I tried holding down Menu and Minus on the remote, to reboot the AppleTV and bring up the diagnostic menu, to no avail: Nothing would come up on the screen. (I also tried blindly pressing “Down, Down, OK” on the remote, in the hopes that I just couldn’t see the diagnostic menu, but that didn’t work either.) I could SSH into the AppleTV, so at least that part of the Boxee install succeeded — and the AppleTV was showing up and syncing in iTunes, so I wasn’t stuck in a loop of rebooting over and over.

I wasted a lot of time Googling for solutions — which, to be fair, provided some useful information about the AppleTV’s disk partitions, how it does a Factory Restore, etc., but also provided a lot of outdated info. Several pages began with “first, download this disk image from mesu.apple.com” and then linked to an image that’s no longer there; others provided useful instructions on how to backup your AppleTV disk, which would have been good advice to follow before trying to install Boxee, but didn’t help with recovery.

So, for posterity, here’s how I forced my AppleTV to re-load its factory settings when all I had was the ability to SSH into the box. (I should emphasize here: There are much, much easier ways to restore factory defaults on an AppleTV if you can navigate the menus, or if holding down Menu and Minus on the remote successfully reboots the box into diagnostic mode. What I’m about to describe here is a process for deliberately breaking the operating system on the AppleTV’s OSBoot partition, which forces the box into recovery mode; this is not for the faint of heart, I’m not responsible for what happens to your AppleTV if you try it, and heaven help you if you run these commands on your Mac.)

The AppleTV’s hard drive is partitioned into four volumes: EFI, Recovery, OSBoot and Media. The AppleTV boots to the EFI partition first, which in turn tries to mount OSBoot and launch the operating system. In my scenario OSBoot is mounting and booting successfully, but something has corrupted the AppleTV Finder (or some other mission-critical app) and so we can’t get to the menus.

The solution here is to render the OSBoot partition unbootable, which causes the EFI partition to go into recovery mode and offer up the Factory Restore option from its own menus. While there are many exciting ways to render an operating system unbootable, I went with a simple one: Renaming mach.sym (and, for good measure, mach_kernel.prelink) so the OS couldn’t find its kernel.

So, here’s what I did to achieve a factory restore of my AppleTV from the Terminal prompt. Text that I typed is in italics.

YourMac:~ you$ ssh frontrow@appletv.local
frontrow@appletv.local's password: frontrow
Last login: Fri Feb 26 22:55:22 2010
-bash-2.05b$ sudo mv /mach.sym /bad.mach.sym
Password: frontrow
-bash-2.05b$ sudo mv /mach_kernel.prelink /bad.mach_kernel.prelink
-bash-2.05b$ sudo shutdown -r now

The shutdown command reboots the AppleTV and logs you out; when the box comes back up you won’t be able to SSH into it anymore, but (at least in my case) it came up with the “select your language” menu, followed by a menu that contained the factory restore option. (For what it’s worth, I also had to unplug and re-plug the HDMI cable to get my AppleTV fully working again.) After the factory restore I had to do the usual steps to get the AppleTV back on my wireless network, connected to iTunes, etc., but once all was said and done my AppleTV is back to normal.

I might have another run at installing Boxee, but I think I’ll give it a week or two; honestly, if Viacom got their act together and allowed me to subscribe to Comedy Central via iTunes — or even just allowed me to buy a season of The Daily Show for less than the cost of basic cable — I wouldn’t bother. Anyhow, if I do try again I’ll at least have these last-ditch instructions for doing a factory restore written down.

I think that observation is true, but it misses the point: The iPhone has a user interface and UI metaphor that Apple designed, from day one, to work with hand-held touchscreen devices, and to provide a subset of general-purpose computing functions. (A very large subset, as the App Store illustrates, but a subset.) The iPhone UI throws out the “desktop metaphor” of files and folders, changes familiar UI elements (scrollbars, etc.) to fit the smaller screen, and introduces a new vocabulary of gestures — the “pinch” to zoom in or out, for instance — that replace the cursor and mouse with your fingers.

Adapting this UI for a device with a larger touchscreen doesn’t actually require any changes: Apple may roll out some minor UI flourishes with the iPad — three-finger swipes that wouldn’t fit on the iPhone’s screen, perhaps — but the core user interface will remain the same. Every gesture you’ve used on the iPhone will work exactly as you’d expect on the iPad.

Other devices in this category have struggled to scale down a too-complex UI designed for desktop computers, or to scale up too-simple interfaces that were made for special-purpose devices. Apple did neither, and the iPod Touch has already shown that the secret of iPhone’s success isn’t the phone. Apple has rolled out the first major UI advance since the original Macintosh in 1984, and that’s the not-so-secret ingredient that makes the iPad a different animal from netbooks, tablet PCs, thin clients, tablet computers and every other attempt on the long list of failures to simplify the desktop metaphor.1

The personal computer has, by far, the most complex interface that the average user encounters on a given day; the automobile is a distant second, and for that you need a license. And, as Fraser Speirs noted, the majority of computer users don’t want that complexity. They don’t need the overhead of maintaining a hierarchy of files and folders, and they’re happy to delegate the grunt work of managing files to an app like iTunes or iPhoto.

