WRT freebase, as is so often the case, its success will probably depend on the unpredictable social dynamics.

I’d add: people generally resist entering any data that they don’t feel a compelling need to enter.

These sites may be good examples of cases where the metadata / data distinction makes less sense. The stuff a person types into fields is always *data* (e.g., at the moment it’s being entered, it’s data), and data can be a type of content that people are interested in generating in that “user generated content” sense of things.

It could make sense to imagine a spectrum between, say, tangible uses and abstract uses of data, e.g., data that I can see as useful to me right now, on one end of the scale; vs data I am informed is going to be useful to me at some point, but for which I have absolutely no tangible sense of its uses, on the other end.

This tangible use / abstract use spectrum may match, or be able to be described in terms of the data / metadata dichotomy. But, I think we’re always going to see new applications that find ways to push metadata into the tangible use end of the scale. And, since I don’t want to sound too idealistic, I should note that I am sure we can look forward to the majority of new applications finding ways to make meta/data seem hard / abstract to use.