First impression on unpacking the Q702 test unit was the solid feel and clean, minimalist styling.
Windows 7 tablets are a terrible idea
- — 16 December, 2010 09:53
I keep hearing about Windows tablets. Steve Ballmer got up on stage with an HP slate at last year's CES, and nobody was particularly impressed. Now we're hearing rumors that he'll give it another shot this year, perhaps even revealing a bit about Windows 8 (I don't think that's likely, given that we don't expect Windows 8 to be released for almost two years). Redmond, if you're listening: stop it. Windows on tablets is a terrible, terrible idea.
At first, it seems like a Windows tablet would be just what the doctor ordered, right? A tablet that can run all the apps you already know and rely on - full-featured applications that don't just "work with" what you use on your laptop or desktop, they are what you use on your desktop or laptop. Video chat with a front-facing camera? Just fire up Skype or any other video conferencing tool you want. Plug in a USB keyboard, and it just works, because the whole Windows driver stack is available. Windows on a tablet would mean completely circumventing many of the limitations and drawbacks of current tablets based on Android or iOS.
But you can't have your cake and eat it, too. Windows as we know it faces two intractable problems that prevent it from ever being a good tablet OS: its hardware requirements, and its core design around mouse and keyboard input.
Current and upcoming tablets are generally based around the same system-on-chip processors found in smartphones. These use ARM CPU cores and package it together with a graphics processor, video decoding and encoding units, audio processing hardware, I/O hardware, and more. It's all bundled into a single low-power chip. We'll see plenty of tablets follow this mold, featuring Snapdragon, Tegra 2, Hummingbird, and similar system-on-chips. Windows, on the other hand, requires an x86 CPU, the lowest-power of which are Intel's Atom and AMD's upcoming Fusion chips. These are battery-sippers on laptops, but use drastically more power than the ARM-based smartphone and tablet chips. Though they integrate the CPU, graphics, and memory controller, you still need other chips on the motherboard for I/O, networking, audio, etc. Oh, and they're more expensive, too. The result: a tablet that requires a larger battery to have decent battery life, a higher BOM (bill of materials) cost and thus a more expensive tablet, and tablet designs that are thicker and heavier than the iPad (which is already at the threshold of what one can reasonably hold in one hand for, say, reading an e-book).
Windows is an OS designed for running many tasks at once, far above and beyond the "pause and resume with limited background processes" style of multitasking we see in most phone and tablet-oriented operating systems. It's made for running applications in, well, windows - a pointless bit of overhead on a device small enough that you'll really want to run everything full-screen. It requires more RAM to function well than Android or iOS, and puts more demands on storage performance. Windows is a miserable experience with tiny, small, slow hard drives, and the SSDs that make it fly are still very expensive.
If It's Good for a Mouse, It's Bad for Your Fingers
Even if we ignore the hardware problem, the fundamental design of Windows and its applications is completely flawed on a touch-based device. Yes, you can mimic clicking with a mouse on a multi-touch display, but a good interface for a pointer is not a good interface for your fingers. Back in the 50s, a psychologist named Paul Fitts devised a model of human interaction with machines that is still very relevant today. People call it Fitts' Law, and it's a mathematical formula that describes the time it takes to interact with something as a function of distance moved to a target, speed, and width of the target along the axis of motion. Applied to a mouse interface, it essentially says that the time it takes to complete an action is a function of the distance the mouse has to travel, the speed it moves, and the size of the target (icon, button, menu, whatever). In other words, the further the mouse has to go, and the faster you move it, and the smaller the target you're trying to click on or drag to, the worse your accuracy is and the longer it'll take you to do that action.
A good mouse interface is therefore one where interface elements that are used together are near each other, and where targets are reasonably large but not so big that they take up too much space from the actual content of the application. Also, the edges of the screen are golden: you can't accidentally "over-shoot" your target because the mouse stops moving at the edge of the screen. This is why the taskbar in Windows and the dock in OSX are so successful - you can easily and accurately move the mouse pointer down to the bottom edge where your target is pretty wide and the edge of the screen keeps you from going too far.
