Monday, 9 October 2017
Stone Age Scorpions And Our Toxic Love Affair With Data Empiricism
It's a truism that interpretations of history (including pre-history) are a sort of weathervane for the contemporary concerns of the historians in question. Just take a look at the weight given to climate change and complex systems based explanations in current thinking on issues such as the Bronze Age collapse, collapse of the Roman Empire, and the origins of the world wars. What's also interesting, though, is what these discourses reveal about contemporary methodologies of thought.
Archaeology and ancient history have become unlikely pioneers in the area of multi-disciplinary studies. Unlikely because these are traditionally such conservative fields, but easily comprehensible in hindsight given the nature of the undertaking. This means that there is an admirably collegiate culture, especially in archaeology, characterised by strong openness to discussion of finds and findings from pretty well any specialist in any field. A good example of this is the recent discussion of the Gobekli monoliths as an ancient astronomical observatory.
It all started, as most archaeological controversies start, with a stupid and poorly researched news article. A whole series of articles, more or less factually incorrect, ran with a paper put together by engineers from Edinburgh University which purported to have used statistical data analysis to match symbols on the gigantic T-Pillars to astrological signs. Apart from variously mis-describing these engineers as 'archaeologists', or vaguely referring to them as 'scientists', every one of these articles described the findings of the paper as if they were incontrovertible fact. As it happens, the team at Gobekli Tepe run an excellent and very informative blog, possibly because they're sick of all the ancient alien morons taking all the oxygen. What can also be found on their blog is the discussion they had with the authors of the report.
Some of this is a little bit abstruse, so I'll provide a quick summary here for context. Basically, the engineers decided (for reasons which are unclear) that a particular scorpion symbol was a sign for the Zodiac constellation Scorpio. Using various data analytics tools, they then cross compared a selection of symbols on various pillars with the calculated positions of constellations in 10000 BCE. Finding that they were able to associate a number of animal symbols with current astrological designations, they then wrote their paper claiming that the site must be an observatory and, further to this, decided that one of the images must represent the still somewhat dubious Younger Dryas impact. So far, so depressingly standard for the use of data in academia. But what's really revealing here is the nature and content of the discussion which ended up happening between the archaeologists and the engineers.
The team at Gobekli Tepe, to whom I am admittedly partial, reacted fairly scornfully to this paper. They pointed out various flaws in the methodology and selection of evidence, and raised a huge question mark over the scorpion/Scorpio thing, making the very valid point that Zodiac signs as we know them aren't really attested prior to approximately 2000 BCE, and that those signs, which form the basis for current Zodiac iconography, are from a very great distance away from the site. Basically, they contended that the distance in time and space rendered the very first and basis assumption invalid, or at least highly dubious. They then went on to point out that the selection of pillars seemed random, was not in any way comprehensive, and had a distinct look of 'cherry picking' about it. And on top of all this, they raised the point that the monument had been altered and reconfigured over many generations, rendering a single purpose unlikely. This done, they pointed them to their own theories about the monuments being indicative of emerging social complexity for consideration.
The response from the engineers was revealing. The engineers airily dismissed the social complexity theory in a single truncated sentence which labelled it as 'opinion'. They then proceeded to blame the archaeologists' slow publication rate (it's actually really fast) for the incompleteness of their data. No coherent defence was made of the basis assumption beyond 'scorpions have always stood for Scorpio', which is ludicrous, and they also embarked on a long, inexpert, and rather sterile discourse on the survival and transmission of stories. All of this was capped off with the bland assertion that “… given the statistical basis [of their] interpretation, any interpretation inconsistent with [theirs] is very likely to be incorrect.”
And there's the kicker. It doesn't matter that the base assumptions for their data analysis are basically pants. The fact that archaeologists and ancient historians spend their lives studying ancient iconography and mythology is utterly insignificant. The fundamental flaws in their evidence selection are irrelevant. All that matters is that their analysis is 'statistical', which must mean it accesses the highest possible level of truth because 'science'. This is unbelievably moronic and, unfortunately, symptomatic of a lot of thinking today.
We can see evidence of this malaise shot through every aspect of our lives. From elaborate psychometric testing to bizarre, data driven theories of law enforcement, the absolute bane of poorly interpreted statistics and numbers in general in politics, and the bizarre and occasionally insane conclusions of data models in genetics, linguistics, and urban planning - data would appear to be the new god. Don't get me wrong - statistical analysis and sufficient data to do it with are vital components of any scientific or theoretical inquiry, but the fundamental component of all of this is humans. Data doesn't think. And if we try to make it do our thinking for us instead of using it as it should be used - to validate or check human thinking - we risk becoming as stupid as the machines we make and use. And that, if you actually think about the last time you asked a computer a question, is pretty damn stupid.