This NY Times article has been making the rounds A Physicist Solves the City
Now that the city is solved, I guess we can all go home.
Apparently physicists have discovered larger cities (1) consume more resources, and (2) possess some economies of scale and agglomeration er "superlinear scaling" (cities increase in per capita productivity by 15% with every doubling). It would be nice if for once a physicist read something outside their own discipline. I won't say the rest is bogus - but I don't think New York City with 8 million people is 1.15^4=1.75 as productive per capita as a city with 1 million people (1 -> 2 -> 4 -> 8) or .03 -> .06 -> .125 -> .25 -> .5 -> 1.0 -> 2 -> 4 -> 8 = 1.15^8 = 3 times as productive per capita as a US city with 30,000 people. It really gets to how you define cities (and productivity), and the bounds within which you are working and have observable, reliable data. There is no reason to believe this is constant over all ranges of city sizes. Further, if this were true, wouldn't everyone in the world live in a single city, with all the productivity that would generate. The positive externalities are offset by negative externalities which reduce productivity. Perhaps there is an optimal city size, given a certain technology level. This is not likely to be constant with changes in technology. Certainly average compensation in NY is not 3x greater than a city of 30000 people.
However I must say Santa Fe Institute has a great publicity machine which the rest of us in academia covet.
For further reading on something that has passed peer-review in our field, see
Cities as Organisms: Allometric Scaling of Urban Road Networks by Horacio Samaniego and Melanie E. Moses. Journal of Transport and Land Use 1:1, pp. 21-39, Summer 2008.