Beyond construction of ordered, equilibrium structures, one of the grand visions of self-assembly research is to organize matter away from thermodynamic equilibrium. [1] The inspiration for such dynamic/non-equilibrium self-assembly, [2] NESA, comes largely from biology [2c,d] (with cells and organisms being perfect examples of non-equilibrium assemblies), while its promise lies in the new types of dynamic materials [3a-d] and chemical systems [3e-g] that could harness and dissipate the externally delivered energy to sense and adapt to the environment, reconfigure, exhibit taxis, or even selfreplicate. Despite recent progress, [2,3] however, NESA remains an experimental challenge and is also in need of unifying statistical-thermodynamic principles. At equilibrium, the Boltzmann relation P(E) / exp(ÀE/kT) provides a powerful and predictive link between the probability P of structure formation and its energy E. In the non-equilibrium (NE) regime, such general principles do not exist, and our understanding of NE thermodynamics is largely limited to the so-called near-equilibrium conditions. [4] There, the formalism developed by Prigogine some seven decades ago [4c] prescribes that the structures that are formed minimize the entropy production dS/dt and the rate e at which energy is dissipated (see Methods). Although the validity of this minimal entropy production (MEP) rule has been questioned, [5] and Prigogine himself clearly emphasized that it applies only "sufficiently close to equilibrium," his Nobel Prizewinning work has subsequently assumed life of its own, with frequent interpretations [6] suggesting that Nature generally forms minimally dissipative structures and systems. Indeed, does it? To answer this question requires a rigorous approach. First, a NE system must be implemented that has a "choice" of evolving into more than one dissipative structure. Second, these possible structures should differ only in their rates of energy dissipation, and not in energies or other parameters; if many