The same could have been said of Palm Pilots and Blackberries over ten years ago. And yet, here we are. PDAs are dead, and Blackberries are irrelevant. Not because they were terrible ideas, but because technology advanced enough that they became irrelevant, and they adapted to being a complementary device and not a device in which to replace the need for a desktop/laptop.
I also think that stating that the PC is dead is way too much of an overstatement, and see some reflection back to the past. Tablets and phones are good for one thing, and one thing only: consumption of content. Try to do anything outside of that sphere, and it just comes across as a rather clumsy device to use.
And I really have a hard time seeing it ever bridging that gap, since this is a problem inherent with the input system of choice. Much like speech recognition software hasn't replaced the keyboard, touch is not a replacement for the old mouse and keyboard. The problem lies in how by its very nature, touch is less precise of an input method.
That doesn't mean that I'm being a neophyte in regards to touch devices, since they do indeed have their place, but they just aren't the be all end all solution, since unless we're going to go to Minority Report like interfaces to make up for the loss in precision, which, BTW, would be completely impractical for long term use because of the ergonomics involved, then there will continue to be a need for the current input paradigms that we have now.
But that doesn't mean that the desktop won't change because of touch though. Gestures might very well become integrated with the desktop without too many problems, and which for that matter, Opera was a pioneer on this in some respects. However, it's not going to replace the need for finer grained controls. What such an input device would look like though, I dunno. It could be rather much like a trackpad on a laptop, could be integrated into the keyboard for a touch area, or even something else entirely. But whatever it ends up looking like, I'm just not seeing the killer advancement to enriching or supplementing the desktop... yet.
What I think the main problem here is that many who are involved in HCI prefer revolutions to incremental improvement, and then call anyone who doesn't want to jump on their latest bandwagon which doesn't want to go along with their revolution as being technophobic, when their new system that they propose either can't or won't replace all of the use cases that they think that it will.
I get it. Developing for older systems can be boring (although I'm a rather strange one and actually love to be on the incremental edge), and continually delving into unexplored areas is much more exciting. However, computing has never worked that way, with every advancement always being some incrementation and refinement of the older ideas which then builds on the work done by the previous generations of tech, instead of trying to replace it entirely.
So in that respect, I fully expect that touchscreen devices will likely end up being in the same position as your Palm Pilots and Blackberries in 10 years time. They will have failed to live up to the hype being given to them, and they will be relegated to being mostly entertainment devices overall, possibly replacing TVs, gaming devices, eReaders, and so on. Heck, we're actually starting to even see that now.
But they won't be replacing the need for an actual computer, much like Blackberries, iPods, and Palm Pilots didn't supplant them either. The input is just too clumsy to do that, and there's really nothing that technology or software can do to change that. But just because I say that doesn't mean that I then think that the desktop then needs to remain unaffected by it, and that it won't change as well because of it, but that such a change is going to end up being more incremental, rather than revolutionary.
Touch won't kill the need for a mouse, but something else which brings the best of both forms of input just might. And that replacement will then likely be just as inappropriate for use with a tablet as a tablet's interface would be for the desktop, even though it might take a lot of hints from the tablet paradigms.