Aggregate results and outcomes cannot be easily predicted from separate, individual actions

From Doing Bad by Doing Good: Why Humanitarian Action Fails by Christopher J. Coyne. Page 149.

Three key implications emerge from an appreciation of systems-type thinking. First, aggregate results and outcomes cannot be easily predicted from separate, individual actions. In a laboratory, a scientist can conduct a controlled experiment, for example combining two atoms of hydrogen with one atom of oxygen to yield water. This is a predictable result that can be replicated in other controlled settings. However, outside of the laboratory, the world is characterized by complex systems that make it impossible to predict the specific outcome of relatively simple actions that are non-additive and non-replicable across contexts. If grasped, this first implication shifts focus from separate actions to the interaction effects between parts of the system. However, this poses new difficulties, because identifying the element(s) responsible for certain outcomes, both in terms of magnitude and direction, can be extremely difficult if not impossible, especially when one remembers that the emergence of effects is often long-term and variable.

The neglect of this implication is evident in the prevalence of what Rory Stewart and Gerald Knaus, of Harvard University’s Kennedy School, term the “planning school” of foreign intervention, which claims to offer a “clear, confident, and unambiguous recipe for success in intervention” based on “clear strategy, metrics, and structure, backed by overwhelming resources.” As discussed throughout this book, it is precisely this type of technocratic linear thinking that dominates state-led humanitarian action.