It really depends on the particular group of algorithms. I'm only considering non-cryptographic algorithms here. As a general rule, any algorithm that involves

It really depends on the particular group of algorithms. I'm only considering non-cryptographic algorithms here. As a general rule, any algorithm that involves a hash or a random/arbitrary choice has historically been based on "assume no adversary" and even now it has only advanced to "assume an incompetent adversary". By contrast, most tree-adjacent algorithms have always been vigilant against competent adversaries.