Improving an agentic system crosses the prose-symbolic boundary

Type: kb/types/note.md · Tags: foundations, constraining, self-improving-systems

In an agentic system — one whose operation runs through a model interpreting retained artifacts — the changes that improve reliability characteristically move behavior across the boundary between prose and symbolic form rather than staying within either. A self-representation covering only one form can therefore represent the side such a change starts from but not the side it must land on, and the change escapes governance at exactly the point where the improvement happens.

The asymmetry sorts behavior between the forms

The engine is an error profile, since scheduler-LLM separation exploits an error-correction asymmetry: work that can be fully specified — counting, membership, state tracking, scoping — is exact on a symbolic substrate and error-prone inside a model, while judgment that cannot be fully specified has nowhere to live but prose. A system under reliability pressure is therefore continuously sorting. New behavior enters as prose, because prose is where a half-understood constraint can first be stated; as operation clarifies which part of it is bookkeeping, that part is pushed into code. The movement is codification, the pressure behind it is the constraining gradient from convention to enforcement, and the sorting is never finished, because codifying a rule requires an oracle for it and the boundary of automation is the boundary of verification.

The improving changes are crossings

The sorting runs in both directions, and each direction is a boundary crossing:

Prose to symbolic. A convention that decays into violations becomes a validator; a membership claim becomes an enforced mark. The improvement is precisely the acquisition of a formal consumer — the unenforced prose version of a completeness claim is a trap, since stale indexes are worse than no indexes: it tells an exhaustive consumer to stop looking while members are missing. What made the system more reliable was not better prose but the crossing itself.

Commonplace's complete mark is this crossing, observed. The claim this head links every note carrying the tag existed as an intention before it existed as a check; ADR 026 turned it into an enforced frontmatter mark and is explicit that the unenforced prose version must never be written. The decision landed in four forms in one commit — prose spec, JSON schema, validator, renderer — and the validator now rejects a complete: true head with a missing member. Nothing about the prose got better. It acquired a formal consumer.

Symbolic to prose. An enforcement cannot be repaired from inside itself, because the standard it should have applied is not in the code — the code is the thing under suspicion. The correction has to land in prose. That happens two ways: the enforcement misfires, rejecting what it should accept or encoding a decision the system has outgrown, and the rule must be re-derived from revised intent; or the enforcement runs correctly and its output shows that the prose was wrong. Enforcement whose rationale is not revisable drifts opaque; prose whose enforcement is not updatable decays into advice.

The same repository supplies the second case: adopting covered_by on the learning-theory head, the new symbolic check caught a tag member the documented rg recipe had missed — the trace retrieval failure is reflection failure reads for what the wire needs to hold. The fix was not to the code. Executing the symbolic artifact produced evidence that the prose was wrong, and the search recipe in kb/reference/navigation.md was corrected, not the check. Had the recipe not been a revisable artifact inside the system, the check would have been left reporting a discrepancy nobody could act on.

Changes confined to one form exist, and some of them matter: a code-only bug fix under an unchanged prose contract genuinely improves reliability. The claim is not that single-form changes are worthless but that they cannot move the line — what the system leaves to interpretation and what it enforces. Moving that line is what a crossing is, and it is where the reliability gains that survive a change of operator concentrate.

Consequence for reflective coverage

Since reflective coverage is graded across representational forms, the question is which grades suffice, and for agentic systems the answer follows from where the improving changes live. Prose-only coverage lets the system revise what it recommends but not what it enforces, so accepted recommendations accumulate as advice. Symbolic-only coverage lets it adjust what it enforces but not why, so enforcement loses its path back to intent. Either way the improving change lands outside the self-representation, so it reaches later behavior without passing through anything the system represents about itself — and nothing in the loop governs it. Governed self-extension in an agentic system therefore needs modification-depth coverage of both forms, and of the mappings between them.

Scope

  • The claim characterizes where reliability gains concentrate under the stated pressure; it is not a universal over every improvement. A system without reliability pressure, or with no formal consumers at all, is outside the argument.
  • It concerns the prose-symbolic pair. The distributed-parametric form sits under selection-grade levers in current agent systems, and this argument does not extend to it.
  • The mechanism carries the claim; the direct evidence base is the single repository trace used in both directions above, set out in full in Commonplace as a reflective system. One trace is not a sample. Classifying external agent-memory systems by which forms their improvements touch would test the claim beyond this repository.

Relevant Notes: