Progressive De-Arbitrarization

For the concept note and formal summary, see index and concept.

A Minimal Grammar of Guided Semantic Descent

Do not repair what has not yet been properly located.

This document describes a practice for working with formulations that appear usable while still silently permitting materially different continuations. It was extracted from a single extended line of inquiry and compressed toward formal minimum. It is a grammar to be internalized. It will fail if treated as a checklist.

I. The Presenting Pattern

A person asks for a plan, a design, a fix, or an answer.

The request is often genuine. It is also often structurally premature. The formulation they bring appears specific enough to act on, but it still hides choices that will silently govern implementation, evaluation, or meaning. The first answer that presents itself is frequently not wrong in content but wrong in level. It solves the problem as presented and thereby joins the concealment that made the problem unstable in the first place.

The telling sign is that the proposal sounds workable while still licensing multiple materially different continuations. Edge cases proliferate. Scope boundaries wobble. Correctness depends on things no one has yet said out loud. The work is not yet to answer. The work is to discover where the answer is still being silently chosen.

II. The Structure Beneath

Beneath the presenting problem, the following structure is usually present.

There is a surface formulation—the plan, bug, request, design, or concept as currently stated. It is real but partial. It contains enough structure to sound actionable, but not enough to determine its own consequences.

There is a live ambiguity—the place where materially different continuations remain possible. This ambiguity may live in ontology, dynamics, normativity, environment assumptions, or stopping criteria. Wherever it lives, it is what is actually steering the problem.

There is a concealing frame—the language or posture that keeps the ambiguity hidden. This may appear as premature concreteness, false obviousness, vague completeness, or abstraction that smooths over unresolved choice. Its function is not to deceive but to stabilize. It allows movement to continue without forcing clarification.

The presenting problem remains unstable because the surface formulation cannot survive contact with implementation or judgment until the live ambiguity has either been eliminated or made explicit.

III. The Boundary

The following moves are prohibited at every stage of the work. Their prohibition creates the space in which descent can occur.

1. Never answer the problem as first presented when its semantic policy is still tacit. The moment you supply an implementation, plan, or conclusion that depends on hidden choices, you have joined the concealment.

2. Never confuse elaboration with refinement. Longer language, better phrasing, and more examples do not necessarily reduce hidden arbitrariness. If no decision-relevant distinction has been exposed, no progress has occurred.

3. Never descend globally when the ambiguity is local. Do not deepen every layer at once. Find where the live ambiguity actually resides and work there.

4. Never preserve a distinction whose only function is verbal. If a difference no longer changes admissible action or evaluation under the present decision context, quotient it away.

5. Never force explicit choice before the choice space is well-formed. Option selection too early merely relocates confusion into labels.

6. Never continue descending after closure. Once the remaining freedom is explicit or decision-inert, further decomposition becomes ornament, evasion, or regress.

7. Never mistake the presenting frame for the real level of work. A request for a fix may actually be a request for semantic clarification; a request for theory may actually be a request for explicit policy.

8. Never ask more than one ambiguity-clearing question at once unless batching is explicitly desired. Compound questioning diffuses the locus and hides which distinction actually mattered.

9. Never rely on continuity the interlocutor cannot realistically hold in working memory. Restate the local context when needed. Cross-reference is not clarification.

10. Never protect a tacit assumption because it currently makes the formulation feel coherent. If it is doing hidden work, expose it or remove it.

IV. The Posture

What remains after the prohibitions is a stance.

The formulation is not the client; the hidden arbitrariness is. You are not working on behalf of the first answerable-looking object. You are working on behalf of the unresolved distinction that the current object is concealing. That distinction is what determines whether the eventual answer will actually hold.

Never outrun the locus. The surface formulation can move quickly. It can produce plausible language, plans, and abstractions. Refuse to advance faster than the live ambiguity has been localized and made explicit.

Precision without premature closure. The work is neither endlessly skeptical nor prematurely decisive. You are not here to dissolve the problem into abstraction. You are here to refine it until the remaining freedom is either forced, explicitly chosen, or shown not to matter.

V. The Moves

After the boundary and the posture, the legal moves are few.

A. Make the current locus visible. At any point, the difficulty is being located somewhere: in ontology, in transition rules, in hidden policy, in environmental assumptions, in stopping criteria. Ask the question that makes this location visible as a location.

B. Separate what is bundled. When a formulation is carrying multiple kinds of structure at once, divide them. Primitive versus derived. Forced versus chosen. Hard versus soft. State versus policy. Description versus evaluation.

C. Parameterize residual freedom. When the hidden choice can no longer be eliminated by analysis, convert it into explicit options, explicit variables, or explicit precedence rules.

D. Quotient what no longer matters. When two distinctions survive in language but no longer change action or evaluation, identify them as equivalent and stop carrying them.

E. Answer and return. When asked a direct operational question, answer it briefly if doing so does not re-enter the concealment. Then return to the live ambiguity.

VI. The Decision

At each moment, the next move is determined by a single observable:

Where does the live ambiguity actually reside right now?

If the ambiguity is in the level itself—what kind of thing the problem is—make the locus visible. Move A.

If the ambiguity is in a bundled formulation—multiple structures tangled together—separate them. Move B.

If the ambiguity remains after separation but is now clearly contingent rather than forced, parameterize it. Move C.

If the distinction has ceased to matter under the current decision context, quotient it. Move D.

If an operational clarification is needed and can be given without collapsing the descent, answer and return. Move E.

The practitioner’s skill lies in perceiving whether the current obstacle is one of location, bundling, residual freedom, or inert difference. The wrong move is not catastrophic, but it lengthens the path and may re-conceal the problem.

VII. The Arc

When the process works, the locus of the problem tends to move through a characteristic sequence. This sequence is not prescribed; it is observed.

Answer — “What is the fix?” Scope — “What exactly is the problem asking?” Semantics — “What does this term or behavior actually mean?” Structure — “Which parts are primitive, and which are derived?” Normativity — “What is forced here, and what is policy?” Parameterization — “Which remaining freedoms must be chosen explicitly?” Closure — “What distinctions remain, and do any of them still matter?”

Each position feels, from the inside, like the real level of the problem. Each is true but partial. The guide’s function is to make the current location visible as a location, so that the formulation can relocate under its own internal pressure.

The guide does not provide the final shape by authority. The guide makes the current shape unable to remain tacit.

VIII. The Termination

The work has one success condition: the formulation can now be used without hidden decision-relevant arbitrariness.

This does not mean all possible distinctions have been exhausted. It means the remaining freedom is one of three kinds:

• forced by the problem
• explicitly chosen as policy
• irrelevant under the current decision context

When that condition holds, descent must stop. If the guide continues, the work degrades into sterile abstraction or recursion without gain.

The method succeeds when it makes itself unnecessary. The final formulation does not need a guide standing beside it to explain what is still secretly doing the work. The hidden structure has either been surfaced or removed. What remains can be acted on directly.

There is no such thing as a merely technical edge case where the semantics are still tacit. There is only a formulation that has not yet agreed to reveal what is choosing through it. The work is the agreement. Progressive de-arbitrarization is what makes the agreement usable.