Need to deliver, or what?
Touch is bogus. It is not at all practical. It needs to evolve into gesture, which avoids fingerprints, and avoids having to "reach" something. Gestures open up the possibility of a computer learning how you [want to] move and reacting accordingly.
RIM blew it, by kowtowing to certain governments who wanted access to their servers. Add that to the outages they've experienced, and their trust factor has almost disappeared. There's nothing wrong with a steady presence in a target market; the arbitrary search for "growth" is just to satisfy the greedy financial sector. This is the year I think that sector will finally be recognized as the fraudulent parasites they are, and treated accordingly; it will be a necessary part of the coming Greater Depression. Back to RIM, they are nothing special since they sold out. Why would a buyer want them, unless it can guarantee the security that used to be there, certain governments be damned?
Siri is a toy, and will be for some time to come. If it becomes a useful tool, that won't mean it's something that should be used often, except in the specific circumstances that warrant it. Asking for directions while driving is one such use, and beats heck out of stopping and pushing buttons and making menu choices.
One of the few I agree with, for Microsoft's sake, is Windows 8. If they don't "get it right," I could see enough momentum building to move in alternate directions (e.g. WINE) that it begins a slow but inexorable slide into irrelevance. Microsoft has no competition because it is way too fractured. As powerful as hardware has gotten, it may become feasible for someone to create a more universal Windows emulator, that will run consistently [across many flavors of Unix].
As for what's missing, what about GPGPU? That would be a remarkable change, if it can deliver. How would that fit in with SIRI and other human-supporting AI? "Shall we play a game?"