Problem matches guide method search; mechanism matches bound transfer

Type: kb/types/note.md · Tags: foundations, methodology

A problem match consists of an independently established target problem and a source tradition with explicit responses organized around materially that problem. The problem must be stated at a discriminating level, so differences in optimization target, feedback, information structure, or failure mode remain visible. A problem match guides search by identifying bounded responses worth examining. It neither predicts greater yield than other sources nor warrants adopting any response.

The soft-bound-traditions survey and the human-writing transfer analysis both illustrate this search move: start from a shared problem, then test source responses against target conditions one by one. A problem match therefore returns candidate responses, not a domain template.

A mechanism match exists when a specific response's operative mechanism and the premises it depends on are independently present in the target part the response would govern. Because borrowed patterns transfer only as far as source and target share a mechanism, only that shared relation carries transferred warrant. Similar labels, outcomes, or domain descriptions cannot supply missing target premises, and any feature beyond the matched mechanism remains a target-side conjecture.

The stages are therefore distinct. Problem match selects candidates. Mechanism match bounds transfer. If several responses pass that test, combining them is still a new target-side construction: each contribution must retain its own conditions and non-transfer boundary, and their interactions must be checked.

Source and target can reach the same control problem through different causes. A causal-origin match shares the condition that generated the source problem. A downstream problem match shares only the later control problem. The deeper match can strengthen the reason to inspect a source, but it is not a substitute for testing the response mechanism. The shallower match still guides search when the candidate response does not depend on the absent source cause. Naming the causal depth prevents a common downstream shape from being mistaken for a broad domain analogy.

Worked case: later-informed delegated choice

The relevant source object is command in war. Its source problem is deeper than ordinary distributed information. As the bounded military reconstruction explains, an intelligent opponent has independent purposes and can observe, resist, and respond. One side's actions therefore help change the other side's choices and the state in which later action occurs. Friction, chance, and information limits add other divergence between plan and execution. The military response retains purpose and non-negotiable boundaries while moving a bounded choice of means toward the actor who can observe the resulting state.

The agent side of the relation must be established independently. Preferential codification concentrates less predictable work at the agent boundary only under a strict condition: within a fixed incoming workload and routing policy, the system codifies a nonzero share of recurring cases and every removed case is more operationally predictable than every retained case. The result concerns the composition of the residual work, not its total effort or intrinsic difficulty, and it does not claim that deployed systems usually meet the condition.

Even under that condition, less predictable does not mean better decided later. Execution-time choice is favored only for the subset where relevant state arrives during execution, or where execution can cheaply resolve the live branch while advance planning would elaborate many unused branches. The later state must be capable of changing which permitted means best serves the fixed intent.

Once that target problem is independently established, the depth of its match must be stated. Genuinely adversarial agent work may share the causal origin when another capable actor adapts against it. Non-adversarial residual work does not. It matches only the downstream problem when execution can reveal evidence that changes which permitted means best serve a fixed intent.

That downstream problem match makes intent-framed delegation a candidate source response: upstream fixes a purpose-bearing result and non-negotiable boundaries, while a competent executor receives defined authority to choose execution-dependent means. It becomes a mechanism match only when the target independently has that information and authority relation. Upstream must hold intent or constraints that execution cannot safely reconstruct. The executor must have decision-relevant execution evidence, and the handoff must govern integration, verification, and return of control. The match transfers the bounded delegation relation. It does not transfer adversarial purpose to a non-adversarial task, military organization or doctrine, or evidence of agent effectiveness.

Compose functions, not traditions

A target method can assign several matched mechanisms different jobs. Four of the contributions below concern deferral and retain the minimum established by productive deferral: a materially live option, later evidence whose possible results can change the named decision, and a timely convergence rule that maps a return condition to an authorized response. These conditions make postponement controlled; they do not show that waiting is worth its cost.

Contribution Bounded target function Conditions and non-transfer boundary
Mission tactics, narrowed to intent-framed delegation Allocate judgment over execution-dependent means. Upstream retains purpose and non-negotiable bounds; a competent executor has a decision-specific information advantage, defined authority, and a governed handoff. Reciprocal opposition transfers only to a target with an adaptive opponent; goal choice, authority expansion, and military machinery do not transfer.
Rolling-wave planning Decide when to add detail: elaborate the near horizon and keep later work coarser until information improves. Later information must inform an operative replanning return. Coarse wording alone does not preserve a live option, and the method does not decide who holds judgment.
Real-options reasoning Preserve one costly-to-recover future exercise choice while informative observation remains possible. The choice must remain exercisable through a later decision point, with irreversibility, expiry, lead time, and waiting costs kept visible. Finance valuation machinery and a conclusion that waiting is valuable do not transfer.
Dynamic adaptive policy pathways Map monitored signposts or triggers to prepared, authorized responses. The observation must discriminate the decision, and the response must remain feasible when the trigger fires. A trigger selects a prepared response; it is not unconstrained delegated judgment.
Set-based practice Keep several materially feasible candidates live and narrow them through bounded tests or observations against shared criteria. Evidence must be capable of changing selection, carrying cost must remain visible, and narrowing needs a boundary. A list of alternatives is not a preserved set.

These functions affect different parts of a decision: who chooses means, when detail is fixed, whether a future exercise choice survives, how an observation selects a response, and how several candidates narrow. Their separate warrant therefore cannot validate the package. Composition must check at least four interactions:

  • Name the choice each mechanism governs and who may select, reopen, modify, or abandon it. Keep alternatives and prepared responses inside the fixed intent and authority bounds.
  • Route later evidence to the actor who holds the relevant decision right. Align replanning, triggers, expiry, lead time, and narrowing, and verify that dependencies or partial implementation have not already destroyed the claimed option.
  • State whether a signal executes a prepared response, reopens judgment for an executor, or escalates upstream. Do not let local adaptation change the alternative set or its comparison criteria beyond the executor's authority.
  • Govern output integration, verification, recovery, and other consequential failure surfaces. Recheck delay, carrying, expiry, and foregone-benefit costs, then evaluate the composed method's own failure modes.

Scope

  • Problem match is a source-search relevance heuristic, not a claim about comparative yield, prevalence, source maturity, severity, longevity, or prestige.
  • Mechanism match transfers warrant only for the shared structural relation. Concrete implementation, unmatched features, and outcome effectiveness need target-side evidence.
  • The agent worked case requires both the strict preferential-codification condition and a separate execution-time information advantage. It does not claim that either condition is common.
  • Adversarial purpose is part of the military source cause, not a general target requirement. It transfers only when the target independently has an adaptive opponent. Hierarchy, rank, force structure, professional culture, violence, and source-specific doctrinal machinery do not transfer merely because the downstream delegation mechanism matches. The other contributions likewise bring no project-management bureaucracy, finance valuation machinery, full scenario machinery, policy institutions, or engineering organization.
  • The five-way method is a target-side construction, not a source doctrine or an empirically validated package. Interaction checks are necessary design questions, not a sufficient composition test.

Relevant Notes: