Each recursion can effectively manage only what it can effectively model.
The variety (choices) that must be dealt with is too large for any model or institution. So every institution will be dealing with an inadequate model. However, others may have the missing models and therefor the missing choices. The big question is how should/can different recursions with different models work together?
**My "solution"** is to have representatives from each essential recursion include members from other recursions, who learn the meta concepts, meta languages and meta systems of the other.
Now the problem resolves to the question: how can decision makers at any particular level trust those among them who hold an "outside" understanding, not understandable by the decision makes, yet allow them to influence decisions.