AR is not VR. I think that VR will certainly enable all these behaviors, eventually. Except in a virtual space, instead of meat space. The behaviors and ideas Dynamicland are advocating should work just as well in both.
And while their points ring true for the current iterations of AR-as-personal-assistant, I don't see why a networked AR where everyone shares the same annotations would work any differently. Isn't that what their projectors are doing, after all?
It's not just the social angle — projections in VR and AR aren't tangible, whereas physical objects imbued with dynamism are tangible. Here's a really good article on why this matters so much: http://worrydream.com/ABriefRantOnTheFutureOfInteractionDesi...
Now, once we have a holodeck, where the virtual objects are made tangible, your points will hold. But until then, tangibility is a key difference (and perhaps not the only key difference, but this should suffice for now).
And while their points ring true for the current iterations of AR-as-personal-assistant, I don't see why a networked AR where everyone shares the same annotations would work any differently. Isn't that what their projectors are doing, after all?