Now, consider what makes a good interface on a touch-based device like a tablet or smartphone. You have no cursor starting position as you do with a mouse, so it doesn't matter if interface elements are close together. You're just as accurate at tapping on an interface element anywhere on the screen. In fact, putting interface elements too close together is actually detrimental to your accuracy, because our fingers are a bit imprecise and it's easy to miss your target by a small amount. Good touch interfaces space objects apart. On a touch device, you want to avoid the edges of the screen because users can, and do, tap outside the display area, which will register no action at all. Then there are those interface elements that are simply vestigial wastes of space on a touch interface but necessary in a good mouse-driven design, like scroll bars.
In other words, the fundamental principles that make a good mouse interface are antithetical to what makes a good touch interface. And I haven't even gone into multi-touch gestures like pinching, multi-finger swipes, and other features that a single mouse pointer can't replicate!
How Microsoft Can Attack the Tablet Problem
I have no idea what Microsoft's tablet plans are, but I can tell you what it shouldn't do, and offer some suggestions. To begin with, Microsoft should absolutely not simply put Windows 7 on a tablet...effectively a netbook with no keyboard. That makes for a tablet that is too large, too expensive, too short on battery life, slow to boot up, and so on. Also, as I discussed earlier, the fundamentals of a mouse-driven interface make for a poor touch-driven interface.
Microsoft might put some sort of standardized front-end for touch on top of Windows 7, as a way to get you to your media or apps more easily. I hate that idea, too. If you've ever used HP's TouchSmart PCs, you know these front-ends are often clunky and slow, and once you get into the application, you're back to an interface designed for a mouse. You can, of course, make a series of touch-centric apps, but then you're throwing out one of the key benefits of running Windows in the first place; the ability to use the millions of Windows apps out there.
To my mind, there are three courses of action Microsoft can take that have a real chance of producing tablets that are actually a joy to use:
1. Build a whole new tablet OS. This obviously lets them make an interface and rely on applications optimized for touch and for the tablet form factor, but it's a huge undertaking that would simply take too much time and money to do right. By the time a hypothetical new tablet OS ships, Apple and Android tablets will have completely dominated the market.
2. Scale up Windows Phone 7 to a tablet version of the OS. This could be a great idea. It's already got a slick touch-based interface, and if we assume that some updates in early 2011 will address missing features like copy-and-paste, missing browser features, and multitasking, it could easily be competitive with other tablet operating systems. It runs on those energy-efficient and cost-conscious ARM based processors, and it has great tools for building touch-centric applications already in the hands of thousands of developers. If I was Microsoft, I would throw huge resources at making this happen post-haste.
3. Take the Windows Media Center route with a Tablet interface for Windows 8. Media Center is a sort of alternate "mode" of Windows designed to do the things you want to do from your couch (watch TV and internet video, listen to music, stuff like that). It's made to be operated with a remote control, never touching a mouse or keyboard. It's got an API of sorts, a framework, for developers to build Media Center apps that run only in the Media Center interface and work with a remote. If you've used it lately, you'll know it's actually a pretty great set-top box experience. In a similar vein, Microsoft could build as a standard part of Windows an alternate front-end for touch based devices like tablets, with its own application framework and completely different UI. It's still Windows underneath, but you wouldn't know it (just as you don't see anything Windows-like in Media Center, beyond the logos). This could be very successful, but the timing is poor. It still relies to tablets running x86 processors from Intel or AMD, and we're still at least a year from that sort of hardware being inexpensive and energy-efficient enough to make a great tablet.
Call me a cynic, but I don't expect to see an announcement in the near future of Microsoft pursuing any of those three options. I'd love to be wrong, but I expect to see thick, heavy tablets with inferior battery life based on hardware that would make for a disappointing netbook, running basic Windows 7, with various disparate touch front-end interfaces designed by the hardware vendors, putting a thin veneer over the fact that the whole system is really built for a mouse and keyboard it doesn't have.