Ballmer (and Microsoft) still doesn’t get the iPad

"The operating system is called Windows," claimed Steve Ballmer when asked about Microsoft's plans for the tablet/slate/pad form factor at the company's annual Financial Analyst Meeting on Thursday. He expressed dismay at the iPad's strong sales figures, "[Apple has] sold certainly more than I'd like them to have sold," he said. Ballmer then promised that Windows-powered devices will be shipping "as soon as they are ready," going on to explain that they would get a boost from Intel's low-power Oak Trail platform next year.

The message was clear: Microsoft still doesn't understand why its Tablet PC concept has repeatedly bombed over the best part of a decade. Apple sold more iPads in its first three months of availability than PC vendors sold Tablet PCs in the whole of last year; in fact, the number of iPads sold in that period is likely to eclipse the number of Tablet PCs sold both last year and this. But still the company is persevering: stick a regular PC operating system on a laptop, give it a touchscreen, and then take away the keyboard and pixel-perfect pointing device. Ballmer even reiterated the company's position: slates are just another PC form factor.

The iPad is a neat package. It's not a device for everyone. There are lots of things the iPad doesn't do well; there are many things the iPad doesn't do at all. But it's not trying to be these things; it's a conveniently sized, highly portable, long-lasting media-consumption device. It's ideal for browsing the Internet, reading e-mail (with the occasional short reply), looking at photos, playing music and videos, and casual gaming. It doesn't need much in the way of configuration. It doesn't run Mac software. Every single piece of software on it is designed to be used with fingers. In no way is the iPad striving to be a PC, but it is because of this—because it's not running software designed for keyboards and pixel-perfect pointers, because it's running software that's simple and restricted, because it uses a slow, but low-power, ARM processor—because of these things that it is so good at the things it does do.

Tablet PCs, on the other hand, have had all the size and weight of conventional laptops, with all the software of regular laptops, but without the human interface devices to make them useful. They contained the compromises of the iPad—touchscreens are never going to be as good for text entry as physical keyboards, touchscreens, even with styluses, are never going to be as precise as mice—but without any of its benefits, including the light weight, impressive battery life, and purpose-built software. They made sense in some vertical markets, but as mass-market devices, they've consistently failed.

And so it is set to continue because, as Ballmer said, the operating system is called Windows.

The situation on the hardware side will improve; the inexorable march of technology, especially now that Intel is targeting these form factors with the Atom line, will allow Tablet PCs to rival the iPad's size and weight. The iPad will still have an advantage in the near term—Intel can't yet match the power consumption of Cortex A8 processors such as the iPad's A4—but that advantage is diminishing. It won't be this year, it might not even be next year, but a time will come that PC-compatible hardware is no longer a penalty in this kind of machine.

The PC-style business model—commodity hardware running a Microsoft operating system—may also have some legs in this market. We argued recently that this wasn't a good fit for smartphones; the restrictions on hardware and software customization give little scope for PC OEM value-adds, and the high margins in that market mean that Microsoft is missing out on a lot of the revenue.

The same may be true of the tablet market, in which case Microsoft should probably produce its own iPad equivalent, but Intel's involvement means that commoditization is likely to be swift. If these devices are to run Windows, there may also be a case for allowing the full range and flexibility of PC hardware. In any case, the wheels are already in motion and PC hardware is becoming more suitable for tablets, even without Microsoft's direct involvement.

Let your fingers do the walking

But the software side—Microsoft's responsibility—is the real problem. Windows is not designed for fingers. Microsoft lauded Windows 7's touch capabilities—the operating system has built-in support for multitouch displays and touch gestures—but that's nowhere near adequate.

Why isn't it adequate? Well, a good place to start would be Microsoft's own Windows Phone User Interface Design and Interaction Guide. This is instructive because it tells us what Microsoft is thinking about finger-based interfaces. One key detail tucked away in there is that any touch target should be at least 9mm×9mm (typically about 34 pixels square) to ensure that it's big enough for a finger to hit. In exceptional cases, they can be a little smaller; 7mm square, about 26 pixels square. This guideline is a simple one, but it's important. With UI elements smaller than this, the user interface simply isn't convenient to use with fingers.

Windows 7, however, is chock full of elements smaller than 9mm. Taskbar buttons and Start menu entries are big enough, but they're exceptional. Icons in the notification area, title bars, toolbars, scrollbars, checkboxes, radio buttons, everywhere you look, there are pieces of user interface that are too small. Not just things that are too small; there are many effects that depend on hovering, there is item selection, there are abundant context menus, there is extensive use of keyboard shortcuts. And though the operating system does try to help with some of them—for example, opening a Jump list with a flick instead of a right click gives the items wider spacing so that they're easier to hit with a fingertip, and Explorer can optionally show a checkbox next to each item to allow selections to be made by clicking—it doesn't solve every problem. It doesn't even try.

