Part VI - Reconstruct Practice and Explain Method History

RMP.10 - Reconstruct a Candidate Method Account from Practice Evidence

Status: Stable

Type: Method pattern

Primary working result: a claim-to-evidence dossier with scoped candidate Method accounts, contradictions and rivals, results of any selected further checks, evidence limits, and the next receiving use—or a justified lowering when the evidence supports only Work description, a local workaround, tool behavior, or an unresolved grouping.

RMP.10:0 - Use This When

Use this pattern only after A.3.1.MR has recovered an ordinary candidate account or honest lower result and a named receiving decision still cannot be made without a larger evidence programme. Typical triggers are heterogeneous settings, disputed variants, tacit cues or judgments, incomplete records, consequential use, or a need for a stronger held-out application.

Begin with the receiving decision and the few candidate-account claims whose uncertainty changes it. Select evidence forms for those claims rather than collecting every available trace or interview.

The first useful result is a claim-to-evidence matrix and sampling decision. It states what would support, contradict, split, or lower each claim and why the selected evidence form can distinguish those outcomes. This makes the specialist burden visible before the study grows.

Use the larger evidence programme only when ordinary candidate recovery is insufficient for the named decision. No observation, interview, log, or synthesis backdates a Method into past Work, promotes a log into a MethodDescription, or admits a Method without A.3.1.

RMP.10:0.1 - Working Distinctions

Evidence or result	Contribution retained here	Boundary
occurrence observation and artifacts	overt actions, results, versions, coordination, and traces for one Work occurrence	observed recurrence does not identify a Method by itself
Critical Decision Method (CDM) probe	recalled critical-incident cues, options, judgments, and counterfactual reflections	recall and interviewer effects remain visible; CDM supplies no event-log claim
event-log analysis	recorded recurrence, sequence, deviation, and conformance relative to declared log semantics	tool traces describe recorded events, not intent, tacit contribution, or a MethodDescription
contextual observation or protocol elicitation	coordination, workarounds, or otherwise inaccessible reasoning when their particular risk is named	these moves are selected separately; CDM and process-mining sources do not support them by proxy
expert synthesis	claim-to-evidence matrix, decision-relevant sampling, contradiction-by-scope rule, choice of further checks, and stopping rule	disclose the synthesis rather than attributing it to the specialist sources
candidate Method account	a scoped episteme about a possible reusable way of doing	Method identification remains with `A.3.1`

RMP.10:1 - Problem Frame

Ordinary records show only part of practice. Logs favor tool-visible events, procedures favor intended behavior, interviews favor memorable episodes, and successful outcomes can hide competent minority variants or recovery Work. A stronger reconstruction therefore needs several evidence forms and an explicit account of which claim each supports.

More evidence is not automatically better. The programme earns its cost only when its combined result changes candidate-account content, scope, rivals, or the receiving decision beyond what A.3.1.MR already returned.

RMP.10:2 - Problem

A trace-only reconstruction turns log regularity into a Method. An interview-only reconstruction turns confident recall into recurrence. A majority-vote synthesis erases variants and failure cases. An expanding study collects material without a stopping decision.

The resulting account looks rich yet cannot say which claim each item supports, what would contradict it, or which remaining uncertainty matters to its use.

RMP.10:3 - Forces

Force	Tension
Tacit contribution	Important cues and judgments may be absent from records, while elicitation can distort them.
Heterogeneity	Several performers and settings improve discrimination, while indiscriminate sampling raises cost.
Contradiction	Variants may be real, while some disagreement is error or incomplete evidence.
Tool visibility	Event logs are precise about recorded fields, while invisible coordination and intent remain outside them.
Further discrimination	An unseen case can test prediction or transfer, while an unavailable or decision-irrelevant case adds no useful assurance. Retrofitting after inspection loses the intended discrimination.
Specialist burden	Consequential decisions may justify a programme, while ordinary recovery needs an affordable stop.

RMP.10:4 - Solution

RMP.2 supplies a selected evidence strategy or a justified stop when that design question remains open; a compatible supplied selection is enough. RMP.3 makes its consequential operational choices executable, and RMP.4 preserves the actual observation, elicitation and log trace with its access limits. The reconstruction below makes the specific claim-to-evidence connection: which uncertainty about a reusable way needs which evidence, how setting-specific variants survive contradiction, and what bounded account the combined material supports. Reuse an adequate design or trace; the programme needs the missing discrimination, not another copy of those results.

