Variables?

I'm reading 'Collapse' by Jared Diamond, author of Guns Germs and Steel. I'll do a longer review when I've finished it (short review fascinating, although seriously limited by his inability to analyse power structures), but something occurred to me when I was reading last night, and I wanted to write a little more about it.

The book first focuses on ancient societies, and then looks at modern societies that have collapsed, or could collapse. I find his analysis of those societies interesting and useful, in roughly that order. I don't think I can put this all down to his complete lack of power analysis.

I wonder if it's actually because he comes from a scientific background, and his skill is in controlling and comparing the variables. We know so little about Easter island, the Anasazi, and the Norse in Greenland, that in a weird way it's easier to compare them. The irrelevant fluff (and probably lots of relevant stuff) has already been filtered out. When he's talking about China, he doesn't seem to put it together to anything coherent, there's too much detail about China coming in and out of the WTO, the one child policy (which he doesn't have a problem with - his lack of power analysis goes quite deep), and just too much stuff for someone who has a smattering of knowledge about the subject to disagree with him about. At the same time it's not really good qualititive work, it doesn't explore the reality of the interaction between people and the environment in China (and you couldn't, not in a chapter), because he sees the environment and people as two variables, and doesn't seem interested in a more nuanced understanding of what's going on within 'people'.

Possibly it's just me being a discipline snob, but I do think that there are problems with trying to explain human behaviour in a quantative manner because ususally there are just too many variables.