The part that distinguishes complex adaptive systems from other systems is that the system, or the actors within the system, learn. They adapt, and in order to do so, there must be a feedback system that reacts to a payoff. For example, in a city, a payoff can be as simple as the survival of one firm over its competitors, or it can be one competitor gaining an edge over another, or one building use gaining an edge, or one industry gaining an edge, or even one subculture gaining an edge. It’s all about the edge, and once an edge is gained from a payoff, increasing returns often happen, turning that edge into even more of an advantage. This is a very good thing for the actor that has gained the edge.
However, the payoff may pay off too well for everyone. Since every actor is enmeshed in a network that incorporates every other actor, no matter how loosely, as the advantaged actor gains the edge, the rest of the network compensates, often in subtle and sometimes in large ways. So, what is good for one actor is not necessarily what is good for other actors… nor for the system as a whole. What is positive payoff for one may be negative payoff for the system at large, and when that happens, the long-term payoff for the advantaged actor may not be as positive as that of the short-term.
For example, if a simplified city has 3 different industries and one of those industries gains an edge, this may help the city grow. If that edge continues to increase, however, the other 2 industries might be out-competed to the point that they languish and, perhaps, die, which further increases the edge of the advantaged industry. Now, it is a predominantly one-industry city. What happens if that industry begins to stagnate, as monopolies often do? What if actors from that same industry but from a different city begin to gain an edge over the actors of the one-industry city? The whole city then suffers because of the previously wild success of that industry. Detroit is a good instance of this type of self-destructive success.