I think one of the hard things about academia (by ways of "pure research"), is measuring impacts. In some positions there are metrics which are more representative of real progress than others. My belief is that measuring an academic's "impact" is a very fuzzy thing in the long term, especially when you bring in people who don't specialize in the same field. That's a reason this can't be generalized fully, though you can probably find corollaries.

But, your last sentence to me, seems like a good one. Presumably, that would be an excellent topic to pursue research in, considering we clearly don't have a working theory on it (but that's not my specialty so who knows, maybe there's a bunch of smart people being ignored about it). Thus, it's hard to measure progress in, and we institute perverse incentives per the article.