Treehugger has a post about a theoretical idea for "food towers," condominiums for farms, to fit into dense urban landscapes. On the one hand, it's a nifty idea that's worth a try--and I have to confess I get a little Tom Swifty-ish when presented with an outlandish sci-fi concept. On the other hand, I've read my Wendell Berry, and know this kind of thing crops up from time to time, and promises breakthroughs that are never realized because ... it winds up being kind of dumb. A better idea might be Berry's call for more horse-based farming. Even something like the Post-Carbon Institute's Local Energy Farms project. (And, of course, any time innovative agriculture comes up, I have to plug the Land Institute, which is working to create an American prairie-based food polyculture.)
Of course, the big problem with Berry's solution, and even the Local Energy Farms and the Land Institute, is that while they're doing great things to improve agriculture, they don't really have much to say about the food system, and about the crucial connection between cities and their hinterlands. (I'm taking this city/hinterland language from William Cronon's book on the development of Chicago, Nature's Metropolis (Powell's).)
One attempt at making this link is the 100 mile diet, started by a couple in Vancouver. It's a nifty idea, but I worry that it doesn't scale up. That is, why does everyone get a hundred miles? I.e., why are New Yorkers and Springfieldians looking at the same foodscape? I tried to scale it based on population once, and the result that I got was that all of Springfield should be eating from Sangamon County alone. That didn't quite make sense either.
Somehow, you have to be able to account for the population of the city, the quality of the surrounding land, and also to provide for open space and natural habitats. And I don't see anyone working on that.