mjI'd say that one of the 1st things that constrains which problems are picked, & how those problems are framed is very often the maths.
#Reductionism helps make simplifying assumptions 'doable,' but that usually gets swept under the rug & forgotten by subsequent generations, & calcifies into dogma.
For example, we see linear "solutions" in models that are openly acknowledged to be #nonlinear. These cannot be but provincial, yet they are rarely framed as such. That does not move the collective ball forward.
Lawyers & publishing scientists have something in common: they ask questions they >
> already know the answer to. Now, this is a few steps away from framing the problem, but the bias is clearly aimed at control of the narrative. This is the day-to-day work that pays the bills.
Actual discovery is very often serendipitous, & we know that #emergence is outside known parameters, >
> by definition.
In summary, we're never going to make careers out of saying "I don't know." So most hold their reservations silently & move on.
From this vantage point, we see survivorship #bias & replication crisis. Anyone can look better by removing the errors/critics that were made on the way.