Complex systems are characterized by entities following fixed rules, and when these entities adapt, the system becomes a complex adaptive system (CAS) [1]. CAS possess properties greater than the sum of their individual parts, cannot be fully understood by analyzing components in isolation, and often self-organize without central control [2]. These systems are deeply interdependent, meaning that interventions can lead to unpredictable, nonlinear outcomes and unintended consequences [3-6]. They also have “memories,” meaning they are influenced by past events, and can learn and change in response to new information [2, 7].

Here are several mechanisms and mental models that describe changes in complex systems, drawing from the provided sources:

1. Dynamics of Change and Adaptation

• Evolutionary Processes (Natural Selection, Extinction, Adaptation Rate, Red Queen Effect): Systems, like organisms, must adapt to changing environmental demands to survive [8, 9]. Adaptation is a continuous process driven by environmental pressures and competition [9-12]. The Red Queen effect illustrates that constant adaptation is necessary simply to maintain a competitive position, as standing still equates to falling behind [10, 13]. Furthermore, exaptation describes how existing traits or components can be repurposed for new functions, offering a mechanism for innovation and rapid adaptation without needing to start from scratch [14, 15].
• Creative Destruction: This model explains how existing structures or systems are disrupted and replaced by new, often superior, innovations or ways of thinking [16, 17]. In economic contexts, it’s an evolutionary process where markets act as ecosystems, and firms must adapt to new, more efficient competitors or face obsolescence [18, 19]. Creative destruction is a ceaseless process that fundamentally alters how societies are organized [20, 21]. In science, this is analogous to paradigm shifts, where anomalies lead to the replacement of old theories with new ones, advancing understanding [22-24].
• Feedback Loops: These are fundamental mechanisms where the output of a system cycles back as an input, continuously refining and improving the system [25, 26]. Balancing feedback loops work towards maintaining stability and equilibrium, such as a thermostat regulating temperature [27]. In contrast, reinforcing feedback loops amplify change, leading to exponential growth or rapid decline, as seen in fashion trends or poverty cycles [28]. Understanding these loops is crucial for directing system changes and monitoring their impacts [26].
• Equilibrium and Homeostasis: A system in equilibrium is in a stable state where all forces are balanced, often dynamically, with continuous adjustments within a certain range [29, 30]. Homeostasis, a biological concept, describes a system’s capacity to maintain stable internal conditions despite external changes, often returning to a functional “good feeling” state rather than necessarily its original state after a disturbance [31-33]. Short-term deviations are frequently necessary to achieve long-term stability [34].
• Inertia: This model explains that systems, including human behaviors and beliefs, resist change [35, 36]. Overcoming this resistance requires sustained force and effort, and the longer a system or habit has existed, the greater its “mass” and thus its inertia [37-39]. Inertia helps to understand why new ideas or behaviors often face significant resistance [40, 41].
• Tendency to Minimize Energy Output: All living beings, including humans, instinctively conserve energy, which can lead to resistance to change or risk-taking behavior [42, 43]. This “least-effort principle” influences how environments are designed and how individuals form habits [43-45].
• Critical Mass: This refers to the point at which a system is poised to transition from one state to another, where a seemingly small additional input can trigger a disproportionate and self-sustaining change [46-48]. It highlights that accumulated effort is necessary to reach a tipping point, after which change can propagate rapidly [49]. For social systems, targeting opinion leaders can achieve critical mass more quickly [50, 51].
• Emergence: This occurs when systems, viewed at a macro scale, exhibit capabilities or behaviors that are not present or predictable from their individual micro-scale parts [52, 53]. A key feature is self-organization, where parts follow simple rules without centralized control, leading to complex collective behavior [54]. Cultural learning, for example, produces a “collective brain” that allows human knowledge and technology to accumulate and advance far beyond what any individual could achieve [55-57].

