But going a step further, there's a huge inefficiency in these hub and spoke models of distribution and shipping stuff long distances. If you can produce stuff locally, and distribute it locally, that gives you a huge advantage.
Er... really? Actually, transportation costs have dropped by orders of magnitude (a point Stross himself made earlier), to the point where they are a fraction of the retail cost of most goods. This didn't used to be the case. Whole cities used to spring up along rail lines or rivers or canals because the transportation cost along those corridors was so much lower than anywhere else, and the low-cost corridors were scarce enough that a lot of commerce was packed into any particular one. These days if you need, say, a computer shipped from Asia and can wait a couple of weeks, it comes - extremely cheaply - by sea to a port, then by truck to a distribution center, and then by smaller truck to our doorstep. There are still low-cost corridors, but they are both ubiquitous (highways and roads) and very cheap. Even if you need that computer more quickly, it can shipped by air in a day or two; the cost of air freight has dropped very rapidly.
However, even if the transportation cost is low, it is still a cost that does not need to be paid if the item is produced locally. So is there an advantage to local production? For some goods (milk, let's say) there is because of spoilage. But for many goods the production cost is simply far lower elsewhere than near wherever you happen to be. Modern economies involve thousands of products, the production of each of which involves dozens or hundreds of intermediate goods, raw materials, etc. There's an extremely complex interchange of materials. The odds are very good that not all of these goods will be most efficiently produced locally. There are economies of scale in most production, after all: instead of having ten separate factories, it's usually more efficient to have one factory with a tenfold greater capacity. But that one factory has to be in one place, which means it isn't in all those other places; if you want the goods that factory produces, you have to ship them there.
This isn't really complicated or mysterious. Which is why I'm mystified that Krugman (Nobel Prize-winning economist) has such a weak response to Stross:
That's by the way one of the mysteries — we don't quite know why there's so much stuff being shipped long distances.... Why were they traveling and — better still, when you're flying between Cleveland and Atlanta, what is it that Cleveland has that Atlanta needs? What is it that Atlanta has that Cleveland needs? Actually, what is that Atlanta has that anyone needs?
Surely Krugman understands comparative advantage? Even if Cleveland produces everything more efficiently than Atlanta, Atlanta must have some good that it is comparatively best at producing next to Cleveland, and that is what Atlanta will tend to export. But Krugman's elitist tendencies get in the way. Perhaps he is playing to his Canadian audience: they enjoy hearing a joke at Atlanta's expense. In a nutshell this has been Krugman's career since become an opinion journalist: he trades on his (impressive) academic economics background to get people to listen to his vacuous, ill-thought-out biases and prejudices. He really should know better.
Stross, incidentally, comes to Krugman's rescue here:
I reckon that complexity, the number of different components, the number of different specialties that you need to run a modern, high tech civilization has mushroomed by orders of magnitude over even the past 50 years. And all of this stuff is really small, very, very specialized, there's only a very few people who do it in one place and it has to get shipped about even though it looks similar.
That's right. The science fiction author is teaching economics to the Nobel Prize-winning economist.