Invokable Structures

What It Is

Invokable structures are external structures (companies, agent bodies, institutions, programs, saved workflows, skills) that exist ahead of any particular goal and can be called like functions when a goal needs them. A person with invokable structures around them turns a new goal into a configuration those structures search for. A person without them turns the same goal into an error signal that nothing can respond to, and the goal dies because nothing exists that can answer it.

"I want to find invokable structures that create a larger decision space. This could be companies, it could be agent-led companies."

The claim underneath is a claim about agency itself. Agency is usually modeled as a property of the actor: drive, capability, willingness to act. The invokable-structures view relocates most of it: agency is largely the ability to create or borrow callable external structures, because those structures are what convert intentions into legible actions. The actor supplies the goal; the structures supply the action vocabulary.

The Goal-Validity Test

A goal is a component of a control system, and cybernetics lists what a control system needs: a goal state, sensors, actuators, a control loop, and resource constraints. Setting a goal builds exactly one of the five. The goal generates an error signal, the measured gap between where you are and where you declared you should be, and an error signal is only useful if a mechanism exists that can respond to it.

"A goal is like you generate an error signal. But if I don't have a mechanism to respond to that error signal, or the error signal is too noisy — nothing to do… If I set a goal like I want to make fifty k this month, I don't have the actions. So I need to build that action vocabulary."

This produces a test that most goal-setting advice skips entirely. "Save $50k" is a valid goal for someone with an income: the error signal maps onto actions that exist today (spend less, move money, cancel the subscription). "Make$50k this month" is an invalid goal for someone with no product, no leads, and no delivery pipeline: not an ambitious goal, an invalid one, in the same way a thermostat wired to no furnace is an invalid heating system. The error signal fires and dissipates as anxiety, because anxiety is roughly what an unanswerable error signal feels like from the inside.

The moralized reading of a stalled goal is a character verdict: not hungry enough, not disciplined enough. The mechanistic reading is a completeness check on the table above, and the row that fails is almost always the actuator row. The repair is making structures exist that can be invoked, which is why the first real action on an invalid goal is usually construction or borrowing rather than effort.

Goals Compile to Configurations

Once enough structures exist, the relationship between goals and actions inverts. Instead of each goal demanding a hand-derived plan, the goal becomes a query against the structures already standing:

"The key thing is they are expressive enough that I could set a new goal and it becomes a configuration on these invokable structures."

The end state of systems-building is exactly this expressiveness. An income goal like "$100k by end of month" stops being a wish and becomes a configuration problem: which ventures run, at what intensity, selling what, to whose leads. The block on a goal like that was never effort. It was the absence of programmable structures for income, which makes it a search problem over structures rather than a willpower problem over days. This is macrostate engineering with the missing half filled in: a macrostate declares what must be true and delegates the microstates to a resolving computation, and invokable structures are what the resolving computation is when the domain is your life rather than a chat window. Declaring a macrostate with no structures underneath it is declaring an acceptance criterion over an empty search space.

The same move generalizes causality programming upward. At the habit scale, the rule is to make desired causality the default path: wire the graph so the gym happens without a daily decision. At the institutional scale, the wiring is made of standing structures: a company is a causal path from "someone needs X" to "money arrives" that runs without you re-deriving it, and building one is rearranging causality into the environment so the world itself helps produce the consequence:

"It is modifying material reality so responsibility and computation move into the environment—rearranging causality so the world itself helps produce the desired consequence."

Borrowing Before Building

The structures do not have to be yours. Every environment already contains invokable structures that most people never enumerate: firms, institutions, cohorts, accelerators, fellowships, government pipelines. Using them is not a lesser form of intelligence:

"If intelligence means getting what you want, it means being able to use the structures present in your environment… It's the definition of stigmergy."

Stigmergy is the biologist's word for coordination through structures left in the environment, and the working image is the spider, which expresses its intelligence as a web that does work for it continuously. The practical corollary is an audit: sit down and enumerate the invokable structures your environment actually offers, the way you would enumerate the functions of a library you just imported. A city, an industry, a stage of life each offer a different structure menu, and someone who has mapped the menu has a larger decision space than someone of equal talent who hasn't.

Borrowed structures also dissolve virtue words. Universities make hard multi-year learning happen not because students possess patience but because the institution surrounds the waiting with social pressure, prestige, and access that make patience rewarding:

"It takes a lot of patience, and patience is not just a virtue. It is a system — when you have other things that make it rewarding."

When a plan requires a virtue you do not reliably have, the substitution is structural: find or build the system that makes the virtue's behavior rewarding, rather than emulating the virtue bare. This is the same translation moralizing vs mechanistic runs on discipline and willpower, applied to the structures outside the skin.

Spinnerets

The view also rewrites what "low agency" means, because the ability to create agency is itself improvable:

"Agency can be created like a spider defines its area of control with a web. What are your spinnerets—the actions that expand your area of influence?"

