Comment by strken | Hacker News Reader

Comment by strken | Hacker News Reader

Parent Full thread

I've been reading through a beginner-level archaeology textbook that devotes multiple chapters to explaining, with examples, that absence of evidence is not evidence of absence.

One of the biggest mistakes that people tend to make is looking at the current population and using it to fill in the gaps in the archaeological record. An example my book gives is Lake Mungo in Australia, where researchers tried to demonstrate that life was much the same in the prehistoric period as it was in the historic period just after British colonisation by proving that grindstones were in use in both the historic period and 15000 years ago, and thus that food sources hadn't changed. This was demonstrated to be wrong by later findings showing changes to technology and food sources over time, particularly in response to conditions of abundance due to higher rainfall during the last ice age.

In this case the factual evidence is (paraphrased) "The village of Pingliangtai had no big houses, no obvious social hierarchy in burials, and a system of ceramic drainage pipes". It's tempting to read this as evidence for having no hierarchy; it's also tempting to think about the obvious ways in which a hierarchy could exist without leaving any record, like a leader who lived one town over, and conclude the opposite; but I think my book would caution against believing either interpretation. Small houses, matching graves, and pipes are the truth and everything else is an educated guess.

'Absence of evidence is not evidence os absence' is a mantra that was instilled in me when I was trying to figure out my uni undergraduate and went with history for a year.

This was also a center point in Historiography, and how, when analysing old documents by historians, one needs to be aware of possible bias, not only of the writers, but also of one self.

In the presence of a model the absence of evidence _can be_ evidence for absence if you have good reason to believe your model is complete. If your model is the oft cited celestial teapot then the absence of evidence is the evidence of absence. But if your model is 1 trillion celestial teapots in orbit around the sun than absence of evidence _is_ evidence of absence.

People would do well not to rely too heavily on this asinine little phrases. They make us feel smart, sure, but they are very simple minded.

The specific issue with archeology is destruction can be selective. You can’t tell which towns failed to keep marriage records based on the lack of surviving records when the place those records ended up could get destroyed in a fire.

Is a structure missing because it didn’t exist, or is it missing because it was made out of useful materials repurposed for something else.

Similarly people dig up the remains of ancient towns because they got buried, if the location had erosion instead there would be nothing obvious left to investigate.

Not to mention that there's only so much the layout of ground floors can tell you about social organization of a civilization which isn't described in written records, even if relatively evenly preserved. As in the famous "Motel of the Mysteries" parody, in which an archaeologist of the future discovers the underground remains of a 20th century motel and ascribes probable religious significance to everything from the television to the toilet.

I mean, "lots of houses, all of them quite small and all nearby burials without ceremony" could describe a hierarchy free anarcho-syndicalist utopia. But it could also describe a monastery, a barracks, a ghetto, a town built for subjects of the undiscovered nearby imperial palace or a slave colony, all of which are usually associated with particularly rigid hierarchies. If apartheid era Soweto was buried under a volcano, it would probably look like a settlement organised on relatively egalitarian lines by late twentieth century standards...

Archeology is also limited by the things that will survive and what access to them the people had. If the people didn't have access to metal you will find very little metal tools (what you do find will also be used until worn out and then remade into something else; or restricted to the very rich. Either way limiting the usefulness of what evidence we find.)

> if you have good reason to believe your model is complete.

When in archeology is that ever true? Frankly, outside of archeology when is that ever true?

> One of the biggest mistakes that people tend to make is looking at the current population and using it to fill in the gaps in the archaeological record.

I think this is good to keep in mind but let's be real, humans have tended to make hierarchy be one of those things they push backwards in time. So yeah, it's best to be as impartial as possible but in this case, the authors are the ones contradicting the common anachronism.

> absence of evidence is not evidence of absence

Wrong mathematically, at least if you care to define A as evidence of B to be the same as P(B|A) > P(B|not A), and intuitively if you think about fairies and other things. But sure, it's right if you instead think of evidence as being the same thing as proof, or an inequality as expressing an insurmountable magnitude. As a way to guard against hubris I can understand the repetition.

There's probably a mathematically rigorous version of the proverb along the lines of "low-confidence observations from a limited number of data points are not always sufficient to update other actors with priors different from yours sufficiently far towards your position for the optimal status-seeking strategy to be for you to give an expression of high confidence", but it wouldn't roll off the tongue quite so well.

In the world of math, that IS rolling off the tongue well

You sort of touch on it, but to be explicit:

The person you're replying to is not using the same meaning of "evidence" as you are. You're meaning it in the strict sense of "holding all things equal, increases probability estimate of perfect bayesian reasoner" while they're using "given actual sociological conditions amongst the advanced monkeys and the eay they interpret words and labels, this should not be given the label Evidence".

I find it baffling how some archaeologists draw detailed narratives of ancient life from just one, often ambiguous, piece of evidence.

Early dinosaur reconstructions (and paleo human ones) are similarly baffling. Many current ones too.

>One of the biggest mistakes that people tend to make is looking at the current population and using it to fill in the gaps in the archaeological record.

That how the Europeans never found the Amazon lost cities, because earlier explorers were looking for cities like European cities.

Which Amazon lost cities do you mean?

There have been many ancient cities in the Amazon rainforest discovered through satellite imaging/LIDAR/other techniques, but before those discoveries large settlements in the area were considered legends and myths.

my city, the amazon driver for some reason can't find my house number and keeps dropping off the package at the post office.

> absence of evidence is not evidence of absence.

But it is. It's not _proof_ of absence, but it's certainly _circumstantial_ evidence of absence.

> people tend to make is looking at the current population and using it to fill in the gaps in the archaeological record.

Well, ok, but that's not the same as opposing the principle you cited.

> tempting to read this as evidence for having no hierarchy

You're again switching up arguments. What you're demonstrating is that evidence to the absence of some expression of X is weaker than evidence for the absence of any aspect of X. (And at the same time conceding that absence of evidence is evidence of absence.)

Reply on news.ycombinator.com