Somehow I missed this article from last year. In it, two professors from Carnegie Mellon describe how despite the enormous distances that modern foods are shipped, transportation as a whole accounts for only 11% of the total greenhouse gas emissions produced by agriculture. Delivery of finished food products to retail stores accounts for only 4% of the total emissions. They conclude that "buying local" is less effective in lowering our carbon footprint than shifting from red meat and dairy to less carbon-intensive poultry, fish and veggies.
The greater efficiency of concentrating agriculture in the best locations is reflected in Jeremy's post at Ag Biodiversity. Check out his stunning maps that illustrate the global distribution of farming in 1995 versus the concentration of agriculture in only the best locations (it's not clear if this is hypothetical or actual). Like any other industry, agriculture is most efficient when located where the competitive advantages lie (e.g. good soil, weather and cheap and available labor). Unlike other industries though, I think agriculture may be an industry that we should sacrifice some efficiency in order to maintain the robustness of a distributed network. No one will starve, overthrow their government or start a war if political or natural obstacles suddenly cut them off from iPod factories on the other side of the world...
I don't have the time (or probably the qualifications!) to dig into the authors' methods to see exactly how they calculated all this, but it's a good reminder of how unintuitive complex systems can be. It's easy to suggest aesthetically-pleasing ideas such as buying local, but the devil is in the details. As James points out, there's not even close to enough land surrounding major population centers like NYC or Philly to feed all the inhabitants within a "local" number of food miles - let alone the fact that very few crops are grown in many of these regions to begin with (e.g. pretty much just dairy in upstate NY). It's often more efficient to grow crops year-round in sunny Arizona or California, or in the deep soils of the Midwest (and ship them thousands of miles) than grow them locally in cool, short summers on rocky soil.
As a big fan of New Urbanism, I'd be happy to vote for legislation that encourages dense, mixed zoning with lots of subsidized parks (including garden plots and farms) in and around cities, but that's more a statement of aesthetic preference than of effective conservation. I think we would benefit from a more clear understanding of not only the facts involving the environmental and social impacts of modern agriculture, but also our personal motivations for our beliefs.
If some sort of concentrated feed lot system turns out to be the most environmentally-friendly form of beef production, would you advocate it?