Gimmicks and whizbangery are not a good way to change the world.
There are two major trends that will dominate "computing" (for lack of a better term than that increasingly antiquated one) for the next couple decades: transparent computing and ubiquitous computing.
Transparent computing is the idea that when you're doing a computery-task-thing, the computer should vanish. This is what inspired multi-touch, in large part: You're no longer interacting with a computer which interacts with digital representations of things. You're directly interacting with things. The Jeff Han demo at Ted captured this idea perfectly: You reach out and touch a photograph, and it responds like a thing. There's a powerful computer behind all of that, of course, but the goal is for the person to be unaware of the machine. More and more, we're seeing these principles put into action. We've got it down for photos and scrolling through lists or documents; we haven't figured out how it should work for other types of interactions yet, but things like accelerometers and AGPS give hints of possible future developments.
Ubiquitous computing is the simple idea that where you are is more important than where the machine is. There are several ways to tackle this problem. "Everybody should have a laptop" is one; an iPhone-like device in your pocket is another. But we're still a long way from the ideal there. If I want to move a document from my laptop to my phone, I have to go through an abstract, mediated process like syncing or mounting a shared filesystem over a network. I should be able to put the phone next to the laptop screen and simply drag the document to the phone with my finger. We're not there yet.
The shortsighted person and I mean no offense here, but what else would you call it? looks at a technology like multitouch and shouts "Bolt this onto the side of my existing computer immediately!" Except that's nonsense, because multitouch is one aspect of a different vision for human-machine interaction. Multitouch by itself is useless; it only has value to the extent that it lets us get closer to transparent computing.
I've got a year-old Macbook Pro. It has multitouch. I can use multitouch gestures on the trackpad to do things like zooming in to a PDF or tell my browser to go back a page. I NEVER use these features. Why? Because they don't do anything for me. They don't make the computer disappear. I'm still interacting with a machine, and the machine interacts with whatever I'm looking at on the screen. It's like those plexi boxes that you see in science labs in movies, the kind with two joysticks and a pair of robot arms. You move the joysticks to control the arms which lift the lid off the barrel of atomic waste, or whatever the heck. It sucks, and we only put up with it because there's no alternative that doesn't involve a slow, agonizing death from radiation poisoning.
Okay, so maybe my simile broke down a little at the end there, but you get my point. Multitouch isn't a feature that should be stuck on to existing clumsy, abstracted, in-ten-years-we'll-laugh interaction paradigms. It should be used as one of many features to create new, better interaction paradigms. Anything else is just painting the brontosaur's toenails as the comet streaks across the sky.