it is better to be vaguely right than precisely wrong
j.g. UT actuarial science
To some degree our perspective, our existence, approach influences the outcome though we might oversimplify our view that we are objectively measuring an independent phenomenon. And so not to butcher this completely and tie back, we can generally parameterize the behavior while not being able to discern all the interrelationships that effect the observed phenomena. More to it but that was my shallow recollection of the concept. Error being unquantified factors and interrelationships.
Einstien hated it but could not dismiss it. It is the freaky unfinished science. Freaky because we can't fully explain the behavior but we know it exists.
Ultimately, I'd say we have to look at the best evidence thoroughly and pursue the least disruptive course whatever that might be. It really can't be a purely quantitive because the models will never spit out the exact answers and I'm not sure we have to. Instead, we would have to make a human assessment of what makes the most sense. Balance short term realities with long term risk mitigation.