> What value is there in "research" that doesn't do anything weird?
Plenty of value, I think. It's e.g. not weird at all to research very mundane things like "can we track as many diabetic people as we can through 10, 15, 20 years and see what other health outcomes happen". Or something like "let's test if this chemotherapy drug is more effective than this other chemotherapy drug".
Sure there is radical, weird research, like rock licking or maybe some sociological study on fetishes. But I think there's space for both the weird and non-monetizeable research (think: 16th century lacemaking in a european country that doesn't exist anymore) and the monetizeable research (some incrementally improved LLM model probably).
By the bar people are calling research "weird" right now, no, "can we track as many diabetic people as we can through 10, 15, 20 years and see what other health outcomes happen" doesn't qualify.
What qualifies is "can we track as many diabetic people as we can through 10, 15, 20 years and see what other health outcomes happen by using only the same procedure and data other people have been using for a while so that it has been on the news more than a few times".
And yeah, you still have a point in that it may still be useful. But it's either something people blatantly ignored for some reason, or it's not really research, but product development.
I think metadata studies are pretty useful actually. Broad questions like "what medical conditions does diabetes cause" can be investigated in that way. It's definitely not weird at all. But when people say "weird" I think people mean things that don't have immediate economic use, for example, a literature phd that is on the depiction of black hair in books from countries with black majorities vs books from countries with black minorities.
Plenty of value, I think. It's e.g. not weird at all to research very mundane things like "can we track as many diabetic people as we can through 10, 15, 20 years and see what other health outcomes happen". Or something like "let's test if this chemotherapy drug is more effective than this other chemotherapy drug".
Sure there is radical, weird research, like rock licking or maybe some sociological study on fetishes. But I think there's space for both the weird and non-monetizeable research (think: 16th century lacemaking in a european country that doesn't exist anymore) and the monetizeable research (some incrementally improved LLM model probably).