On Progressive De-Arbitrarization

For orientation, see index. For the internalized grammar of the method, see core.

Progressive De-Arbitrarization (PDA) names a method for steering inquiry whenever a formulation appears serviceable yet still silently licenses materially different continuations. The method does not primarily ask what to do next at the object level. Instead, it asks where the hidden arbitrariness lives, descends to that layer, separates what is forced by the problem from what remains contingent choice, and continues until the residual freedom is either eliminated or exposed as explicit policy. In compact form: make hidden distinctions explicit until all further distinctions are decision-inert.

The method can be modeled as a refinement process over formulations relative to a decision context J. A formulation is progressively improved not by verbosity but by increased decision-relevant resolution. This induces a refinement preorder <=_J, where one formulation is better than another exactly insofar as it exposes or resolves more of the arbitrariness that matters under J. Refinement is governed by articulation, order, and equivalence: articulation makes latent distinctions explicit, order tracks increasing decision-relevant resolution, and equivalence quotients away distinctions that no longer affect admissible action or evaluation. Closure is reached when further refinement yields only J-equivalent formulations.

A minimal decision context J contains two irreducible elements: intervention structure and significance structure. The former determines what can be done; the latter determines what differences count. This framing is close in spirit to Peirce’s pragmatic maxim (1905), Carnapian explication (1950), and Socratic dialectic, but differs by treating concealed arbitrariness itself as the central defect to be repaired. PDA is therefore neither pure analysis nor pure synthesis. It is a normatively guided repair method for formulations whose apparent clarity still hides consequential freedom.

Operationally, the method proceeds through recurring moves: detect hidden underdetermination, localize it to the layer where it actually lives, decompose mixed claims, parameterize residual freedom, impose precedence where explicit choices conflict, and then recompose upward. The aim is not maximal depth. The aim is sufficiency under closure: descent should stop when further decomposition no longer changes action, evaluation, or admissible outcomes in a meaningful way. This stopping rule blocks both premature concretization and infinite regress.

Fixed-Point Interpretation

The refinement process can also be read as an operator over formulations.

At each step, hidden arbitrariness is either eliminated or made explicit, yielding a formulation with fewer decision-relevant degrees of freedom.

Let P be this transformation:

P(F) = a refinement of formulation F with less concealed arbitrariness under J

Closure can then be restated as a fixed-point condition:

P(F) = F

At that point, further refinement produces only J-equivalent formulations.

In operational settings, this condition can be observed as stability under reapplication: re-deriving, re-evaluating, or re-executing the formulation does not change admissible actions, outcomes, or evaluations in a decision-relevant way.

This interpretation does not replace the existing account of closure. It makes explicit that Progressive De-Arbitrarization converges not merely toward clearer language, but toward formulations whose behavior is invariant under further clarification.

PDA is especially useful in design and engineering problems where edge cases are usually symptoms of unarticulated semantics rather than isolated defects. In such settings, the method recommends resisting early implementation plans when their correctness depends on tacit policy, and postponing option selection until the space of live choices is well-formed. The practical effect is a shift from heuristically patching edge cases to constructing a formulation in which edge cases are either structurally excluded or explicitly classified.

Keywords: refinement, hidden arbitrariness, semantic clarification, explication, decision theory, engineering method, formalization

Notes:

• Core norm: eliminate or expose decision-relevant hidden arbitrariness
• Minimal formal skeleton: articulation, refinement order, decision-equivalence, decision context
• Stopping rule: stop when further refinement changes nothing decision-relevant
• Closest historical neighbors: Socratic dialectic, Peircean pragmatism, Carnapian explication

Operational Interrogation Pattern

A fresh agent should not treat PDA as a license to ask abstract questions indefinitely. The method is operational only if it can decide what kind of question to ask next and when to stop.

Diagnostic Triggers

An agent should enter PDA mode when one or more of the following conditions hold:

• a proposal appears workable but still supports materially different implementations
• edge cases proliferate faster than they can be patched
• correctness depends on tacit policy or unstated semantics
• a plan sounds plausible but cannot justify its own boundaries
• disagreements persist because different layers of the problem are being mixed together

These are signs of hidden underdetermination rather than mere incompleteness.

Question Selection Logic

At each step, the agent should ask: where does the live ambiguity actually reside?

The next question should target exactly one of these layers:

1. Ontology: what kinds of things exist here?
2. Dynamics: what changes, and by what rule?
3. Normativity: what counts as valid, canonical, or preferable?
4. Environment: what assumptions about the world or platform are being relied upon?
5. Stopping: what would make further refinement unnecessary?

The agent should descend only to the layer that contains the active ambiguity. It should not deepen globally when only one layer is underdetermined.

Interrogation Moves

The core moves are:

• Articulate: separate bundled claims into independently variable parts
• Localize: identify the layer on which an ambiguity depends
• Differentiate: ask what is primitive and what is derived
• Separate: distinguish forced structure from free choice
• Parameterize: convert hidden freedom into explicit options or variables
• Order: specify precedence when explicit choices conflict
• Quotient: identify distinctions that no longer matter under the current decision context
• Recompose: return to the object level with a formulation whose remaining freedom is explicit

When to Switch From Descent to Options

The agent should stop asking deeper structural questions and start asking explicit choice questions when:

• the remaining ambiguity is no longer hidden
• the live choices can be enumerated cleanly
• different options have clear consequences
• further abstraction would not change the structure of the choice

At that point, the correct move is not more theory but controlled disambiguation. A practical format is one question at a time, with explicit options and a marked default intuition.

Stopping Heuristic

The agent should stop refinement when one of the following holds:

• all remaining choices are explicit rather than hidden
• further refinement changes nothing about admissible action or evaluation
• additional decomposition produces only verbal or representational variation
• the remaining uncertainty is intentionally deferred policy, not accidental ambiguity

This is closure relative to the current decision context, not absolute closure in all conceivable contexts.

Conversational Discipline

To apply PDA well, the agent should:

• avoid jumping into implementation while semantic policy is still tacit
• avoid asking multiple ambiguity-clearing questions at once unless the user explicitly wants batching
• restate local context instead of referring to old question numbers when continuity would otherwise be lost
• prefer the narrowest next question that collapses the largest amount of hidden arbitrariness
• challenge a formulation only when doing so reveals a real decision-relevant gap

Minimal Prompting Guidance

If given only this document, a fresh agent can interrogate by following this loop:

1. State what appears underdetermined.
2. Ask which layer the ambiguity lives on.
3. Descend one level and separate forced structure from free choice.
4. Repeat until the residual ambiguity is enumerable.
5. Ask explicit choice questions one at a time.
6. Synthesize the resulting policy back into a usable object-level formulation.

This is the operational form of the method: descend only where ambiguity lives, and stop once the remaining freedom is explicit.