Organize evidence around decision-changing claims, keep evidence forms distinct, resolve contradiction by scope, and select further checks only where their attainable result warrants the burden. Return scoped accounts or a lower result when the receiving decision has an adequate basis.

RMP.10:4.1 - Pattern-Use Unfolding

Name the receiving decision and unresolved claims. Include only intended result, entry conditions, reusable operations or invariants, cues and decisions, participants and capabilities, information and artifacts, support, variation, recovery, and stop claims that can change the decision.
Build the claim-to-evidence matrix. For each claim, state what observation could support, contradict, split, or lower it and why the evidence form can distinguish those outcomes for that claim.
Sample for decision-relevant variation. Select cases that differ in success or failure, routine or exceptional conditions, performer, setting, and tool or provider. Add a disconfirming case when the emerging account explains only the convenient sample. Use no universal participant count.
Keep evidence forms distinct. Use occurrence observation/artifacts for overt actions and results; CDM for recalled critical-incident cognition; event-log analysis for recorded recurrence and deviation. Select contextual or protocol evidence only for a named gap and risk.
Keep occurrence and candidate layers separate. Record each Work occurrence before abstracting an account. Preserve missing fields, observer effects, recall limits, tool coverage, and source access.
Resolve contradiction by scope, not vote. Compare performer, setting, outcome, and evidence form. Correct a supported error; otherwise split variants, narrow applicability, retain rivals, or lower the claim.
Choose whether a further check is needed. Identify the unresolved claim and the possible answers that would change the receiving use. Compare the attainable discrimination with evidence access, practitioner capacity, delay, and the whole acquisition burden. Select a held-out case when prediction or transfer to that case matters and can be examined usefully. State each serious account’s predictions about cues, decisions, variation, and result before inspecting it; record surprise and revision. Another claim may instead need a source comparison, a targeted elicitation, or no additional check because the existing evidence already supports the bounded decision. An unavailable check leaves its claim uncertain; it does not oblige an unproductive study or license a stronger claim.
Stop or lower. Stop when the scoped accounts and explicit uncertainty dispositions support the receiving decision. Rivals may remain unresolved when distinguishing them would not change that decision. Narrow the use or defer it if a load-bearing uncertainty remains. Return no candidate account when the evidence supports only a Work description, local workaround, tool behavior, or unresolved grouping.

RMP.10:4.2 - Record the Result

Result position	Required content
receiving boundary	Decision, situation, consequence, candidate accounts, and unresolved claims.
claim-to-evidence matrix	Claim, discriminating observation, selected evidence form, support/contradiction/split/lowering rule, and stopping use.
sample and occurrences	Decision-relevant variation, each Work occurrence, source access, missing fields, and evidence-form limits.
synthesis	Scoped accounts, retained variants/rivals, contradiction disposition, and attribution of each synthesis move.
selected further checks, when needed	Claim and use at issue, why the attainable answer warrants the burden, actual result and remaining uncertainty. For a held-out check, also preserve predictions made before inspection, surprise and revision. When existing evidence suffices, state that basis in the return; no empty held-out record is needed.
return	Account(s) or lower result, evidence limits, next receiving use, and reopen condition.

RMP.10:4.3 - What Changes in Practice

Teams stop equating a discovered sequence or interview narrative with the Method. They spend specialist effort only where a receiving decision needs it, know which evidence form supports which claim, and can retain two scoped variants instead of averaging them into a misleading universal account.

RMP.10:5 - Archetypal Grounding

RMP.10:5.1 - Ordinary `A.3.1.MR` Stop

A maintenance team has six source-traceable calibration Work records using the same released instruction, input schema, equipment configuration, decision authority, and result meaning. One failed record contains an explicit missing-input stop already described by the candidate account. The receiving decision needs only a source-traceable account for comparing two instructions; no tacit cue, disputed variant, consequential transfer, or held-out prediction changes that decision.

A.3.1.MR returns the source-traceable candidate account, any real rival and the important gaps, with its provisional status and comparison-use limits. No distinguishing question or available follow-up is needed for this instruction comparison. RMP.10 is not entered; interviews and process mining would add cost without changing the result.

