Complex change is often described as "evolutionary" in economics, policy, and technology, yet most system dynamics models remain constrained to fixed state spaces and equilibrium-seeking behavior. This paper argues that evolutionary dynamics should be treated as a core system-thinking problem rather than as a biological metaphor. We introduce Stability-Driven Assembly (SDA) as a minimal, non-equilibrium framework in which stochastic interactions combined with differential persistence generate endogenous selection without genes, replication, or predefined fitness functions. In SDA, longer-lived patterns accumulate in the population, biasing future interactions and creating feedback between population composition and system dynamics. This feedback yields fitness-proportional sampling as an emergent property, realizing a natural genetic algorithm driven solely by stability. Using SDA, we demonstrate why equilibrium-constrained models, even when simulated numerically, cannot exhibit open-ended evolution: evolutionary systems require population-dependent, non-stationary dynamics in which structure and dynamics co-evolve. We conclude by discussing implications for system dynamics, economics, and policy modeling, and outline how agent-based and AI-enabled approaches may support evolutionary models capable of sustained novelty and structural emergence.
The term evolution is widely used in economic, policy, technology, and organizational studies to describe long-term change. Firms are said to evolve, markets evolve, institutions evolve, and strategies evolve. However, in most system dynamics models (Meadows, 2008), the mechanisms underlying this change remain constrained to fixed state spaces, static transition rules, and equilibrium-seeking dynamics. As a result, such models capture adaptation within predefined structures, but struggle to explain how new structures, strategies, or categories emerge over time. Evolution is invoked descriptively but rarely implemented mechanistically (Nelson & Winter, 1982). This tension reflects a broader issue in scientific explanation. Idealized models (Cartwright, 1983;Mitchell, 2009) often succeed by deliberately removing history, path dependence, and pop-
Figure 1 illustrates how broadly and persistently the term “evolution” is used across domains.
Using Google Trends data over a five-year period, Figure 1(a) shows sustained interest in domainspecific phrases in biological, technological, cultural, linguistic, chemical, and astronomy contexts. Despite this widespread usage, the term is rarely accompanied by a clearly specified dynamical mechanism outside biology: there are no explicit genes, no genotype-phenotype distinction, no mutation or recombination operators, and no inherited replicators that transmit information across generations. Instead, “evolution” functions as a descriptive shorthand for gradual or pathdependent adaptive change in systems. While such usage is often intuitive and rhetorically effective, it leaves unresolved what, if anything, is being selected and how selection is implemented dynamically.
In socio-technical contexts such as technology, culture, language, and economics, evolutionary change is commonly attributed to diffuse processes including competition, learning, imitation, or adaptation to external pressures. These processes can generate historical contingency, but they are rarely formalized in ways that allow population composition to reshape the space of future possibilities. As a result, evolutionary language typically describes outcomes rather than mechanisms, offering narrative coherence without a corresponding population-level dynamical account.
Even in formal scientific and mathematical settings (Nowak, 2006), the evolutionary terminology is frequently used analogously. Evolutionary game theory, for example, adopts the language of strategies, fitness, and selection while generally assuming fixed strategy spaces and static payoff structures. The resulting dynamics describe changes in relative frequencies under predefined rules, not the endogenous emergence of new strategies, representations, or categories.
These usage patterns reveal a consistent gap between the informal evolutionary discourse and formal modeling practice. Evolution is invoked to describe long-term change, yet the underlying models remain constrained to fixed state spaces and equilibrium-seeking dynamics. This gap motivates the need for a general evolutionary systems framework capable of capturing open-ended, non-equilibrium dynamics driven by endogenous population feedback rather than predefined optimization criteria.
Stability-Driven Assembly (SDA) was previously introduced (Adler, 2026b) as a minimal nonequilibrium framework in which differences in persistence bias the accumulation of structure over time, with complete mathematical derivations and symbolic simulations demonstrating entropy reduction and scaffold emergence. A companion paper (Adler, 2026a) extends SDA to chemical symbol space using SMILES molecular fragments, showing hallmark features of evolutionary search like scaffold-level dominance and sustained novelty over thousands of generations. The term “Assembly” here refers to pattern formation through combination, distinct from Assembly Theory (Sharma et al., 2023), which addresses molecular complexity measurement rather than populationlevel selection. Rather than reproducing those technical results, this section focuses on the core dynamical mechanism: how persistence-weighted feedback reshapes population distributions and induces evolutionary search, treated as a general process applicable across domains.
An SDA system consists of a population of interacting entities generated through stochastic interactions and removed through decay in an open-flow setting. The outcomes of the interaction differ in stability, expressed as a finite persistence time. Patterns that persist longer remain in the population for more generations, become more abundant, and therefore participate more frequently in subsequent interactions. Decay is implicitly modeled by removing expired patterns, while patterns with zero stability are created and eliminated within the same generation, representing interactions that cannot persist under prevailing conditions. This asymmetry in persistence introduces selection without
This content is AI-processed based on open access ArXiv data.