And that's before one even considers third-party software, where the same problems are repeated over and over again.

Windows, and Windows applications, are not designed for tablets. Windows is designed for a pixel-perfect pointer (be it mouse, trackpad, or TrackPoint), and a keyboard. Band-aid fixes to give a thin veneer of touch-friendliness are not good or deep enough. To make Windows a touch operating system requires every single aspect of its user interface to be overhauled. The same is true of every application.

This is a change that Microsoft is making in its phone software. After years of trying to shoehorn desktop interface concepts into its smartphone platform, Microsoft has finally seen sense, thrown away every aspect of the old interface, and started from scratch. Much to the dismay of those who wrote Windows Mobile software, the company is forcing the same change on third-party applications too.

It's a monumental undertaking, but it's the only option for a touch device. Yes, it means a break with the past, but as the Tablet PC and Windows Mobile experience should have demonstrated beyond any doubt, it's the only way.

Windows can still be part of the answer

This does not mean that Microsoft cannot use Windows 7. Behind the scenes, the operating system could indeed be Windows, especially when x86 hardware catches up with ARM-powered form factors. But it cannot look like Windows or act like Windows. It cannot require the user to go into Explorer, or Task Manager, or Control Panel, for essential tasks. It probably should not allow access to those things at all (though a good case could be made for "unlocking" access to the traditional UI when connected to traditional interface hardware). Windows would just be there for basic operating system functionality: hardware support, protected memory, multitasking, networking, and so on. The front-end, and every application, would have to be all-new.

Microsoft does have a slight precedent here. Its Media Center software can work in a similar mode—Windows PCs can be configured to boot directly into the Media Center front-end, bypassing the conventional desktop and going straight into a user interface designed for a remote control. This is by no means perfect: the Media Center interface is not fully featured, and some aspects of system configuration and maintenance need a conventional mouse and keyboard. But it does give a hint of one way in which Windows would actually work as a touch operating system.

Whether this is worthwhile is hard to say; a similar front-end built on top of Windows Embedded CE would achieve the same but with lower hardware requirements. The big value of desktop Windows—compatibility with Windows software—is not so useful in the tablet form factor, since that software isn't tablet-friendly anyway. However, Microsoft is quite adamant that it wants real Windows to run on these devices, so it would be a solution of sorts. Having the full power of a PC may also make such devices more versatile, and the ability to revert to the standard Windows UI when equipped with suitable hardware could also be appealing.

The big risk of such a Windows-based approach is that there would always be a temptation to be lazy, and just rely on the existing Windows interface for some aspect of the device's functionality.

Regardless of whether Microsoft uses Windows Embedded CE or Windows 7 as the base platform, the company also needs to write high quality software for it. The old Tablet PC approach might have gained more traction—at least in corporate markets—if Microsoft had ensured that, at the very least, a tablet-friendly version of Office existed. Though OneNote was compelling on tablets, the remainder of the suite was not. Insiders suggest that this was deliberate; although the Windows group wanted the Office group to produce a tablet-suitable suite, the Office group saw no point, and was unwilling to go out of its way to help out the Windows group.

As for what the interface itself should be, that doesn't really matter. A Windows Phone 7 derivative would be nice from a familiarity perspective, but Microsoft has demonstrated other concepts that look promising too. The important thing is that it must be built from the ground up for touch, and that must extend to software.

What's remarkable—and what should be, for any Microsoft shareholders, a deeply troubling sign—is that Ballmer, apparently, wants to do none of this. For him, the PC model is the only option. It doesn't matter that it has never worked for this market in the past. It doesn't matter that the tablet needs a new approach to user interface design, one that is fundamentally different from that of traditional PCs. Ballmer wants the PC business model—a Microsoft operating system on commodity hardware—running PC software, and the unsuitability of that software is seen as an irrelevance.

Tablets are, like smartphones, another growth market that Microsoft is going to fail to capitalize on, thanks to a failure to understand the company's past failures, and a stubborn refusal to recognize that not everything is a PC. The company has, belatedly, woken up to the realities of the smartphone market and will, at last, produce a phone operating system that works properly with fingertips, and this fall will produce a credible, if imperfect, smartphone platform. When will it do so for tablets?