Entities, Connections, and Organization Through Time
The Foundational Perspective of Connection Theory · 11 October 2026
1. What Is a Person?
We ordinarily understand the world through nouns: a person, a tree, a company, an intelligent system. These expressions are necessary and useful, but they can tempt us to mistake the properties visible at one moment for the whole of the object.
A living person continually exchanges matter and energy with the environment. The body's constituents change; knowledge, memory, habits, and social relationships also evolve. Decades later, this person differs enormously from the infant they once were, yet we may still regard both as stages of one person. Such a judgment does not rest on comparing two snapshots of molecular composition. It concerns a real bodily, causal, and organizational history.
This does not mean that nouns are grammatically wrong, or that a mathematically complete state description is useless. The mistake we wish to avoid is different: treating the static snapshot available to an observer as a sufficient account of a system that persists through time and of its capabilities.
This essay uses Dynamical Realism to clarify why the study of life, organizations, and intelligence within Connection Theory must consider entities, causal relations, and organization over time together. Process philosophy, dynamical systems science, and organizational research offer substantial precedents. Our purpose is not to claim philosophical novelty, but to make explicit the commitments on which subsequent theoretical work rests.
2. Neither Physical Entities nor Causal Organization Can Be Omitted
Relations do not operate without real bearers. A person needs a body; a company needs people and facilities; artificial intelligence depends on physical computation and information storage. After an organizational member dies, that person can no longer act as an active participant. Their knowledge, institutional position, and influence may persist through others, records, and institutions, but those effects still require real carriers and mechanisms of implementation.
Yet an inventory of parts is not an explanation of a system. A hundred people with comparable qualifications may remain isolated or form a stable, cooperating team. Identical computers and base models may have different effective capabilities when combined with different memory, routing, validation, and recovery arrangements.
We therefore claim neither that physical entities are always more fundamental nor that relations are always more important. Given a specified system, environment, time horizon, and intervention conditions, we ask which physical conditions and organizational mechanisms actually change the outcome. When resources are the bottleneck, supplying resources may matter most; when resources are comparable but coordination fails, changing organization may matter more. Their effects are often nonadditive and cannot be partitioned into fixed percentages without justification.
Here, relation in its general sense refers to causally effective interactions and arrangements in actual systems, not every mathematical or logical relation. Connection is a narrower, technical term in Connection Theory (CT).
3. Connection: From Actual Consequences to Stable Relations
Connection Theory distinguishes Interaction → Feedback → Connection → Organization. Interaction can change a state. Feedback is the history-dependent, selective influence of past interactions on subsequent organization. Connection denotes a relation shaped and consolidated through feedback. Organization concerns how such connections and other stable relations are coordinated through timing, routing, roles, synchronization, and other means to produce system-level activity. This is a conceptual progression, not a prohibition on recursive dynamics: an existing organization also affects which consequences can return and which updates become consolidated. See the formal concept pages for Connection, Feedback, and Organization.
Consider a company manager who observes delays and errors in the actual collaboration of two employees, changes how work and communication are divided, and tests the arrangement over several work cycles. A particular working relation eventually becomes stable. The manager participates in feedback processing and organizational updating. There is no principled distinction between human and algorithmic participants for this purpose. By contrast, a manager's reassignment made solely on personal preference does not itself demonstrate that actual consequences selected the new relation.
The relevant consequences must be traceable to the operation of the relation under examination or a previously declared and actually tested type of the same relation. General experience with vaguely similar systems, or predictions not tested in operation, cannot substitute for that evidence. Consequence sensitivity follows the criterion already stated in CT's Adaptation: a subsequent change must be differentially conditioned by the consequences of prior interaction, rather than merely being the next physical state caused by that interaction. Wear, random drift, and regulation under a fixed response rule do not, by themselves, establish this form of selection. Consequences must affect the selection, modification, or consolidation of a relation through records, return signals, or other real causal pathways, leaving a result stable over the relevant disturbances and time interval. Assessment must state the system boundary, relation level, object or type under test, timescale, and evidence.
This gives engineering manufacture and model copying a symmetric interpretive rule: a relational configuration selected and consolidated through actual tests of its consequences may carry the result of that feedback shaping into a new instance through faithful copying or manufacture to validated specifications. Replicated weights of a trained network and engineered couplings manufactured after repeated prototype testing are not different in principle on this point. A new instance need not repeat the original sequence of trials. But a manufactured instance that does not faithfully implement the validated relation, or an untested design that merely resembles one, does not automatically qualify. The copied structure may instantiate the same selected type without making two instances numerically identical individuals.