And, even if the user embraces the desktop metaphor, bringing that UI to a pocket-sized device is difficult: UI elements and mouse gestures that made sense a million pixels ago are now too small to use. Either we keep our icons and toolbars above a minimum size threshold, which means they dominate the pocket-sized screen, or we equip the user with a stylus — a special tool for tweezing tiny buttons and scrollbars — which changes how we interact with the device. (And this is assuming that, aside from the screen size, we can fully replicate all the expected functions of a desktop PC in our handheld device.) A device that implements part of a familiar, well-known UI metaphor, and then substitutes new behaviors elsewhere, violates the principle of least surprise in every case where it deviates.

At the other end of the scale we have handheld devices with special-purpose interfaces. The original iPod has a click-wheel, which works very well for selecting and playing music — but using that interface for a task like word processing would be impossible. A typical cell phone has a twelve-button keypad for dialing numbers; in a pinch this keypad can be used for thumbing out text messages (five button-pushes to get an “S”), but trying to browse the web with it results in a very sub-par experience.2

These interfaces are too small and too limited; they can’t scale up to a general-purpose device. “Smartphones” are a vaguely defined category, but one telling feature of these devices is that they used a non-phone UI as their starting point: The Palm Treo was originally a PDA, the BlackBerry was a two-way pager, and Windows Mobile was a desktop OS. The Treo failed because Palm was unable to push the UI any further; the original BlackBerries were barely usable as phones, much less day-timers or anything else beyond email (they’ve gotten better since, but nonetheless); and Windows Mobile is squarely in the “scaling down the desktop UI” camp. The closest we’ve seen to a device that started with the telephone UI and scaled up was the Motorola ROKR, Moto’s 2005 attempt to bring iTunes to the cell phone… which flopped, in part due to the interface.

So if scaling up a simple UI isn’t possible, scaling down the desktop UI isn’t effective, and adapting a UI meant for some other device yields mixed results, then we’re left with the direction Apple has taken: Designing a new user interface for general-purpose computing, from scratch. Apple’s touchscreen UI solves a problem that has bedeviled the industry since the days of Microsoft Bob: How to reduce the complexity of the desktop metaphor without trading off all the power and features. In fact, the desktop metaphor itself was Apple’s baby 26 years ago, when they replaced the command line interface with a graphical UI — and I seem to recall people at the time saying the Mac was “just a toy,” much like the naysayers are dismissing the iPad now.

The iPhone wasn’t dismissed as a toy when it came out, because its competitors were clearly struggling to build a device that combined power and simplicity — and Apple leapfrogged them all with a phone that contained a first-tier web browser, a first-tier music player, and an all-new user interface that made those functions easy to access and use. Adding over 100,000 third-party apps, as Apple did in subsequent years, has just been insult to injury: The iPhone is a blatantly disruptive technology. Blazingly, screamingly disruptive, forcing AT&T and Verizon and Microsoft and Nokia and Palm and multiple entire industries to scramble and react; there have been trendy cell phones before, like the RAZR or the BlackBerry, but we really are going back 26 years to the original Macintosh to find a single product rollout that turned this large of an industry on its ear so badly.

So, yeah, the iPad is a bigger iPhone — if by bigger you mean more disruptive. The newspaper industry is already in a death spiral, and here Apple’s rolling out a device that wants to be the iPod of print media: What do you think is going to happen? Never mind that it’s a sub-$500 device that does most of what people want from a computer, without hassling them with malware or driver conflicts, and never mind that book publishers are practically salivating at the thing: The question is where the iPad stops being a disruptive technology. Computers have been getting smaller and faster for decades, following the steady curve of Moore’s Law — but making computers easier to use has been more of an incremental process, and mostly a side effect of “faster.” Not many people saw the Mac in ’84 and promptly predicted it would turn the publishing industry on its ear, and most prognosticators dismissed the idea that we’d all be using a GUI ten years later. But both things happened, and the UI improvement is what made them possible.

1. Arguably netbooks don’t belong on a list of efforts to simplify the computing experience: They’re just making it cheaper, trading off power and screen size for a lower price point. And there’s a market for that, but it’s more a sub-category of laptop than a new type of device.
2. And yet the cell phone industry spent years working to deliver that sub-par experience, designing a protocol called WAP and building cut-down web sites that you could navigate using a telephone keypad, before the iPhone arrived and blew them out of the water.

I think the iPhone Lite will simplify the iPhone’s overly complex one-button interface — it’ll have no buttons, and no microphone or speaker. It will require the use of a headset, and Apple will introduce a new Bluetooth headset with volume controls to accompany it.

Take an iPhone, remove the button on the front, the internal speaker, the microphone, the volume controls and maybe the camera, and you’re left with a device about the size of a credit card and the thickness of the new iPod Shuffle. The most popular carrying case doubles as a bifold wallet.