Admittedly, this is because "homo economicus" is sophomoric in the first place: it proposes an actor who has no computational or informational bounds and sees the world only in terms of a single real-valued variable called "utility", which is generally constrained only to be any monotonic, continuous function of their monetary net worth. This actor then optimizes expected utility with no regards to modeling causal structure or to which observables are actually ergodic.
No such creature ever has, or in fact ever can, exist in the real world. In limited domains where ergodicity holds, computational needs are few, and small sample sizes can yield very good inferences, we can do thought experiments about what such a creature would do, and act according to those if we please. But holding that such behavior is the normatively correct way to act when it's actually impossible is the kind of religious thinking one gets when composing "normative theories" without experimental basis.
Humans are not merely "irrational", they positively behave and think in very specific ways. No human is immune to this. It is, therefore, perfectly rational to discuss rational ways of managing the specific human behaviors that lead to irrational behaviors.
This is also while ritual denunciations of "homo economicus" are really quite sophomoric... the nonexistence of rational economic actors does not mean that you can simply model humans in whatever way you please to make your preconceived notions work. It means that you've got to go learn what humans actually do, which is going to be very challenging, and take that into account, which is almost certainly going to shock anyone who tries it. It means the problem is suddenly immensely harder, not easier.