Complex systems have a way of reproducing themselves quite agnostic of the methods we use to influence or shape them. Where we assert causal ontologies we tend to do so under the scope and rubric of many prior assumptions. Data is a function of observations that tend to reproduce and self-validate the assumptions by which measurements are made.
If we are to adhere to simplest explanations and definitions we will only ever reproduce the low-hanging fruit that validates the language(s) used in defining them. Necessity overrides everything else, of course, but from within language and the hyper-inflating matrix of conceptual vocabularies that such pragmatism inhabits, it will often fail to see those things which might actually maximise choice or any other aspiration to utilitarian benefit.
All causal ontologies are complex tautologies. If it were not that human subjectivity and group psychological momentum was built upon the constructive obfuscation of the uncertainties that anchor it upon language as systems of belief, we might notice it.
So, we are bound to inhabit the shallow waters of median description because they are effective, but it is a deeply conservative and utterly brittle strategy as philosophy of life.