2. Influences and Constraints on Change

• The Map Is Not the Territory: This model reminds us that our abstract representations (maps) are not reality (the territory) itself [58-60]. Since reality is dynamic, maps must be continually updated based on new information and experience; otherwise, we risk making poor decisions and failing to adapt to a changing environment [59, 61, 62]. When the map is mistaken for the territory, efforts to simplify complex realities can lead to negative, unintended consequences [63, 64].
• Second-Order Thinking: This involves deliberately considering the consequences of the immediate consequences of an action or decision [65]. A failure to engage in second-order thinking frequently leads to “unintended consequences” that exacerbate existing problems or create new ones, profoundly shaping how systems evolve [65-68].
• Bottlenecks: These are the slowest or most constrained parts of a system that limit its overall output [69]. Identifying and strategically addressing bottlenecks is vital for improving system flow and can stimulate efficiency and innovation. Removing one bottleneck will inevitably reveal another as the new limiting factor [70-73].
• Scale: Systems change in fundamental ways as they scale up or down; what is effective at a small scale may not be at a larger one [74]. Growth often introduces increased complexity, new problems, and unanticipated outcomes, necessitating a re-evaluation and re-engineering of processes [75-77].
• Friction and Viscosity: These forces impede movement and slow down progress within systems [78, 79]. Reducing friction can significantly enhance productivity and facilitate change, whether in physical systems or in organizational processes and innovation [78, 80-82].
• Incentives: Incentives are powerful drivers of behavior, guiding actions towards rewards and away from punishments [83, 84]. To effect change within human systems, altering the underlying incentives is often necessary [84]. Aligning incentives is a critical leadership challenge in directing collective action [84].
• Law of Diminishing Returns: This principle states that outcomes are nonlinear; beyond a certain point, additional inputs into a system yield progressively smaller improvements [85, 86]. In complex societies, increased complexity can eventually cost more in energy and resources to maintain than the benefits it provides, potentially leading to disintegration [87, 88].
• Chaos Dynamics (Butterfly Effect): Chaotic systems exhibit extreme sensitivity to initial conditions, meaning minuscule differences can lead to vastly divergent and unpredictable outcomes over time [89-91]. This implies that precise, long-term predictions in such systems are inherently difficult, and surprises should be anticipated [90, 92].
• Irreducibility: This concept suggests that some systems or ideas cannot be broken down into smaller parts without losing their essential qualities or emergent properties [93, 94]. Recognizing these irreducible limits is crucial when designing or attempting to change systems, as further reduction would alter their fundamental nature [95, 96]. Gall’s Law, a related principle, advises against building complex systems from scratch, as they invariably evolve from simpler ones [97].
• Multiplying by Zero: In a multiplicative system, if any single component contributes “zero” (is completely dysfunctional or absent), all efforts in other areas will ultimately yield no results [98]. To achieve positive change, this foundational “zero” must be identified and addressed first [99].
• Surface Area: This represents the extent of a system’s contact or interaction with its environment [100]. Increasing surface area can foster creativity and innovation by increasing exposure to diverse ideas and information [101, 102]. However, a larger surface area also increases vulnerability and the energy required for maintenance [100, 103, 104].
• Setting: The environment in which actions occur significantly influences what can happen and the choices made [105, 106]. Changing the environment can be a powerful mechanism for changing behavior within a system [107].

3. Tools for Understanding and Navigating Change

• First Principles Thinking: This is a method of breaking down complex problems to their most fundamental truths, separating what is known from assumptions, to build new solutions [108-110]. It encourages challenging the status quo to find innovative paths [111, 112].
• Thought Experiment: These are imaginative devices used to investigate the nature of things, evaluate potential consequences, and explore alternative scenarios [113, 114]. They help clarify thinking, reveal hidden assumptions, and uncover unintended consequences by simulating reality [115].
• Probabilistic Thinking: In complex and uncertain systems, this tool helps to estimate the likelihood of various outcomes, thereby improving the accuracy of decisions [116]. It necessitates continually updating beliefs as new data becomes available [117].
• Inversion: This approach involves thinking backward, such as asking what would guarantee failure or what prevents a goal from being achieved [118, 119]. This can reveal overlooked obstacles and lead to simpler, more effective solutions that traditional forward-thinking might miss [119, 120].
• Occam’s Razor: This principle advocates for preferring simpler explanations over complex ones, especially when they have comparable explanatory power [121, 122]. It helps in avoiding unnecessary complexity and focusing on more robust solutions, though it acknowledges that some truths are inherently complex [123-125].
• Alloying: This model refers to combining different components (e.g., skills, ideas, people) to create something with enhanced properties, making the whole greater than the sum of its parts [126, 127]. It’s a valuable approach for innovation, team building, and strengthening knowledge bases [128, 129].
• Learning as a Margin of Safety: Engaging in continuous learning and actively reducing blind spots creates a buffer against unforeseen events and enhances a system’s ability to adapt to changing circumstances [130-132].
• Churn: Understanding that system components constantly wear out and are replaced is key [133]. When managed effectively, deliberate churn, like regular turnover in an organization, can inject fresh ideas and boost adaptability, preventing stagnation [134-136].
• Algorithms: These are clear, step-by-step instructions that reliably transform inputs into outputs [137, 138]. They are instrumental in organizing systems, identifying effective inputs, and scaling solutions [138-140]. Some algorithms are designed to evolve and learn over time [139].
• Randomness: Embracing true randomness, rather than imposing artificial order, can make systems less predictable and more creative. It’s a useful tool for problem-solving and generating new ideas, especially when conventional approaches are blocked [141-145].
• Equivalence: This model highlights that different things can achieve the same outcome, meaning there are often multiple paths to success [146, 147]. It’s particularly useful when traditional solutions are no longer viable, encouraging the exploration of alternative, yet equally effective, approaches [146].
• Global and Local Maxima: This model helps determine if a system has reached an optimal state (a peak) and whether there is potential for further, greater improvement (a higher peak) [148]. It implies that sometimes, a temporary decline (moving through a “valley”) by changing fundamental structures is necessary to achieve a significantly better outcome, rather than just fine-tuning within a current, sub-optimal state [149-151].