wiki/concepts/context-carrying-capacity

Context Carrying Capacity

wiki/concepts/context-carrying-capacity/index.md

Rendered from markdown source. Open raw source on GitHub.

Context Carrying Capacity

Parent lineage: The Telic Field Papers

Status: candidate Source role: semantic wiki layer Processing tier: 4 Content canon status: unset

Working read

Context carrying capacity is the amount and complexity of relevant telic difference a loop can hold in active relation without suppressing standing, losing coherence, or collapsing into coercive simplification.

It is not simply memory.

A person may remember many facts while being unable to hold two conflicting needs at once.

An institution may possess every record while allowing only a quarterly metric into executive context.

An AI may retrieve a large archive while acting from a narrow selected context.

Core claim

No element, person, or system can hold the full telic field of everything it depends upon.

Active context

Context includes the differences that change what a signal, action, request, or consequence means.

Active context is the portion currently available to the process governing action.

What falls outside active context may remain real while losing practical influence.

Candidate formulation:

What falls outside active context can lose effective standing without losing constitutional standing.

Context capacity is relational

Context capacity depends upon:

  • the number of relevant centers;
  • the complexity of their fields;
  • contradiction;
  • ambiguity;
  • time pressure;
  • emotional load;
  • trust;
  • available witness;
  • language;
  • processing ability;
  • power;
  • reversibility;
  • stakes.

The same system may have high capacity in one relation and low capacity in another.

Internal example

The frightened part may know danger but not the present.

The ambitious part may know the opportunity but not the body's cost.

The caretaker may know another person's needs but not the self's.

The conscious narrator may receive signals without the histories that make them intelligible.

Internal conflict may therefore arise from unequal access to context rather than merely incompatible values.

Types of omission

Protective omission

Context is temporarily withheld because the receiver cannot safely or usefully metabolize it.

Capacity omission

Context is absent because the system cannot presently hold it.

Strategic compression

Only the information needed for a scoped decision is transmitted.

Distortive omission

Removed context changes the apparent meaning or purpose.

Constitutional erasure

A relevant element is systematically excluded from decisions that recruit its energy or impose consequences upon it.

Not every omission is breach.

Omission becomes constitutionally dangerous when the system repeatedly benefits from a center while excluding what that center protects.

Overload responses

When context exceeds capacity, a healthy loop may:

  • slow down;
  • sequence;
  • partition;
  • summarize visibly;
  • preserve unresolved items;
  • delegate;
  • consult an outer loop;
  • pause;
  • stop;
  • release the current objective.

If it does none of these, it may begin dropping standing silently.

Consentful context exchange

More context is not automatically better.

Some context was partitioned for protective reasons.

A consentful system should increase context exchange without exceeding the ability of participating centers to remain present.

Candidate formulation:

A healthy self can route the right context to the right relationship at the right resolution.

Context and consent

Consent requires enough relevant context for the scope of the decision.

It does not require complete disclosure.

A center should not be required to reveal more than the receiving relationship can safely hold.

A receiver should not claim informed consent while omitting context necessary to understand the active telos, cost, risk, or boundary.

Context and AI

An LLM context window is only one technical form of capacity.

The more important questions are:

  • which context is selected;
  • who selected it;
  • whose standing is represented;
  • what was omitted;
  • whether compression loss is visible;
  • when the system knows it has exceeded capacity;
  • what it does next.

A large context window does not guarantee constitutional adequacy.

Context overflow and governance

When a loop can no longer preserve protected stakes, it must halt, degrade, reroute, or recruit a competent outer loop.

Context overflow provides one explanation for why.

The loop is no longer capable of holding the differences required for legitimate action.

Risks of misuse

  • using capacity as an excuse for permanent exclusion;
  • forcing disclosure in the name of completeness;
  • assuming more data creates more understanding;
  • hiding omission inside summaries;
  • treating a technical token limit as the whole problem;
  • allowing high-status voices to consume all available context;
  • calling a collapsed single objective “clarity.”

Open questions

  • Can context carrying capacity be measured?
  • Is capacity multidimensional?
  • How should overload be signaled?
  • What context must never be compressed away?
  • How should a system preserve unresolved standing?
  • When should an outer loop intervene?
  • How do trust and witness increase capacity?
  • What is the role of forgetting and release?

Relationships

Upstream

Downstream

  • Semantic Polytelometry
  • Polytelometric comparison
  • AI context governance
  • Temporal telic relations
  • Stop / degrade / reroute invariants

Source lineage

This page remains a candidate semantic surface. The source artifacts govern provenance and exact wording.