Actually, they're not really about natural language--that's what I'm trying to get across in my post. They're more about a place to store and use structured dat

Actually, they're not really about natural language--that's what I'm trying to get across in my post. They're more about a place to store and use structured data and access it formally. That they have an NLP interface right now is, in my mind, a kludge. And they aren't taking an open source / wisdom of crowd approach. Rather, they are taking a very retro curated data approach, more Britannica than Wikipedia. I'm a huge Wikipedia fan (see my recent post reacting harshly to Nick Carr's calling Wikipedia a "Potemkinpedia"), so my reaction was highly skeptical. But, for objective data like the populations of countries and atomic weights of elements, curated works just fine. In fact, for structured data, I suspect curated is probably the way to go.