C.19.2 - Use-Bounded Apparatus Application
Type: Architectural (A) Status: Stable Normativity: Normative
C.19.2:0 - Use this when
Use this pattern when one practical result matters and a relevant method, model, formalism, assurance technique, ontology, or other direct-kind apparatus is available, but the work needed to configure and apply it may cost more than the result warrants. Start here whether one apparatus is already selected or a real choice among available alternatives has become current.
The first useful move is to name the practical use, result kind, claimed guarantee, constraints, and reuse horizon, then ask whether the next adaptation and application work can reach a useful result within the available budget. This keeps a small, adequate path small while letting repeated or high-consequence use justify richer configuration.
Not this pattern when. If an adequate way to obtain the result is still unexplained, use C.39. If usable material needs feasible variation and examination, use C.40. Use C.18 when the question concerns the generation account or archive/front stewardship. If the live question is a local choice over an existing option set, use C.11. If the blocker is an ontology conflation, use A.7.1; if it is a material conflict among FPF premises, use A.7.2.
The primary reader is an engineer, method or model selector, or technical lead. Use the instructions below to guide bounded application; the problem-facing result remains with the pattern that defines or tests it. Reader position, guidance and performed Work have different meanings. Section 4.1 gives the identity-bearing conditions when a claim about a Method, MethodDescription or dated Work is needed.
C.19.2:1 - Problem frame
A team can have a rich apparatus and still lack an economical way to use it. A maintenance group may need only one typed relation distinction for a one-off repair, an architecture team may need a configured model across hundreds of handoffs, and an assurance team may need a formal technique only after the required guarantee becomes stronger. In each case, displaying more apparatus is easier than proving that its setup work changes the result.
The concern is one bounded apparatus application for one declared use and horizon. The apparatus retains its direct kind; this pattern does not introduce a generic U.Apparatus kind. The application question is distinct from candidate generation, local choice, planning, performed work, and the domain result.
C.19.2:2 - Problem
Without a use-bounded application method, users make two symmetrical mistakes. They configure an entire rich basis because it is available, or they keep a lightweight path after recurrence, automation, transfer, or consequence would repay a more capable configuration. They may also call candidate generation “choice”, call a choice record a work plan, or treat an application note as the practical result.
The failure is not merely excess documentation. It obscures who performs the work, what result the work must produce, which guarantee is being claimed, and which neighboring pattern contains the defining content for the next question.
C.19.2:3 - Forces
| Force | Tension |
|---|---|
| Early value vs richer assurance | A useful result should arrive before exhaustive configuration, but higher consequence may justify deeper setup. |
| One current path vs live alternatives | Inventing rivals creates bureaucracy; ignoring genuine alternatives can lock in avoidable cost or loss. |
| Local economy vs reuse | One-off work favors a small path; repeated work can amortize configuration and improve transfer. |
| Direct kinds vs shared comparison | Unlike candidates must retain their kinds while being compared against one declared use and guarantee. |
| Guidance vs actual work | The instructions guide configuration or application; their description does not perform it. |
C.19.2:4 - Solution
C.19.2:4.1 - Identify the use, apparatus, method, performer and result
Before choosing how to proceed, identify what the application involves.
- Declared use: the practical question, direct result kind, claimed guarantee, non-negotiable constraints, and horizon.
- Selected or candidate direct-kind object: the method description, model, ontology module, formal technique, or other governed object being considered.
- Application guidance and Method: use this pattern’s instructions to guide the application. When identifying a particular Method, apply A.3.1. Treat a claim-bearing episteme as its
U.MethodDescriptiononly when A.3.2 establishes that the already admitted Method is its EntityOfConcern and it makes substantive claims about that way of doing. Using the guidance does not itself require that classification. - Performer and work: when a claim identifies particular dated configuration or application
U.Work, recover the admitted performingU.Systemthrough A.13 and independently admit the Work under A.15.1, including its enacted Method under A.3.1. Add F.6 afterward only when the present use needs precise assignment-bound attribution. In a short account, expose assignment identity, species, participants, or attribution detail only when the use relies on them, attribution is ambiguous, or source wording must be repaired. - Problem-facing result: the domain, engineering, assurance, architecture, or other subject-pattern result inspected after the work.
A person or an AI system can read these instructions and also perform the work. To attribute a particular performance, establish who actually carried it out.
C.19.2:4.2 - Select the application branch
One current apparatus. For an already selected method, recover its required operations, the inputs and settings it accepts, and the result and guarantee it supports. Check the conditions of this application and supply its inputs. The existing configuration may already suffice. If extra setup is needed, establish what it will enable and whether its cost fits the budget and the expected use. The single-apparatus path needs no OptionSet or ChoiceResult.
