I think that coupled systems of positive and negative feedbacks "naturally" become cybernetic systems. Any part X of the system that "learns" to react to neighbouring inputs in ways that increases stability of X, logically leads to increaed autonomy of X. Subsystems may "learn" to control their own growths and ocsillations as well. "Learning" may mean just stumbling upon a functional configuration of feedabcks (that then tends to preserve itself), or it may mean development of a cybernetic-mechanical mechanism to copy past reactions in certain distinguishable situations. Once a perception-reaction network is established, the system has a "mind" of its own.

Subsytems with expressive functionality put more structure to the whole system - positive feedbacks will be controlled or exploited by "someone", cooperative and competative feedbacks between functional subsystems will gradually occur. Self-preserving functionality of the whole system may emerge.

In the light of this, we should not assume that our destructive growth can physically continue until it crosses boundaries of its own validity - the boundaries may "come" forward themselves. The positive feedback of our run-away civilization should provoke a reaction of the Earth system. Do we see it? Is the explosion of our civilization so unique that the Earth system, even if existing as a smart self-preserving Gaia, "would not know" what to do with it? Or is human impact already so great that an unmistakable stress signal is already received by a core network of geological/atmospheric processes, but the reaction will take some time (a decade or so)?