RMP.10:5.2 - Genuine Escalation: EC-417 Evidence Reconciliation

The EC-417 decision cannot determine whether internal and supplier-originated evidence reconciliation are one account or scoped variants from ordinary records alone. Here the consequential evidence-closure question, available case access and practitioner capacity justify the larger programme and its whole burden. This constructed reconstruction uses a twelve-case development sample plus one held-out case; that count is not an entry rule for other recoveries:

four firmware-only and four internal harness-plus-firmware releases accepted versioned provisional pinout evidence before safety closure;
four supplier-originated releases required signed evidence at closure;
four reconciliation boards were observed;
CDM probes involved two supplier, two software, one hardware, and one safety practitioner; and
a thirteenth supplier-originated change was reserved unseen for the held-out discriminator.

Evidence move	Scenario contribution	Limit retained
observation and artifacts	overt evidence versions, board records, integration bundles, approval states, and closure results	no population estimate or tacit-cue claim
CDM probes	cues, considered options, judgments about provisional versus signed evidence, and counterfactual reflections in critical incidents	retrospective recall and interviewer effects remain
event-log analysis	recorded recurrence and deviations in evidence-version, integration, approval, and closure events	log fields do not establish intent or a reusable Method
disclosed expert synthesis	twelve-plus-one sample, internal/supplier scope split, contradiction rule, held-out discriminator, and stop	not attributed to CDM or process-mining sources

Before inspecting the thirteenth case, the team recorded three rival predictions:

Account or rival	Cues and decision predicted	Variation and result predicted
`C-Evidence-Reconcile-Internal` applied beyond its observed scope	internally controlled versioned evidence would be the decisive cue; the internal reconciliation role could decide the branch without a distinct supplier approval	supplier origin would not change the branch; closure could proceed from the internal reconciliation result without signed supplier approval
`C-Evidence-Reconcile-Supplier`	supplier-owned configuration and restricted geometry would be decisive; supplier responsibility and signed approval would remain required for closure	reversible early integration could occur, but a signed-delta comparison and affected verification would precede either closure or an explicit missing-approval stop
`R-Rig-Capacity-Only`	rig unavailability, rather than evidence origin, would trigger the hold and next-slot decision	the same branch would occur for internal and supplier changes and would clear when rig access returned

The held-out case then showed cues of supplier-owned geometry, restricted provider access, and available rig capacity; the decision allowed reversible early integration but withheld closure pending signed supplier approval; the variation required a signed-delta comparison and one affected verification; and the result was truthful closure after signed approval arrived at D-8. No load-bearing observation surprised the supplier-scoped account. The available rig contradicted the capacity-only rival for this case, and the retained supplier-approval requirement contradicted the internal-account extrapolation.

The resulting disposition is to keep C-Evidence-Reconcile-Internal for the eight internal cases, keep and narrow C-Evidence-Reconcile-Supplier to supplier-owned configuration with the signed-evidence branch, and retain rig availability as a support condition rather than the account that explains this split. The scope split is not decided by majority count. One held-out supplier case leaves transfer to other suppliers, other configuration classes, and missing-approval situations uncertain; neither account is admitted as a Method, and no effectiveness or population claim is made.

RMP.10:5.3 - A Specialist Programme Can Finish without a New Case

A retired repair service used two routing practices. Ordinary A.3.1.MR recovery from its mixed logs and instruction editions cannot tell whether an urgent route was a repeatable cue-based branch or an occasional supervisor override. The receiving museum archive needs a qualified comparison of the two candidate accounts for this service’s final year. It is not deciding to reopen the service or transfer either account to current work.

The larger programme is justified because the disputed cue is absent from the logs. Its claim matrix selects the dated routing sheets and targeted critical-incident interviews with the two available former dispatchers. The sheets record whether replacement stock was present; the logs record only the chosen route. Linking them to the same occurrences shows that the disputed route accompanied absent stock in the accessible cases. Each dispatcher recalls using the stock cue and describes one exception requiring supervisor involvement. Their recalled reasons remain elicited accounts; the routing sheets support recorded stock and routing, not unrecorded intentions or universal recurrence.

The evidence supports a stock-conditioned candidate branch with a separately retained supervisor exception. The archive can now compare that account with the earlier instruction and state the unresolved gaps. A new live repair cannot be observed because the service has closed. Simulating one would test present performance under new conditions, which the archive’s comparison does not need. The programme therefore returns the bounded account, source links, recall limits and unresolved coverage outside the accessible records. It makes no claim about transfer or effectiveness, and it does not identify a Method through these observations.