This choice has a real consequence: some engineered products operating under fixed rules may still contain CT Connections without thereby being adaptive in their present operation. Connection helps explain why particular stable relations have the form they do as a result of consequence-sensitive selection. What the system can do now must additionally be explained through Organization, Active Organized Dynamics, and Effective System Capability. Two systems with the same current physical arrangement and operating rules may have identical present performance even if their histories differ; a historical classification of Connection must not be mistaken for an extra physical power. A related distinction between selected history and current causal performance appears in the selected-effects literature in philosophy of biology (Millikan, 1984; Neander, 1991), though we do not import its theory of identity wholesale into CT.
4. Fixed Regulation and Adaptive Change Are a Different Distinction
A conventional thermostat can regulate temperature through a pre-specified feedback rule. Even if its design type has improved through testing, the fact that this particular device switches on and off does not itself show that it is adapting. Conversely, a manager who changes organizational arrangements in response to real team outcomes, or a training process that changes model weights in response to error, may exhibit consequence-sensitive organizational change.
To distinguish Regulation from Adaptation, we must first specify which level of state belongs to the organization or response mechanism under examination. Whether changes in a PID controller's integral state constitute adaptation at the relevant level depends on the predeclared variables and criteria for subsequent interaction. The mere observation that it accumulates error is insufficient. CT's definition of Adaptation allows consequence-sensitive change in state, policy, position, resource use, routing, or causal organization; it does not require every adaptive change to alter physical wiring. Nor should an instance-level test for Adaptation be imposed as a requirement that every Connection receive constant online maintenance.
This level distinction lets us study historically selected stable structures, useful control under fixed rules, and continually changing adaptive organization without conflating them.
5. Physical Technology Changes What Organization Can Do
Human collaboration once relied heavily on face-to-face speech. Writing and printing expanded the preservation and replication of information; telephony transformed long-distance exchange; computer networks enabled geographically distributed information sharing, automated coordination, and extensive asynchronous collaboration. Material technologies change not only speed and cost, but also which organizational arrangements are realizable.
Yet identical facilities can be used very differently. Physical conditions constrain or expand feasible forms of coordination; organization determines how resources are put to work; organizational needs may in turn drive technological innovation. This is an established insight in sociotechnical-systems research. We use it here as an accessible example of the co-evolution of entities and organization.
6. Snapshots, Structural Memory, and Persistent Identity
To understand a person, we cannot rely solely on today's medical measurements; to understand a company, solely on this quarter's revenue; or to understand an intelligent system, solely on one test score. Still, partial observation is not a complete state. If a full state, governing dynamics, and environmental conditions are sufficient, the effects of relevant history must already be represented in that state or its transition mechanism, rather than added again as a mysterious cause.
CT's Structural Memory refers to differences left by earlier causal interactions that remain available to affect later operation. The past does not act on the present without mediation; a causally effective trace must remain.
For a system described as “the same” across time, CT also separates Organizational Persistence, the maintenance of organization during continuing operation; Organizational Continuity, a qualifying causal lineage across interruption under a predeclared boundary, lineage level, and interval; and Identity, a classification constrained by real continuity facts. Functional similarity does not establish the same lineage. Copying feedback-shaped weights or an engineered configuration does not turn separate instances into one individual.
Currently manifested capability, retained capability-relevant state, Recoverability, and applicability in the present environment are also distinct. A recoverability claim must specify in advance the tracked system and individuation frame, the target regime or capability, environmental conditions, admissible interventions, and the success criterion. These distinctions let us ask precisely what a system retains or loses across shutdown, component turnover, and environmental change.
7. From Present Performance to Future Capability
Two companies may produce the same output today yet face different future action possibilities because their resources, organization, retained states, and transition mechanisms differ. Given identical sufficient states, dynamics, action sets, and constraints, different future reachable sets cannot arise merely because we assert that the systems had different pasts.
Control theory already supplies reachability, viability, and recovery analysis. For social systems whose dynamics cannot be specified exactly, we can instead estimate bounded future possibilities under explicit scenarios, state the uncertainties, and provide bounds where feasible. A qualitative claim about what “might happen” must not be passed off as a mathematically established reachable set. General future reachability remains a direction for subsequent CT research, not a term already formalized in v4.6.
For intelligent systems, any claim that organization adds capability must also test competing explanations involving the base model, compute, context, and external coordination. Effective System Capability is multidimensional. Present performance, longer-term adaptation, recovery, transfer, and reorganization must be evaluated separately under appropriate tasks and constraints.
