[dredmorbius]

Bingo. I've been discussiing this on G+. In paarticular, adding topographic maps helps show transport logic pointedly ignored by the paper.

The areas may simply be more amenable to prosperity: good land, water, resources.

Roads were little used for transport. Goods moved by water. Rivers. Sea. Canals, mostly after 1500. Costs were 1/20th or less of overland drayage.

Transport routes are established between points of interest, and those develop according to potential. Several of the major roads follow coastlines or rivers. Others bridge river valleys, generally through other valleys and over passes.

The Roman empire itself grew into areas offering food, lumber, or other trade.

The paper's conclusion of causality is grossly premature and overstated.

https://upload.wikimedia.org/wikipedia/commons/thumb/a/a6/Fr...

http://floodmap.net/Elevation/ElevationMap/CountryMaps/?cz=F...

https://upload.wikimedia.org/wikipedia/commons/d/dc/Topograp...

[growlist]

I've only had a quick skim but what worries me is that the authors are investigating a fundamentally geographic problem yet don't apparently have a background in GI science. For example, the authors use fixed grid cells for analysis, yet the modifiable areal unit problem states that varying the boundaries of your analysis can change the result, and especially given the coarseness of their chosen grid I'd like to see at least some attempt to handle this (perhaps repeat with smaller/shifted grid cells, compare results), or failing that at very least evidence of awareness of the problem. I've an MSc is GIS and one thing I learned is that even with the best data and methodology (PhD level spatial stats) it can be extremely difficult to establish statistical significance, and I'm not convinced here. I'd like to see a far more sophisticated treatment of the GI science methodologically before the leap into the historical/economic domain. I strongly suspect the geo side would fall apart under close analysis, thus undermining everything beyond that point.

[dredmorbius]

More or less my take.

Misuse of statistics.

[vidanay]

The images from the article distinctly show the roads and population centers which then shows the mountain ranges in "negative space."

(I agree with you, just pointing out another observation)

[et1337]

What about the negative result TFA found for the Middle East and Africa?

[dredmorbius]

Topology and climate of both regions differs markedly. The significant population centres in both regions developed along their respective major river systems. Most especially the Nile, and Tigris-Euphrates.