Back

This is such a good summary of "the problem with 'data science'", as it tries to replace any domain expertise with the misguided notion that some form of statistics.

And so you end up with (maybe even well-meaning) researchers that are so far in over their head that they don't even know what they don't know.

Case in point: the authors of the preprint come from mathematics, computer science, physics, and 'future studies' (lol)…

@theluddite

@gedankenstuecke It's so frustrating too because the existence of large datasets and the computation to look into them are new phenomena that call for new techniques, which could offer genuine insight, but in conversation with existing work, not replacing it!

It's the perfect extension of the mindset of software development in general. Computers are a fucking miracle and somehow I spend all my time annoyed at everyone who uses them.

@theluddite yes, the problem is that those large data sets are in a way 'too accessible', in the sense that everyone with a half-baked understanding of how that data was gathered and what limitations it might have can now write up their findings.

When I was a biologist I used to complain about physicists trying to do biology without understanding biochemical limitations. Now I'd love if they'd stick to that instead of doing fake social sciences!

@theluddite Instead, any time that someone tries to point out the mistakes made when dealing with such data it's pushed aside because 'look at how much data they had', as if garbage gets better just because the pile is bigger.

@gedankenstuecke amen! It's even funnier in the social sciences too. Imagine thinking that you could understand "misinformation," aka politics, using fucking statistics on Twitter data without engaging with hundreds if not thousands of years of theorizing about exactly this from some of humanity's best and brightest. They're out here labeling things as true or false without a care in the world. Have you never even heard of philosophy??? Good god.

@theluddite yeah, but of course the people most likely to engage in such a superficial way have never thought of that history they could make use of. Partially because they never come to think of other disciples having a tradition too. And maybe even more problematic, because Most researchers are never taught the history of their own domain even

@jaranta @theluddite the good tech isn't dead yet!