I have zero doubt that all of the rumors about Apple doing a tabletesque device are true and that it will be the company's answer to netbooks and the explosion of demand for general purpose computing devices in the sub $500 range (and no, the iPod Touch, great though it is, is not the answer here).
If I am wishing for one feature though, it is that the aforementioned device does not in fact run iPhone OS, but instead some form of their desktop OS X. Given the company's preference for closed-and-controlled environments, I find it unlikely, but here are two good reasons why:
1. Having a device that has a different display size and other hardware will muddle a relatively nice developer story on the iPhone (Symbian anyone?). You can already see this beginning to happen in the 3G/3GS split and that is for CPU reasons alone; for instance, we've got this application that we've been testing at work that runs 4x faster on the 3GS than on the 3G. This basically makes the difference between usable and complete garbage. And this is just the beginning of the pain that we'd face if we had to be checking for other hardware differences, or worse still, designing "scalable GUIs" that somehow always end up looking shitty.
2. A general purpose computing device depends on a lot of two-decades old infrastructure that a new OS like iPhone OS just doesn't have. Printing. An ecosystem of 3rd party applications. Other input peripherals. By leveraging desktop OS X, the tablet would get all of these out of the box.
Incidentally, even mighty Google seems to have realized this with Chrome OS— after a year of everyone trying to shoehorn Android into a general purpose form factor, they just came out and said "whoops!, we really just want a regular Linux distro in there after all."
One final note: several folks have argued that the desktop model for computing is broken and that consumers need something simple like the iPhone OS but I don't buy this at all. While the PC OSes may be challenged (especially on the Windows side) when it comes to usability and maintainability, they have been incredible platforms on which to build for two decades— as Zittrain would say, "truly generative," and to throw them out now seems both pedantic towards the users who we all supposedly serve and very short-sighted.