My steelman is this (without having read anything downstairs, so I apologise if there's a better on extant): the world is a complicated place, and we all form beliefs based on the things we think are important in the world; and since we are all horrible reasoners, it's impossible to believe about some things that they are important movers of the world without seeing it actually happen and viscerally feeling it change things.

Cognitive biases in yourself are like this, methinks. Your thought processes really need to be broken down repeatedly for you to be able to start seeing the subtle shifts happening inside you - and anticipating that they happen even when you don't see them (generalising from many examples here, but not nearly enough).

Another difficult tripping point for me was intuitive reasoning. Till I saw people who couldn't make any sense do significantly better than me I could not possibly believe it, even fighting against people who told me I over-analysed and spoke too much.

I'm slowly coming around on dishonest rhetorical stances, because of the amount of time I've spent trying to convince hostile arguers. Let me soothe the raised hackles of your inner LW-cat by saying that I can't endorse anything like this without finding a Schelling fence,* and am willing to consider anyone who takes such a stance on LW (or in LW-related contexts) evil.

*In fact, based on the world being as it is, I strongly suspect there isn't one.

So, Romila Thapar is not a Dalit activist, just a historian (I'm guessing this is a source of confusion; I could be wrong).

I'm saying they should have read up before starting their project.

I can't find the study for some reason, so I'll try and do it from memory. They randomly picked from a city Dalits (Dalit is a catch-all term coined by B R Ambedkar for people of the lowest castes, and people outside the caste system, all of whom were treated horribly) and people from the merchant castes to look for genetic differences. Which is all fine and dandy - but for the fact that neither 'Dalit' not 'merchant-caste' is an actual caste; there are many castes which come into those two categories. So, assuming a simple no-inter-caste-marriage model of caste, a merchant family from village A thousands of kilometres from village B has about as much (or, considering marginal things like babies born out of rape, even less) genetic material in common than a merchant and Dalit family from the same village - unless there's a common genetic ancestor to all merchant families. And that's where reading historical literature comes in - the history of caste is much more complicated, involving for example periods when it was barely enforced and shuffling and all sorts of stuff. So, they will find differences in their study, but it won't reflect actual caste differences.

Actually, with the caveat that I don't have any object-level research, I doubt it; they assign a rigidity to the whole thing that seems hard to institute. My point was that 'do there exist genetic differences' is not the issue here.

a) Actually Thapar's point wasn't that there were no genetic differences (in fact, the theory of caste promulgated by Dalit activists is that it's created by the prohibition of inter-caste marriage and therefore pretty much predicts genetic differences) - but that the groupings done by the researchers wasn't the correct one.

b) I should actually check that what I surmised is what she said. Thanks for alerting me to the possibility.

When I said humanities I didn't mean social sciences; in fact, I thought social sciences explicitly followed the scientific method. Maybe the word points to something different in your head, or you slipped up. Either way, when I say humanities, I actually mean fields like philosophy and literature and sociology which go around talking about things by taking the human mind as a primitive.

The whole point of the humanities is that it's a way of doing things that isn't the scientific method. The disgraceful thing is the refusal to interface properly with scientists and scientific things - but there's no shortage of scientists who refuse to interface with humanities either, when you come down to it. My head's canonical example is Indian geneticists who try to go around finding genetic caste differences; Romila Thapar once gave an entertaining rant about how anything they found they'd be reading noise as signal because the history of caste was nothing like these people imagined.

And, on the other hand, we have many Rortys and Bostroms and Thapars in the humanities who do interface.

In the abstract. Though, undoubtedly, many of the people can do wonders too.

Why? This looks as if you're taking a hammer to Ockham's razor.