A spinneret is any action whose output is a standing structure rather than a consumed result: incorporating a venture, publishing a system others can invoke, building an agent body that keeps running, joining an institution that grants standing capabilities. Ordinary actions spend effort and end. Spinneret actions spend the same effort and leave behind something callable, so the decision space is permanently larger after them. The ladder of agency describes climbing through demonstrated mastery; from level 4 upward, what the climbing physically consists of is this — forcing functions, coordinated teams, institutions are all invokable structures of increasing scale, and each rung is a new class of structure the climber can now create or command.

Marshalling, Not the Clever Push

There is a common mental model of working smart as a single high-leverage act: find the clever solution, execute it once, collect the multiplier. The structural view corrects it:

"I used to think working smart was a one-time thing — find a clever high-leverage solution. ... But I realized it is not just a one-time push but also leveraging pre-existing structures, or marshalling, or compounding them."

Marshalling is ongoing: keep the inventory of available structures current, route each new goal through them, and compound them so the next goal finds a richer vocabulary than the last one did. It is also largely divorced from felt cognitive effort: when the structures help you search and prune the space, the work feels light precisely because the structures are carrying the computation. Effort spent building a structure is external and persists, while the same effort spent pushing directly at the goal is internal and evaporates with the day. This is the same variance-containment logic as the agent body's cell framing: you build structures around yourself because your own willpower is random variance, and the structure is what carries the compounding.

Advice Becomes Tooling

The most compressed version of the whole idea shows up when a mental heuristic gets externalized. Working backwards, SMART goals, the Eisenhower matrix — these are policies over a representation of goal-space, and as long as goal-space lives only in a head, a policy can only ever be advice, re-run by hand each time someone remembers it. Externalize the representation and the policies change type:

"The meta-heuristics we teach as human advice… were always operations on goal-space that you had to run by hand… Advice becomes tooling."

Once pursuits exist as an explicit external structure (an indexed graph of experiments an agent can operate on), every heuristic that used to be advice becomes an executable skill invoked against that structure. This is the invokable-structures move performed on cognition itself, and it marks the general pattern: whatever currently exists only as knowledge-you-must-remember-to-apply is a structure waiting to be built, after which it stops costing memory and starts being callable.

Boundary: Structure over Request

Structure over request is a sibling law on a different substrate, and the two are easy to blur because both say "build structure instead of wishing." The distinction is what the structure determines. Structure over request is about output quality: depth, rigor, and voice are properties of the computation topology, so you build the invocation graph rather than asking the model to think harder. Invokable structures are about action capability: whether a goal can produce motion at all is a property of the callable structures standing around the actor, so you build the company, the body, the pipeline rather than asking yourself to try harder. One governs how well a computation runs; the other governs whether an actor has actions to take. They compose, since an invoked structure will often contain computation topologies inside it, but they answer different failures.

The action-vocabulary framing also connects to legibility: a goal without invokable structures produces no legible actions, which is why an invalid goal feels like fog rather than like a hard task. When the fog is about your own plan's undefined variables, clarity's diagnostic applies; when the plan is clear and there is still nothing to call, the missing piece is a structure, and no amount of further planning will conjure it.

Related Concepts

• Cybernetics - The five-component completeness test; a goal without actuators is an open loop
• Agency - The capacity that invokable structures externalize and enlarge
• Ladder of Agency - Climbing from level 4 up is the creation of successively larger invokable structures
• Macrostate Engineering - Macrostates need a resolving computation; standing structures are that computation at life scale
• Causality Programming - Making desired causality the default path, lifted from habits to institutions
• Structure over Request - The sibling law for output quality rather than action capability
• Agent Body - The structure-around-the-self that contains willpower variance
• Moralizing vs Mechanistic - Virtue words decompose into structures here as traits decompose into systems there
• Legibility - Structures are what make a goal's actions legible enough to execute
• Clarity - The first-person diagnostic for when the plan, not the structure, is what's missing
• Error Signal - The object a goal emits; this page builds what answers it, that page builds the signal itself

Key Principle

A goal is only valid once structures exist that can respond to its error signal, and agency is largely the ability to create or borrow those structures. Run the completeness test on any stalled goal: goal, sensor, actuator, loop, constraints — the actuator row is usually the empty one, and the repair is construction or borrowing, not effort. Map the structures your environment already offers before building new ones (stigmergy), prefer actions that leave callable structures behind (spinnerets), and keep marshalling them so each new goal compiles to a configuration instead of demanding a hand-derived plan. Virtues decompose the same way: patience is a system, not a trait. The end state is expressiveness — a set of structures rich enough that setting a goal is enough to start the search.

Set the goal, then check what answers it. If nothing can be invoked, the next action is never trying harder — it is making something exist that can be called.