My proposal is that the referee ought to:
1) Keep track of recurring questions that the PCs ask them
2) Decide whether these have interesting but uncertain answers
3) Establish rules that balance PC agency and randomness for the ones that do
4) Explicitly dismiss the ones that don't have interesting answers as irrelevant to play
This sounds simple and yet is widely ignored in actual rules design. For example, one can find dozens of minor tweaks of how combat works in various versions of D&D, but very few designers have developed anything covering "Can we find a good spot to camp for the night?" One might say that's not an interesting question, though that position is difficult to maintain if one is also rolling encounter checks overnight.
A few years ago, I created my own rules for chases because I found that despite the frequency that chases appeared both in games and in related material, no one really wrote rules that covered them very well (a few people had written rules for fleeing that resolved doing so in a single roll and were mainly intended to be used by the PCs when fleeing monster).
On particularly appalling oversight is lacking teamwork rules. "Can I help?" is such a common question PCs ask, and yet anything other than a blind methodological individualism is uncommon. Long-time readers will remember that I was shocked to discover that the One Ring game doesn't even have rules for PCs to help one another with ordinary tasks. I've spent a lot of time coming up with teamwork rules for Mythras, Openquest, and Into the Depths because their absence in other games drove me up the wall.
I think there are many more examples of recurring questions at tables that have no rules or procedures to resolve them that probably should, and that effort to design new rules would probably be much better spent on them than elsewhere.