Windows 8: Maybe Microsoft Should Give Mango a Go?

[Author's Note: I actually wrote this article a couple of months ago, intending to publish it this month (February 2012), so I was interested when Microsoft posted the article "Building Windows for the ARM processor architecture" a couple of weeks ago. Because the Microsoft article sort of directly addresses my thoughts in this article, I thought, "Hey, maybe I actually predicted the future this time!" But unfortunately that didn't turn out to be the case, and in fact the reality seems kinda grim.]

I realize I'm not the first person to have this "amazing" insight, but after working with the "pre-beta" version of Windows 8 and seeing the mini-demo of a Windows 8 tablet at the CES keynote, I've got to ask: Why on earth isn't Microsoft offering a version of the Windows Phone OS on low-cost, low-power tablets? I mean, isn't it the obvious, clearly lower-risk, more far-sighted answer to just roll out tablets running Mango?

[Author's Note: This was my thought two months ago, and I still think it's a good idea. What Microsoft did, however, was the exact opposite. Instead of exploiting the power of the company's nifty new phone OS to make its Windows tablets cheap, small, and low-power, Microsoft decided to exploit the power of the Windows desktop OS to make the company's phones expensive, large, and battery-sucking. Read on.]

In case the combination of "Mango," "Windows," and "tablet" don't all compute in a familiar way, let me back up a bit.

In a nutshell, I (and, I think, everyone else watching Windows 8) see the Windows 8 desktop's reason for existence summed up in one word: iPad. The main and perhaps only reason that we're even talking about Windows 8 in a world where most folks are quite happy to still use Windows XP (a decision I'm not defending, but that I do observe every day) is because Microsoft has to respond to the iPad's threat to Windows desktops. Because the iPad is reasonably light, offers a good-sized and bright screen, runs for up to eight hours on a charge, and costs in the $500-$800 range, any return volley from Microsoft must meet or beat the iPad in those categories. (Sorry, Samsung. I love your 1,366-pixel-wide tablet, but it's a trifle messy to carry with me compared to the iPad, and its battery life is better than a laptop but a mite short to compete in the "tablet I live with" category. There's definitely a market for a high-power, big-RAM, wide-screen, with-a-stand-it's-a-laptop-replacement tablet -- but the market for Windows-running-iPad competitors is, I think, significantly larger.)

Beyond its size, battery life, and cost, the iPad is attractive because it does a great job with multi-touch. Every currently available version of Windows, however, doesn't really get multi-touch, with one exception: the latest version of Windows CE, "Windows Phone," also known by its erstwhile beta name, "Mango."

And then there's the application base. The iPad has more apps available for it than YouTube has videos of toddlers doing cute things, so Windows 8 tablet buyers will similarly expect to see a large application base. The good news on that score is that Windows 8 tablets like the Samsung I mentioned will run just about any Windows application, and there are -- ahem --  more than a few Windows applications. But the fact that fingers aren't mice means that running traditional Windows apps can be something of a pain on a multi-touch tablet, and that's one reason why Windows 8 supports a completely new kind of app that's touch-centric, a "Metro" application. Unfortunately, there aren't too many Metro apps yet, but that'll probably change by the time Windows 8 RTMs. In fact, Microsoft is so sure that people won't care that their iPad-ish Windows tablet doesn't run standard Win32/.NET apps that the company is building a special version of Windows 8 that can't run non-Metro Windows apps at all.

The thought here is that although a Metro-only tablet wouldn't run Win32 apps, it would be able to run on hardware cheap enough and skinny enough to deliver great battery life. It's not a crazy idea, as iPhones and iPads can't run Mac software and they're doing fine, and although the Mango phones haven't exactly conquered the market, I've never talked to any Mango user who cared that Windows Phone couldn't run full Windows apps, or even apps from the earlier Windows Mobile/Windows CE operating systems.

