The key issue is “structural coupling.” When connections are too tightly coupled, then changes in one part of the network cascade into other parts of the network (banks, forest fires, power grids, etc.) These cascades typically leave the system better off (Perrow called these “normal accidents” in his work on high-risk technological systems). For example, yes, disease can spread more quickly but so can immunity. Conversely, systems that have very weak connections can scarcely be called systems in the first place and they suffer from a host of other issues.

In the middle, “small pieces loosely joined” (Weinberger) maintain adaptability by being able to 1) respond to cascades when they are positive, and 2) resist cascades when they are negative. Sometimes this is referred to as “highly optimized tolerance (HOT)”. In world politics, Keohane and Nye propose concepts like “sensitivity” vs. “vulnerability” on an axis of susceptibility (also used in epidemiological network studies).

This is not to say that Collapse is not a real possibility (e.g. Jared Diamond’s “Collapse” c.f. Joseph Tainter’s “Collapse of Complex Socieities” or a conference paper by Norm Yoffee and I for the Santa Fe Institute “Collapse in Early Mesopatamian States: What Happened and What Didn’t” (http://www.santafe.edu/events/workshops/index.php/The_Co-Evolution_of_Behaviors_and_Instututions).

Stu Kauffman and others tackle the bigger mystery of how networks and complex adaptive systems seem to tune themselves to the point of maximum robustness, i.e. the oft-touted “edge of chaos.”

Panarchy is just far more complex than simple utopian or dystopian visions can adequately describe.

Let’s not forget that the human brain is the most hyperconnected living system we know of (per unit area) and it _can_ be very resilient.

The key to looking at complex living systems that perpetuate through ‘build up and sudden collapse’ is the theory of “self-organized criticality.” While a massive and sudden collapse is indeed made possible because of the hyperconnectivity, such a collapse is essential to keeping the system stable and it does not result in a new system or new design.

So I think the author is taking the ‘robust response to tiny triggers’ to mean a lack of resilience, which is wrong.

Marc