8. Intellectual Lineage and Limits of the Perspective
Process philosophy and process biology direct attention to the temporal extension of objects. Structural realism emphasizes that a component inventory may omit explanatory causal structure. Collier and Hooker's Dynamical Realism, Maturana and Varela's work on autopoiesis, sociotechnical-systems research, and interventionist approaches to causation all provide useful background. We adopt what helps explain real systems, without asserting that relations have a universal ontological priority over entities.
Connection Theory v4.6 is the normative source for the technical vocabulary used here. The symmetric interpretation of copying, manufacture, and feedback evidence is an applied clarification under its feedback-derived definition, not a new theorem about identity across instances. Where the evidence or normative meaning is insufficient, that uncertainty must be stated. An explanatory essay must not become a second, conflicting source of definitions.
9. Six Foundational Commitments for Subsequent Connection Theory Research
1. Reality and physical realization. Actual causal effects require physical realization; models must be distinguished from the reality they represent. Institutional identities may be constituted through collective recognition, but their real effects still require carriers and mechanisms of enactment.
2. Conceptual boundaries. Interaction, Feedback, Connection, Organization, and Adaptation must not be treated as interchangeable. Connection follows the v4.6 definition of a stabilized, feedback-derived relation. Whether consequence-sensitive updating occurs presently is assessed separately under Adaptation.
3. Feedback evidence and symmetry across instances. Human and algorithmic agents can both participate in feedback shaping. The relevant consequences must originate in actual operation of the relation under examination or a declared, tested type of that relation. Subsequent updates must be differentially conditioned on those consequences; wear, random drift, or prescribed response alone is not sufficient. Validated relational configurations may carry the results of feedback selection through faithful copying or manufacture, under the same standard. Analogous experience, prediction, or an untested design alone is insufficient. Every assessment must declare the boundary, level, type or instance, time window, and evidence of selection and consolidation; uncertainty must be acknowledged when evidence is missing. Carrying feedback-shaped structure does not establish individual identity continuity in the sense of Organizational Continuity.
4. Organizational explanations face competing tests. Having specified organizational variables, claims about their effects require appropriate comparisons or interventions and checks against simpler explanations based on compute, component capability, or external coordination. Physical and organizational effects need not admit a unique numerical allocation.
5. Temporal extension and layered identity. Partial observation is not a sufficient state; Structural Memory requires currently usable causal traces. Organizational Persistence, Organizational Continuity, Identity, Recoverability, and Current Applicability have distinct criteria.
6. Capability is conditional. Effective System Capability calls for multidimensional evaluation under specified conditions. Work on future reachability must state its model, admissible actions, environmental and resource constraints, timescale, and estimation method; no reachable-set claim follows directly from a snapshot.
These commitments are not a replacement glossary. When a term has a formal definition, the frozen Connection Theory v4.6 takes precedence.
Conclusion: Returning to the Person Who Keeps Changing
Return to the person with whom we began. We can photograph them, measure their blood pressure, or analyze the composition of their body. These records are real and useful, but each describes a local state in a life history. Through ongoing exchange of matter and energy, organizational maintenance, environmental interaction, and the retention of experience, that person continues through time. Their abilities are formed, altered, and sometimes lost through these processes.
Recognizing this continuity does not deny the physical reality of the body. Recognizing relations and organization does not give them a mysterious power independent of matter. Entities make real causal processes possible; causal organization coordinates the use of resources; ongoing dynamics allow effects of the past to enter the present and change what may happen next.
Dynamical Realism asks us to see all three together: real entities, real organizational relations, and their continuing evolution through time. This is a starting point for understanding the world, and the common foundation for Connection Theory's further study of life, organizations, and intelligence.
Principal References and Normative Sources
- Wang, K. (2026). Connection Theory: A General Framework for Adaptive Network Dynamics, v4.6, §§3–6; formal version.
- Collier, J. D., & Hooker, C. A. (1999). Complexly Organised Dynamical Systems.
- Ladyman, J., & Ross, D. (2007). Every Thing Must Go: Metaphysics Naturalized.
- Nicholson, D. J., & Dupré, J. (eds.) (2018). Everything Flows.
- Maturana, H. R., & Varela, F. J. (1980). Autopoiesis and Cognition.
- Orlikowski, W. J. (1992). The Duality of Technology.
- Woodward, J. (2003). Making Things Happen.
- Millikan, R. G. (1984). Language, Thought, and Other Biological Categories.
- Neander, K. (1991). Functions as Selected Effects: The Conceptual Analyst's Defense.
- Normative concept pages: Connection · Feedback · Adaptation · Organizational Continuity · Effective System Capability.