To create this Metro-only Windows 8 platform, Microsoft is creating a version of Windows 8 that runs atop a processor platform that Windows hasn't made much use of thus far -- the Advanced RISC Machine or ARM processor. You probably don't own a PC that uses it, but if you've used a PDA or a smartphone in the past decade and a half, there's a decent chance that it ran on an ARM processor. ARMs are a designer's joy in that they're well-understood (they've been around since the mid-1980s) and aimed specifically at low-power controller and small-device computing uses. Getting all of Windows on it would be tough because you're basically limited to a half-gigabyte of RAM for code and, well, how many of you would outfit a Windows 7 box with just a half-gig of RAM? (That's based on the memory layout of the current ARMv7 processors. Something bigger is on the way, but, again, it won't be nearly as fast as a standard Pentium/AMD64, and, again, a slower-than-Pentium chip is exactly what we want for our Metro-only Windows 8 tablet, because a slow processor is a power-sipping processor. So for ARM boxes, Mango's a good fit, and regular old Windows is a classic case of trying to fit 20 pounds into a 5-pound bag.)

Although I'm impressed that Microsoft would tackle shoehorning a subset of the modern "give me 64 bits of space" Windows into a sub-32-bit architected chip, it makes me wonder what that'll mean for future versions of Windows. If Windows 8 is a tight fit for ARM, what of Windows 8.5, 9, 10, whatever? No matter how you slice it, the ARM-friendly version of Windows will have to support the same WinRT as the full-blown Windows, and if ARM Windows is cramped in its first iteration . . . eek.

 Yes, I'd love to be able to buy an inexpensive, long-battery-life system running a Windows-ish operating system, but I'm not sure that putting parts of Win32 and WinRT on ARM is the way to do it. The whole thing sounds a little like an automobile manufacturer delivering a small, stylish, option-replete car that gets 130 miles to the gallon -- but that requires sperm whale oil to operate. It's interesting, but not really practical, and it doesn't promise a bright future. And heck, maybe not spending a lot of effort on building a limited-value port to ARM would be a smart use of resources, since we've already seen a couple of Windows ports that netted Microsoft very little -- and here I'm thinking of the implementation of NT on MIPS and PowerPC chips.

If ARM tablets are a neat-sounding but not all that promising option for cheap tablets, is there another way? Sure -- which brings us back to my original point. We already have a Microsoft operating system that can't run Win32 stuff but can run Metro-ish applications, and even better, it already works -- Windows Phone. I hadn't yet had a chance to play with a Mango phone, but I got my hands on one recently and I have to say that it's pretty darned impressive. It runs what's essentially a Metro interface on a very low power device, and it does so quickly and fluidly. (More fluidly, in fact, than does my Samsung tablet, but again I'm only working with a pre-beta, and in fairness Metro will get some serious tuning and tweaking before Windows 8 releases -- but it's still quite impressive how snappy Mango runs on a little device.)

Every time I talk to someone "in the know" about this seemingly obvious strategy, I always get the same disturbing answer -- a knowing smile, a rueful shake of the head, and something like "Windows Phone isn't under the Windows Desktop division." Decoded, this means that the guys running Windows Desktop know that if Microsoft comes out with a Mango-based tablet OS that runs on inexpensive tablets, then an awful lot of people who currently own laptops running Windows 7 will switch to Mango tablets, bruising the egos and perhaps year-end bonuses of the folks running the Windows Desktop division. This seems kind of crazy in that it should be very, very clear that (1) the iPad has the juice to take a serious bite out of Windows -- and by serious I mean "great white shark bite"; (2) such a bite would be very bad for Microsoft in general; (3) Microsoft needs an iPad competitor, and an inexpensive one; and (4) Mango already performs wonderfully on lower-power CPUs (and probably runs fine on ARM right now, for all I know). I mean, I understand that large organizations have internal turf wars and such, but are the players here really so short-sighted as to say, "If I can't win, then I don't want the rest of the company to win"? Stay tuned for the answer -- I know I can't wait to see what happens!