| HN Mirror

Careful how those stats are collected/computed. I don’t think statistics techniques and the methodology that’s used to run these kinds of experiments is anywhere good enough to make such bold conclusionary statements. It sounds nice in theory but in practice social sciences are to other science disciplines (that can make such definitive statements) as tabloids are to news papers. It’s a cheap imitation from the viewpoint of the strength of the conclusions. I hope they get to a better footing eventually but that entire field is way to arrogant right now in the strength of the conclusions they make based on bad science (limited population studies, small effect studies, rarely/never publishing negative results which is a kind of p-hacking en made, lack of reproducibility, etc). That also ignores the fact that epistemologically they have a weak footing of not actually concretely describing their terms (ie happy, sad, good, bad, etc) or how they can measure these things (what is 1 happiness unit?).

Once you start seeing links establishing how to actually measure this stuff with technology, then the results will be more meaningful (easy to replicate experiments, easy to do things en masse more cheaply and quickly).

That being said I still have respect for the work being done. It should still be done rather than wait for better tech but the incentive structures for that field need to be drastically different to better reflect the quality of using that data to make decisions (maybe a generally agreed upon ban to feature it in “public” media/journals, more careful reporting by science journals to limit to the results of meta studies or major reversals of what might be orthodoxy, publishing lists like Math does of what are open problems and what are assumed to be the conclusion even if not proven yet, etc). There are reforms to be made but until that’s done these studies do more harm than good as people latch on to whichever they want that sounds good to them rather than what’s actual reality).