On Thursday, July 27, 2006 04:40 MEST, Michael Perelman replied to Paul: >> Temin's "proof" is a linear regression with all of 6 prices/locations! >> [could a lefty could get away with that at MIT?]. > > Me: He explains why Roman data is so scarce in the article, so that > defect might be forgiven.
Alas, there is a minor mistake in the graph at http://michaelperelman.wordpress.com/2006/07/24/resource-cost-of-cities/ : The distance from Rome to the Po valley is only about 500 km and not as shown more than 1500 km. Hence the relationship between distance and wheat price discount is not as straight as Temin guesses, and the excellent fit of the linear regression has gone. hk