Candidate generation or reframing. If no adequate current object is available, use C.39 to develop a missing explanation of how to obtain the result, or C.40 to vary usable material and examine the difference. Use C.18 when generation-account or archive/front claims are needed. The declared use and eligibility basis can guide that work.
Local choice. Only when two or more already-available eligible alternatives, or another genuine local-choice question over a live set, are current, use C.11 for OptionSet, ChoiceRule, probing, and ChoiceResult.
Post-choice enactment. For a singular selected direct-kind object, use A.15.2 to plan its application when a plan is needed and A.15.1 for the dated application Work. C.24 is the pattern for sequencing, budgeting, checkpointing, and replanning only when the application involves tool-call work.
C.19.2:4.3 - Admit candidates by one use-bounded predicate
UseBoundedApparatusCandidateEligibilityPredicate@Context is a local eligibility predicate, not a U-kind, relation kind, or candidate-generation method. A candidate is eligible only when it has a credible adaptation path to the same declared use, direct result kind, claimed guarantee, scope and horizon, and non-negotiable constraints. A candidate that cannot meet one of those values stays outside the current option set rather than becoming a “weaker” member of it.
When the claimed guarantee or requirement is disputed, use C.11.DUA to appraise its purpose, contribution, burden and current force. Keep apparatus eligibility under the current declaration separate from that appraisal. Resume application under a revised declaration only when the change is available through the authority or agreement that governs it.
When choice is current, preserve the exact C.11 contract:
choose nownames one selected option or a retained tie-set;reject current setreturns to a named candidate-generation pattern or closes with no current application;probe againretains one probe and its epistemic budget because it can still change the choice;reroutenames the neighboring question and its subject pattern.
These four dispositions form the complete current C.11 result set. “Configure the rich basis”, “adapt again”, and “use a lighter method” are option or plan contents, not extra choice-result values. A tie is not a hidden winner.
C.19.2:4.4 - Perform the bounded application
- State the practical use, direct result kind, claimed guarantee, constraints, horizon, and current apparatus state.
- For an already selected apparatus, check its configuration and conditions of use (§4.2). Under
A.3.1, determine whether the original Method admits the proposed inputs, parameters, operations, required participants, result and acceptance conditions. If a defining condition changes, identify the revised way of doing the work or state what remains unresolved. Use the subject method to establish what guarantee the revised procedure supports. If candidates are missing, useC.39for an unexplained way to obtain the result orC.40for feasible variation and examination of usable material. AddC.18when a generation account or archive/front claim is needed. - When alternatives are current, identify them by their direct kinds and apply the shared eligibility predicate.
- For each current path, identify what must be learned, supplied, integrated or checked to meet the required guarantee. Estimate the work still needed. An existing configuration that meets the requirements can be used immediately.
- Compare only the dimensions that can change the current path for the declared use and horizon. These may include time and budget, prior exposure, post-threshold efficiency, transfer, retention, interoperability, downside and reversibility; use their subject-pattern values when needed. Do not compress them into an undeclared scalar.
- When choice is current, use its
C.11 ChoiceResult(§4.3). A request for another probe or a different decision identifies that next action. A rejected set leaves no selected application. A retained tie-set leaves open which member will be used. - When one apparatus is selected for use, carry out its application. Use A.15.2 when a plan is needed and C.24 when tool-call planning is the question. A precise dated-Work claim uses the performer and admission conditions in §4.1.
- Inspect the result against the criterion for this use. For another application, check that the accepted method, inputs and assumptions cover its conditions. To establish that the recipient can apply the method with a stated degree of help, examine that recipient’s own performance under those conditions. Keep an application or configuration note when reuse, dispute, automation or consequence makes it useful.
C.19.2:4.5 - Stop and reopen
Stop when the direct result is usable at the claimed guarantee and no further distinction or setup is expected to improve the declared use enough to justify its burden over the stated horizon. This is a positive result, not a claim that the unused apparatus is inferior.
Reopen when a consequential counterexample, failed result, changed use or guarantee, changed recurrence horizon, new candidate path, automation need, or changed adaptation cost alters the lawful path. Reopen only the affected application, candidate, choice, plan, or work question under its subject pattern.
C.19.2:5 - Archetypal Grounding
Repeated audited handoffs. A fleet has 400 maintenance handoffs each year. The required result is an audited maintenance-decision episteme with a fixed safety and interoperability guarantee. Candidate generation is complete: an ontology-backed Method and a lighter local decision Method are eligible; their descriptions are available. A spreadsheet macro is excluded because it cannot preserve required relation-occurrence identity.
