of Microsoft’s customers that means running desktop Windows applications. Fortunately Windows 8 also supports traditional desktop Windows. But if we will be spending all our time in desktop Windows, then what is the point of Metro?

The answer is that Microsoft is trying to break out of a tight spot. Windows is imprisoned by its legacy; it will never be truly secure or truly touch-friendly because it was designed before either of those things mattered. Apple’s iOS has shown that there is an alternative, based on a locked-down operating system, a curated app store, touch control that works, and overall design excellence. Windows is slowly but surely losing market share to the iPad and other tablets, and the primary intent of Windows 8 is to recover that lost ground. At the same time, everything that is in Windows 7 is still there in Windows 8, with the possible exception of the old Start menu. If you start up a desktop app, it runs in the desktop environment. There are even two versions of Internet Explorer 10, a metro-style version which runs full-screen and in which plug-ins such as Flash and even Silverlight are disallowed, and a desktop version which runs in a resizable window just like IE9. By delivering both Metro and desktop Windows in one operating system, Microsoft hopes to keep faith with its legacy while also transitioning to a new model that could in time dominate client computing. The obvious difficulty is that if Windows 8 machines run full-fat Windows alongside Metro, then users will need suitably powerful hardware as well as keyboard and pen or mouse in order to operate desktop Windows applications. That will make Windows 8 machines expensive: the Samsung tablet handed out to conference delegates, complete with Bluetooth keyboard, would likely cost over £1,000. And what is the point of making Metro secure when users can easily pop into desktop Windows and bypass all its protection?

Microsoft's iPad alternative

The answer (though this was not clearly articulated at BUILD) is that Windows 8 on ARM, not on x86, will be Microsoft’s iPad competitor. ARM systems on a chip (SoCs) are the industry standard for tablets and enable low power use and relatively high performance. Windows Division president Steven Sinofsky said in his keynote that “the demos that we’re showing you today are equally at home on ARM and on X86.” Despite that, there will be important differences. The first is that existing Windows applications will need to be recompiled to run on ARM. The second is that Microsoft may be intending to lock down Windows on ARM to a greater extent than on x86. There were no BUILD sessions on recompiling for ARM, and Microsoft is not encouraging this; we even heard that the Windows 8 online Store might be the only way to install apps on ARM, and that they will all be Metro apps. If Microsoft follows through with this, then Windows 8 ARM tablets will be price-competitive with Android and iPad, and will not require keyboard and mouse or pen, because none of the old applications which require this will be present. However this also implies that the only reason to buy a Windows 8 ARM tablet will be to run Metro apps, of which none currently exist beyond a few samples. This means that Microsoft will be keen to promote Metro app development to populate the store for the launch, and of course we heard plenty about Metro-style development at BUILD. Every Windows 8 machine will have Metro, ensuring the ubiquity of the platform. One point of particular uncertainty is what will happen to Microsoft Office on ARM. It seems implausible that Office will be ported to Metro in time, though we may see efforts to make the product more touch-friendly in the desktop environment. Incidentally, desktop Windows does exist on ARM: we saw this in the ARM samples on display. Metro is not yet a complete operating system and access to desktop Windows is necessary for some tasks, such as access to the full control panel. It would not surprise me though if Microsoft has in mind to remove desktop Windows from a future version running on ARM. Personally I am looking forward to Windows 8 tablets. They will solve the tricky problem business travellers face: do you pack the iPad with its convenience, instant-on and long battery life, or a Windows laptop with the applications you need, or both? If Microsoft do it right, a Windows 8 tablet will be the ideal combination, though I will be looking for x86 and its compatibility rather than ARM. So it is obvious that Microsoft’s new direction is risky. Metro will be a hard sell to businesses, many of whom will look at Windows 8, see little change in the part of Windows that they actually use, and stick with Windows 7. It took Microsoft ten years to displace Windows XP, and I foresee an equally long life for Windows 7. As for the consumer market, even if its OEM partners deliver attractive Windows 8 tablets at a competitive price, Microsoft will not find it easy to displace iPad or Android. Much depends on the quality of the Metro apps in the store when Windows 8 tablets start appearing in, I presume, autumn 2012. That in turn depends on developers, which is why Microsoft was happy to give them shiny new machines at BUILD. Whether Windows 8 Metro succeeds or fails is uncertain, but the one thing beyond doubt is that client computing is changing radically and that Windows 7 is in one sense the last of its line.