Agency Is Not a Feeling, It Is a Complexity Machine

The strange link between control and complexity

What if the thing we call agency is not just a private feeling of being in charge, but one of the main engines by which the world becomes more complex?

That question sounds almost backwards. We usually treat agency as something we experience from the inside, the sense that I chose this, I caused that, I meant this outcome. Complexity, by contrast, sounds like a property of systems, ecosystems, technologies, organisms, economies. Yet these two ideas may be the same phenomenon seen from different angles. To have agency is to persist in a world of uncertainty by building ways to act that are flexible, goal directed, and increasingly sophisticated. To grow in complexity is, in part, to accumulate more such ways of acting.

This matters because it changes the story we tell about progress. Complexity is not just stuff getting more intricate. It is the result of systems learning how to do more with more options, more environments, and more futures. Agency is the mechanism that makes that possible.

---

Agency is not a switch, it is a structure for surviving uncertainty

A shallow view of agency treats it like a binary. Either a thing has control or it does not. But real agency is more like a structure made of several interlocking capacities: intentionality, monitoring, equifinal variability, and rational means end selection.

That language sounds technical, but the underlying idea is intuitive. A truly agentic system does not merely react. It represents a future goal, notices what is happening now, compares the two, and chooses among different possible routes to get there. If a door is blocked, it uses the window. If the window is shut, it tries another tool. If the tool fails, it revises the plan. That ability to reach the same goal through different means is not a side effect of agency. It is the core of it.

Consider a chess player. A novice may have a few fixed moves in mind. A strong player does something very different. They imagine a future board state, scan the current position, and evaluate multiple routes to that future. They are not chained to one procedure. They are organized around a goal. The more sophisticated the player, the more their agency looks like a search process across possibilities rather than a single line of execution.

This is why agency cannot be reduced to simple cues. A system can look active without truly being agentic. A thermostat changes behavior, but it does not model a future in the way a person does. A dog can be goal directed in a richer sense. A human can not only act, but also reflect on acting, imagine alternatives, and revise the very goals being pursued. Agency is not a mere motion detector. It is an inferential system for navigating uncertainty.

Agency is not the absence of constraints. It is the ability to keep moving toward a goal despite constraints.

That distinction matters because it reveals the deep connection to complexity. The more a system can adapt its means while preserving its aims, the more complex its action space becomes. Agency is not opposed to complexity. It generates it.

---

Complexity grows when successful action leaves behind new possibilities

The claim that everything in the universe tends to become more complex can sound mystical until you ask what complexity actually is. One useful answer is this: complexity is stored functional information. It is the accumulated structure that makes some actions possible, efficient, or reliable.

A river channel is more complex than random water movement because it channels flow into stable paths. A city is more complex than a village because it supports specialized roles, transportation networks, supply chains, and layered forms of coordination. A biological organism is more complex than a cloud of molecules because its parts are organized around survival, repair, and reproduction. In each case, complexity is not just more parts. It is more functional organization.

This is where agency enters the picture. An agent that solves problems does not merely expend energy and disappear. It often leaves behind scaffolding. A beaver builds a dam, and the dam changes the environment in ways that alter future behavior. A human invents writing, and writing transforms memory, coordination, governance, and culture. A software team creates a system, and the system creates new workflows, new bottlenecks, and new kinds of expertise. Successful action becomes a platform for further action.

That is the feedback loop to notice: agency externalizes itself into structure. Once a goal is achieved, the method of achievement can be preserved, refined, generalized, or combined with other methods. This is how complexity accumulates. Every solved problem becomes part of the machinery for solving the next problem.

Think about language. At first glance it is just a communication tool. But language is also a compression engine for agency. It lets one mind influence another, coordinate distributed action, transmit plans across time, and create institutions that no single person could run alone. Once language exists, new forms of agency become possible, and those forms leave behind even more complexity: law, science, bureaucracy, education, markets, literature.

In that sense, complexity is not random elaboration. It is the sediment of repeated successful agency.

---

The deeper tension: agency creates complexity, but complexity also creates agency

Here is the paradox that makes this question interesting. Agency may be one of the drivers of increasing complexity, but complexity is also what makes higher forms of agency possible.

A bacterium has a narrow repertoire. A bird has more options. A person has vastly more, partly because the surrounding world is more structured. Roads, tools, symbols, institutions, and technologies expand the space of possible actions. You can only write a novel because a civilization has already created alphabets, keyboards, literacy, publishing, and an audience. Your agency is not isolated from complexity. It is amplified by it.

This creates a recursive loop. Simple agency creates modest complexity. Complexity then supports stronger, more flexible agency. Stronger agency then creates still more complexity. The process is cumulative, but not linearly so. It is a ratchet. Once a new level of functional organization exists, it becomes the basis for the next layer.

A useful way to think about this is through three stages:

1. Goal formation: A system represents a future state.
2. Adaptive execution: It finds multiple ways to move toward that state.
3. Scaffold accumulation: The successful pathways become durable structure, making future action easier and more varied.

Seen this way, agency is not a magical inner force. It is a process of future shaping the present, and then the present being built into something that can better serve future goals. Complexity is the residue of that process.

This also explains why the feeling of agency matters, but is not enough. People often feel in control when they are following scripts, and feel out of control when they are improvising in genuinely hard conditions. The feeling is a clue, not the whole picture. Real agency is verified by the robustness of the system’s goal directed behavior under changing conditions. Can it adapt? Can it monitor? Can it recover? Can it choose among multiple means?

The hallmark of agency is not certainty. It is graceful response to changing conditions.

That is why the most capable agents often look less like rigid controllers and more like living systems. They can absorb noise, revise plans, and remain oriented. Complexity is not a sign that agency has failed. Often it is evidence that agency has matured enough to handle a more complicated world.

---

A new mental model: agency as functional compression

The most powerful synthesis here may be this: agency is functional compression.

A complex environment contains too many possible states to manage one by one. Agency compresses that space by creating goals, models, habits, tools, and institutions that allow action without exhaustive calculation. A person does not need to recompute every movement from scratch to cross a street. A team does not need to invent collaboration from zero every morning. A civilization does not need to rediscover farming, accounting, or navigation each generation. Agency builds compressed structures that preserve action across time.

But compression has a cost and a benefit. The cost is rigidity if the compressed structure becomes too fixed. The benefit is leverage if it remains adaptable. That is why the best forms of agency are not just efficient, they are equifinal. They can reach the same goal by different paths. They store not only a solution, but a range of possible solutions.

This is a profound criterion for understanding living and intelligent systems. The more a system can preserve purpose while varying method, the more agency it has. The more it can preserve effective structure while adapting to new conditions, the more complexity it generates.

A business illustrates this perfectly. A brittle business depends on a single product, a single leader, a single market condition. A resilient business develops processes, feedback loops, and modular teams. It becomes more complex, but that complexity is not clutter. It is compressed agency. It allows the organization to respond to shocks without losing itself.

The same is true for a person. A person with agency does not simply