It is widely believed that an equilibrium framework based on simple models, such as the

representative agent and bilateral bargaining, in a frictionless economy provides a consistent

framework in micro, macro, finance, and institutional economics. However, equilibrium

believes in self-stabilizing market and institutional convergence broken down when inherent

instability and non-predictable uncertainty emerge under nonlinear and social interactions.

Equilibrium illusions in economics and econometrics are pretty in math modeling but

dangerous in policy decision. Known examples are the Frisch model of noise-driven cycles,

the Lucas model of microfoundations, and the Coasian world of zero transactions. These

models not only violate basic laws in science but also lack evidence in economic history.

Their common problem is associated with linear Hamiltonian economics with symmetric

information without history. Economies are dissipative systems in nature, characterized by

symmetry breaking, information flow, a time arrow, and history. The many-body problem is

fundamentally different from the one-body and two-body problem in mathematics. Both

computational and natural experiments, such as a stock market crash and a transitional

depression, reveal the severe limitations of equilibrium thinking and structural changes from

evolutionary dynamics. The new science of complexity offers new tools of nonlinear

dynamics and non-stationary time series analysis. Existing puzzles in equilibrium

economics, such as persistent cycles, interruptive crises, market resilience, social

movements, and organizational diversity, can be better understood by nonlinear dynamic

models. Like the paradigm shift after Einstein in physics, the evolutionary perspective

provides a general framework, while equilibrium models serve as its special cases, since the

equilibrium picture is an approximation of economic complexity in a short-time window

taken from a long-term historical current.