In this constructed planning account, both eligible Methods satisfy the same guarantee and material non-time conditions. The complete expected setup and per-handoff effort are comparable person-hours, including the required learning, review and integration. Forty setup hours are available. The ontology-backed Method needs 40 hours initially and 6 minutes per handoff; the lighter Method needs 2 hours initially and 14 minutes per handoff.
| Horizon | Ontology-backed Method | Lighter Method |
|---|---|---|
| 400 handoffs | 40 + 400 × 6/60 = 80 hours | 2 + 400 × 14/60 = 95 hours 20 minutes |
| 100 handoffs | 40 + 100 × 6/60 = 50 hours | 2 + 100 × 14/60 = 25 hours 20 minutes |
The setup difference is 38 hours and the per-handoff saving is 8 minutes: the methods tie at 285 handoffs. At the declared 400-handoff horizon, C.11 returns choose now for the ontology-backed Method on the supplied equal-guarantee basis, saving an expected 15 hours 20 minutes. If the feasible horizon falls to 100, the same basis favors the lighter Method. These are conditional estimates, not observed maintenance results; changed costs or guarantees reopen their respective questions.
An admitted maintenance-information System performs the dated configuration and application Work using the selected ontology-backed Method. The account recovers the System’s A.13 core and independently admits the Work under A.15.1. Because the current question is whether setup cost is repaid across the handoffs and consumes no assignment-bound attribution, this example does not open F.6 or expose assignment identity, species, participants, or attribution detail. A later attribution-bearing claim adds F.6 through the same obtaining A.13 assignment.
A separately admitted maintenance-decision System then performs decision Work using the selected Method. Its A.13 core and independent A.15.1 Work admission are recoverable; this setup-cost example consumes no assignment-bound attribution, so F.6 remains unopened. Under A.6.1, application MaintenanceDecisionApplication-1 of operation DecideMaintenance, declared by MaintenanceDecisionMechanism-E1, returns AuditedMaintenanceDecision-1 under result declaration DecisionResult; no Work-to-result or production claim is made. To check the forecast, compare actual setup and effort per audited handoff on comparable work and verify that the required guarantee still holds. Reconsider the application if the revised costs, guarantee or expected number of handoffs change which path is worth using.
One-off naming repair. A team already has the typed-relation method for one local, reversible naming decision. No rival is live and the current method has a credible small path, so no option set or choice result is created. The team performs the minimal wording and typing work and returns a scoped terminology decision. Recurrence, integration, a failed result, or a stronger guarantee may later open candidate generation and choice.
Non-use. If the blocked result comes from missing telemetry while the method, state kinds, and action distinctions are already clear, return to measurement and evidence work. Apparatus selection cannot manufacture the missing observation.
C.19.2:6 - Bias-Annotation
The main bias is prestige-by-apparatus: richer form, newer tooling, or familiar terminology is treated as practical superiority. The mitigation is one declared result and guarantee, direct-kind candidates, adaptation/work cost, and a positive one-apparatus path. A second bias is automation optimism; cheaper computation does not erase evidence, human attention, integration, or consequence review.
C.19.2:7 - Conformance Checklist
| ID | Check |
|---|---|
CC-C19.2-1 | The declared use names a direct result kind, guarantee, constraints, and horizon. |
CC-C19.2-2 | A current single-apparatus case proceeds without a fabricated OptionSet or ChoiceResult. |
CC-C19.2-3 | Candidate generation, local choice, planning, tool-call planning, dated work, and direct result retain their subject patterns. |
CC-C19.2-4 | Every candidate retains its direct kind and satisfies one explicit eligibility predicate for the same use and guarantee. |
CC-C19.2-5 | Any C.11 result is exactly choose now, reject current set, probe again, or reroute. |
CC-C19.2-6 | The intended-reader position, guidance, any independently identified U.Method and qualifying U.MethodDescription, performing U.System, dated U.Work and problem-facing result remain distinct. Every precise performer has an A.13 core, and the Work is independently admitted under A.15.1. F.6 is required only when the receiving use also needs precise assignment-bound attribution. A short projection exposes assignment identity, species, participants, or attribution detail only when that use relies on them, attribution is ambiguous, or source wording must be repaired. |
CC-C19.2-7 | The first useful result and stop are practical, and every reopen condition changes a named next action. |
CC-C19.2-8 | Candidates retain their direct kinds and comparisons use declared, admissible operations. |
C.19.2:8 - Common Anti-Patterns and How to Avoid Them
| Anti-pattern | Repair |
|---|---|
| Configure everything because the basis is rich. | Name the useful-result threshold and retain only setup work with expected return. |
| Invent a rival to make the method look comparative. | Use the one-apparatus path until candidate or choice work is current. |
| Call candidate generation a choice. | Use C.39 to explain a missing way or C.40 to vary and examine usable material; let C.11 operate on the existing eligible set. Add C.18 when the generation account or archive/front claim matters. |
Treat the apparatus-choice ChoiceResult as an application plan or the problem-facing result. | Keep selected object, plan, dated work, application note, and domain result separate. |
Let a U.MethodDescription episteme, its described Method, a plan, option row, publication, or reader position perform Work. | State in ordinary language that an admitted System performs dated Work using the Method. Recover its A.13 core and independently admit the Work under A.15.1; add F.6 only when the present use needs precise assignment-bound attribution. Expand assignment and attribution detail only when that use needs it, attribution is ambiguous, or the source wording must be repaired. |
| Rank heterogeneous candidates under one hidden “depth” score. | Preserve direct kinds and compare only declared use-bearing dimensions without hidden scalarization. |
C.19.2:9 - Consequences
The pattern makes cheap use a first-class result. Teams can apply one current method without ceremony, yet repeated or consequential uses can justify configuration or a real choice. The cost is naming the result, guarantee, applicable condition or predicate, and enough work economics to support the path. That bounded burden prevents apparatus display from replacing practical return.
