I would have liked a more bayesian approach to the steelmanning: for example there is nothing wrong per se in starting with a hypothesis (”the conclusion”, as you frame it), and movinh forward from that to seeing how that explains and predicts the evidence. That is in fact exactly what you should be doing! The problem is there only when you (1) overestimate the priors for your favorite hypothesis, (2) are not rigorous in the way you evaluate the likelihoods and convert them into posteriors, or especially (3) don’t take into account the evidence that goes against your hypothesis.
I don’t seem to understand your argument from missing pottery or writing systems:
Assuming that civilizations tend to first grow around agriculture in river valleys, near the shore, and river valleys are by definition planes with very little elevation gradient, and first cities tend to grow on the banks of these rivers in the valley, we would expect most if not all of their heritage to be concentrated at the low altitude shoreline. This includes writing, if any would have survived (are we assuming they must have preferred clay tablets and stone walls?). This would also include most if not all pottery.
Göbekli Tepe being built by hunter gatherers is to me an extraordinary claim. How should this ”extraordinarity” be treated in a proper bayesian evaluation? By giving the ”hunter gatherers built it” a low prior probability, and giving the ”agricultural surplus culture built it” a higher prior probability, as this is in our uniform experience the way monolithic structures come into existence. You really need quite a lot of man power to do that, which means some kind of a surplus.
But is ”hunter gatherers built it” a scientific hypothesis? How would you falsify it?
I would have liked a more bayesian approach to the steelmanning: for example there is nothing wrong per se in starting with a hypothesis (”the conclusion”, as you frame it), and movinh forward from that to seeing how that explains and predicts the evidence. That is in fact exactly what you should be doing! The problem is there only when you (1) overestimate the priors for your favorite hypothesis, (2) are not rigorous in the way you evaluate the likelihoods and convert them into posteriors, or especially (3) don’t take into account the evidence that goes against your hypothesis.
I don’t seem to understand your argument from missing pottery or writing systems:
Assuming that civilizations tend to first grow around agriculture in river valleys, near the shore, and river valleys are by definition planes with very little elevation gradient, and first cities tend to grow on the banks of these rivers in the valley, we would expect most if not all of their heritage to be concentrated at the low altitude shoreline. This includes writing, if any would have survived (are we assuming they must have preferred clay tablets and stone walls?). This would also include most if not all pottery.
Also the approach to extraordinary claims requiring extraordinary evidence is muddled up, as we would expect, given this stuff comes from Hume, who quite frankly is horrible when talking about anything approximating probability calculus. (See for example https://www.researchgate.net/publication/301854680_A_Bayesian_Baseline_for_Belief_in_Uncommon_Events)
Göbekli Tepe being built by hunter gatherers is to me an extraordinary claim. How should this ”extraordinarity” be treated in a proper bayesian evaluation? By giving the ”hunter gatherers built it” a low prior probability, and giving the ”agricultural surplus culture built it” a higher prior probability, as this is in our uniform experience the way monolithic structures come into existence. You really need quite a lot of man power to do that, which means some kind of a surplus.
But is ”hunter gatherers built it” a scientific hypothesis? How would you falsify it?
Cheers!