The rise in the environmental costs of production under a "buying local" regime would more than offset the environmental benefits of reduced transportation.

More here.

A few things...

It seems that these posts largely (if not intentionally) ignore two of the three pillars of sustainability -- economic and social concerns.  Furthermore, I think the energy arguement ends up wrong because it's incomplete (as energy analyses inevitably are).  And I don't agree that the energy makes the "biggest difference."  I also can't make the numbers any less squishy--there's too many factors to make hard and fast statements about agriculture (all the more reason agriculture can't be run only as a business, it's a lifestyle).

Anyway, the statement that there's probably a huge range of on-farm energy costs across the distance grade is correct.  However, it is only with the local that it is so easy to actually find out how much energy is expended on the farm in the production of each commodity (in soft terms, of course).  You can ask the farmer, even go to the farm!  Further, you can find out what pesticides and fertilizers are used (pollution of air, water, and soil and fossil fuel implications).  Smaller farms can easily use less energy than large farms who are bound to huge diesel rigs by using a combination of manual labor, small tractors (possibly using renewables???), and maybe even traction animals (gasp!).

I'm sorry, but this just doesn't add up to me:
So by this reckoning, growing all of your food in your own backyard isn't as important as improving your car's gas mileage by a mere 3 percent. Or, put differently, all else being equal, it may be wiser to choose a home within walking distance of a grocery store than one that's adjacent to the fields where your food is grown.How do you figure?

This one also doesn't cut it:
We lower the environmental costs of food production by growing each kind of foodstuff where it is more efficient (on average) to do so. For example, we tend to match plant varieties to climates and soils. If we grew everything everywhere, so that everybody could always buy local, we would need more pesticides, fertilizer, water, energy and land to produce the same amount of food.
Do you have anything to show that this is actually what we do?  Potatoes are from South America, but I believe Idaho is famous for them.  How about corn--I think it's more driven by reliable subsidy payments than by being climate-appropriate.  "If we grew everything everywhere," a nonsensical notion from the start, there would indeed be some challenges.  But how about the option of dietary restraint?  Furthermore, growing a wider variety of crops in all areas is possible with current conditions and could easily be advanced with THOUGHTFUL RESEARCH into breeding varieties to thier locales rather than for ease of storage and trasportation.  That is to say it could be done with brains (and current technology) rather than with "more pesticides, fertilizer, water, energy and land" as is argued without a trace of support.

Now back to what you're missing-
The social issues cannot be overlooked.  Rural communities turning into slums and ghost towns.  Not to mention children not knowing what a potato plant looks like (do you?).  Then there's the plight of farmworkers on massive farms.  Add in the health, taste, and security concerns with this and it begins to look quite remarkable.

Then there's the economic side.  Money circulating in a local economy is good for consumers and producers (a.k.a Citizens).  Most food dollars funnel into the vertically integrated corporate hands; they supply the seed, fertilizer, pesticide, and machinery, and then by the commodities and process, store, ship, distribute, and sell the "value added products" back to the farmers who feed their families with food they bought at Wal-Mart because they can't afford to shop elsewhere on the margin.

Finally, the references to "back-to-the-land" put a spin of quaintness to the idea of meaningful local economies that is undeserved.  It's pretty obvious that every system we have ever had (or tried to have) has failed when held to the standard of "sustainable."  I think something much more complex needs to develop from astute imagination if food, economics, and energy are to come into some sort of sane relationship.
Your comments about the inefficiencies of a diffuse population are correct when looking backwards, but not necessarily true when looking forwards.  They also fail to recognize the social and environmental costs of concentrated populations who are touted as more efficient.

True local economies will not just be based on food--diffuse populations now rely on centralized supplies of energy, goods, and services.  It is when the sources become as diffuse as the sinks that we can mark one in our column in the struggle against entropy.

So it's not just about buying those grains and veggies (and maybe a little meat) locally--it's got to be about locally producing and buying that fridge and stove you mention, too.

Lots of food for thought (pun, uh, apologized for).

One other thing to toss into the mix. Part and parcel of our current food system, which relies on long distances, is heavy processing of foods. Processed, packaged foods last longer and can be transported farther more easily.

As a result, most Americans eat an enormous amount of carb-laden, processed food. In general, Americans are eating more and more, and getting more and more obese, but at the same time getting more and more malnourished, because the foods we eat simply don't have the nutritional content they once did (this is true of non-processed stuff too, like veggies grown on mega-farms).

All of which is by way of saying: There are enormous health costs to our current food system. Diabetes, obesity, heart disease, etc., are all exascerbated by the crappy food we eat.

So that's an external cost that should be factored in the mix, though I admit it's a long thread to follow and difficult to put hard numbers to.

(This assumes that local food, grown on small-scale farms, is healthier.)

www.grist.org