C.19.2:10 - Rationale
Application is the common job; selection is conditional. Starting from application preserves the frequent case in which a subject pattern is already selected and only the next useful adaptation matters. When alternatives are real, existing generation and choice patterns provide better result semantics than a fifth local decision vocabulary. Direct-kind discipline prevents an umbrella comparison word from becoming a new ontology.
The aphorism is: make the apparatus earn its setup work.
C.19.2:11 - SoTA-Echoing
| Practice question | Current practice and source | FPF alignment | Disposition |
|---|---|---|---|
| How much reasoning effort repays its cost? | Resource-rational analysis ties computation to action value under bounded resources (Lieder & Griffiths 2020, with Russell & Wefald 1991 as lineage). | The useful-result threshold and non-dominated next move govern setup and application work. The repeated-handoff and one-off cases show the practical difference. | Adapt. FPF retains multiple separately defined dimensions rather than one universal value-of-computation scalar. |
| Should ontology/formality be complete before use? | Demand-driven incremental formalization starts from useful work and deepens when use repays it (Shipman & McCall 1999 as a historical HCI/knowledge-base precursor). | The one-apparatus path allows value before total configuration, with reopen on recurrence, failure, automation, or stronger guarantee. | Adopt as continuing lineage. No HOS tool or single formality ladder is imported. |
| Do different questions warrant different apparatus and outputs? | Current competency-question work distinguishes purposes and expected products rather than imposing one ontology checklist (Keet & Khan 2024). | Eligibility is tied to one declared use/result/guarantee; the pattern does not universalize one method or checklist. | Adapt. Question differentiation changes candidate eligibility and return. |
| How should concern and purpose constrain a model or architecture description? | ISO/IEC/IEEE 42010:2022 keeps concerns, stakeholders, viewpoints, and architecture descriptions explicit without prescribing one tool. | Declared use, guarantee, and direct result keep apparatus application purpose-bound while subject patterns retain their kinds. | Adapt. Use the purpose/concern discipline without importing architecture-description ontology into every apparatus. |
These sources change the positive method: the user starts from a result, may obtain value before total configuration, opens choice only for a live set, and reopens when use economics or guarantee changes.
C.19.2:12 - Relations
- Coordinates with:
C.39for a missing explanation of how to obtain the result;C.40for feasible variation and examination;C.18for the generation account and archive/front stewardship;C.19for explore/exploit policy over still-live candidate pools;C.19.1for scale-amenable bearer preference;C.22.1for adaptation signatures;E.23for repeated improvement; andC.31.ASAPfor architecture-scale preference. - Uses conditionally:
C.11only when a local-choice question over a live eligible set exists. It consumes, but does not extend, the fourChoiceResultdispositions. - Hands off enactment to:
A.15.2for work plans,A.15.1for dated work, andC.24only for tool-call enactment planning. The domain pattern contains the defining content for the practical result. - Description-level specialization:
A.7.1narrows the method claims stated here for consequence-guided ontology analysis. It retains the declared use, problem-facing result, claimed guarantee, horizon, useful threshold, separation among plan, Work, and result, stop, and reopen. It retains the candidate and choice branch only when that branch is triggered. This wording adds no relation occurrence between the described Methods and asserts neitherU.SubkindOfnor a world relation.