This is a great period for fans of ancient industrial history; the scientific world is beginning to take the discipline seriously, as recent papers in Nature about the Antikythera mechanism, another this week on the chemistry of violin-making, and the atomic structure of the legendary Damas blades (also reported in Le Monde) - along with Greek Fire, was the best kept (and best lost) military secret of the middle ages, demonstrate.

As an aficionado of old-school industrial archeology, though I am delighted by this renewal of interest, I find that the high-tech hype around it (CT scans, carbon nanotubes, etc.) may be something of a facade dissimulating the paucity of really novel historical insight: it is a pity for example, that in covering the Antikythera paper, news@nature does not even mention Derek Price, who discovered the mechanism more than 30 years ago, and presents as paradigm-shifting some very old (and beautiful) conclusions that were drawn by Joseph Needham and his team a long time ago on the nature of antique clockwork.

Anyway, my official Favorite Paper of the Week is not the Antikythera reconstruction, it's a paper by three economists, Diego Comin, William Easterly and Erick Gong about the technological causes of the wealth of nations. It deserves much praise for two reasons: first, it takes great risk by trying to assess technological progress over the last 3 millenia in the entire world. This ambition is, as far as I know, unique, even among specialists of universal long-term history such as Mokyr, Diamond, McNeill or Crosby. It implies exposing oneself voluntarily to countless criticisms from the disciplines you're drawing from, which is simply heroic (no irony there). Second, because it vindicates the idea of technological determinism, which, though widely popular, is systematically derided among scholars in history, the humanities, and most of economy.

Technological determinism has a strong version which is obviously false: it's the idea that technological activities directly cause all other human activities without being influenced by them, and that technology is the sole engine of History. The influence of science over technological evolution these last centuries proves this strong version wrong. But this does not mean that technology is not a powerful force in history in its own right (weaker version of technological determinism). Even this weak version is systematically downplayed, the authors say, in development studies and economic history, in favor of institutional or econmic factors. Technological history anterior to the arrival of European colonizers is typically judged irrelevant.

Comin et al.'s study strongly suggest that it is not: if you take the territory enclosed behind the borders of today's nation-states, then try to assess what their technological level was back in 1000 BC, 1 A.D. and 1500 A.D., using a digest of development measures resembling indices of "cultural complexity", or standardised entries in the Human Relations Area Files, you find that the situation in year One is a significant predictor of per capita income and population size in today's nation states. For some differences in wealth, this is even true of 1000 B.C. technological levels; and of course the tendency strengthens as you reach 1500 A.D.

If you consider that the existence of the US and Canada is a huge departure from the tendency underlied by the authors, the tendency should be very strong to overcome it (or perhaps this is a statistical fallacy which I am committing now; Simon will correct me if I'm wrong):

"once we control for the most obvious historical example of replacement of the indigenous technology by technologies brought by new settlers, technology in ancient times becomes a significant predictor of per capita income today."

The authors bravely face the major problem that plagues very large-scale studies: vapid data, sometimes vague to the point of vacuity. They acknowledge that they have no bullet-proof shield against the problem, but they confront it with a very variegated and well chosen set of data, ranging from military to agricultural history (on-line material is accessible for free) - the diversity of points of view being the only way of achieving redundancy. Too many points remain flawed, however. I disagree in particular on 2 key assumptions of the study: the first one is that technological evolution is unidirectional, and the second, that it is gradual.

Perhaps that is a convenient simplification that does not distort the data that much, but they should certainly not assume that it is a law of culture. Technological evolution is not unidirectional, as the Tasmanian and Japanese cases suggest (Jo Heinrich and the late Gerry Martin respectively have written wonderful things about these two regressions. The existence of catastrophic reversals and accelerations in the history of technology is obvious (and that condemns gradualism too).

The most fascinating aspects of the paper, apart from the main thesis, are the interactions of the technological dynamics with geography and population. The authors don't find that technological development correlates with latitudes, as so many authors driven by climatic theories (Ecuador=germs=no progress) had proposed. But continents, on the other hand, seem important, in the African and American vs. eurasian case in particular: this should come as a delight to Jared Diamond, who says that biological and technological developments in continents is driven by the orientation of their axis. In short, the weight of technological differences in 1000 B.C. disappears when they add a continent dummy to the analysis.

The authors also claims that technological development weighs on population and the total expanse of arable land, but Institutional factors, insofar as they can be measured through such indexes as the Polity IV factor, are inert.

The paper is a little short on theoretical justification of the results (but then, it is already so general, one cannot expect them to reveal all secrets of history in 15 pages); they simply opt for the existence in some cultures of some ancient, pervasive and far-ranging technological momentum:

We do not claim to definitively resolve WHY technology in 1000 BC or 1500 AD still matters so much today, a question on which we hope to gain insight from further research. We think that the simplest explanation is that technological experience has an important effect on the ability to adopt the new technologies that have come along since the industrial revolution.

An unsatisfying conclusion, to be sure, but a tantalizing one; shall we measure and track down technological momentum in the future? A first move could be trying to pin down exactly what kind of technologies explain the differences in actual wealth:



The main finding is that different sectors are driving the positive relationship between overall technology adoption and current development for different periods. The sophistication of agricultural technology in 1000 B.C. seems important for current development but the importance of agricultural technology declines for later periods. Technology adoption history in communications and industry in 1000 B.C. and 0 A.D. does not have a positive effect on current development (communication in 1000 B.C. is negative and significant.) In 1500 A.D., however, the degree of technology adoption in communication has a strong positive effect on current development. Military technology adoption history in 1000 B.C. and 1500 A.D. have important positive effects on current development. Finally, technology adoption history in transportation in 1000 B.C. and 0 A.D. has a positive effect on current development but the transportation technology adoption variable in 1500 A.D. does not affect current development.

(that said, I wish the reader to know that I am but a very mediocre statistician - worse, a psychologist, and a know-nothing in economic maters, so don't take my word for that paper... you've been warned).