The programme entered RMP.10 and completed useful specialist work. Its stop rests on the adequacy of that work for the receiving comparison. If a later operator wants to reuse the reconstructed branch, that changed use reopens the evidence and check choice; the archive’s descriptive result cannot settle it.

RMP.10:6 - Bias-Annotation

Recurring bias	Likely drift	Repair
trace bias	Recorded sequence becomes the Method or its complete description.	State log semantics and combine only with evidence needed for the claims.
interview bias	Memorable explanations become recurrence or causal evidence.	Keep CDM claims to cues, options, judgments, and reflections with recall limits.
majority bias	The largest variant erases a competent minority or setting-specific branch.	Resolve contradictions by performer, setting, outcome, and evidence form.
method-brand bias	Using a named evidence Method lends authority to the whole synthesis.	Attribute each sourced move and disclose the project synthesis separately.
programme bias	More cases are collected because the study is open.	Tie sampling and stop to the receiving decision and load-bearing claims.

RMP.10:7 - Conformance Checklist

A.3.1.MR has already returned an ordinary result, and a named receiving decision justifies the larger burden.
The first useful result is a claim-to-evidence matrix and decision-relevant sampling choice.
Observation/artifacts, CDM, event-log analysis, contextual/protocol evidence, and expert synthesis keep distinct contributions and limits.
Each Work occurrence is recorded before candidate-account abstraction.
Contradictions are resolved by scope or retained, not averaged by vote.
Any further check addresses a use-changing uncertainty and has an attainable result worth its burden; an adequate existing basis can support a bounded stop.
A selected held-out discriminator states predictions before inspection and records surprise/revision.
Scoped variants and rivals remain visible.
The result can lower to Work description, local workaround, tool behavior, or unresolved grouping.
No evidence form identifies a Method, creates a MethodDescription, backdates Work, or establishes causality by itself.

RMP.10:8 - Common Anti-Patterns and How to Avoid Them

Anti-pattern	Repair
“Mine the log and recover the Method.”	Recover only recorded events; use the claim matrix to identify missing intent, tacit, and context evidence.
“Interview experts until their stories converge.”	Sample for decision-relevant variation and preserve contradiction, recall limits, and minority variants.
“Use every evidence form.”	Select only forms whose possible results can change a claim or the receiving decision.
“A specialist programme must always obtain another unseen case.”	Select the check from the unresolved claim and receiving use; finish on adequate existing evidence, or narrow/defer the use when uncertainty matters.
“The held-out case fits after we revised the account.”	Record predictions before inspection and report surprise rather than retrofitting.
“The reconstructed account is now an admitted Method.”	Return the account and send Method identity to `A.3.1`.

RMP.10:9 - Consequences

For consequential decisions, practitioners reconstruct candidate accounts claim by claim, make the sources traceable, and can retain variants. The receiving decision gains stronger accounts or an honest lower result rather than a single fluent reconstruction.

The cost is specialist sampling, evidence access, and analysis. Some evidence will remain unusable for the desired claim, and a well-run programme can still return no Method candidate.

RMP.10:10 - Rationale

No single evidence form covers overt action, recorded recurrence, recalled judgment, tacit coordination, and cross-setting variation. Combining these evidence forms is useful only when each is connected to a claim and a receiving decision. The matrix, scope rule, justified check choice, and stop make that combination testable without claiming a universal research protocol.

RMP.10:11 - SoTA-Echoing

Source	Retained contribution	Use boundary
Current FPF `A.3.1.MR`	Ordinary several-occurrence recovery, real rival accounts, source-to-claim trace, useful distinguishing questions and honest lowering. A provisional account can finish without irrelevant or unavailable follow-up.	RMP.10 begins only when a named decision needs and warrants its larger programme; the ordinary question is not a study commitment.
Klein, Calderwood, and MacGregor, Critical Decision Method	Retrospective critical-incident probes for cues, options, judgments, and counterfactual reflections.	CDM supplies no event-log, recurrence, population, or causal result.
IEEE Task Force, Process Mining Manifesto	Discovery, monitoring, and conformance analysis over recorded event logs.	Event-log regularity establishes neither intent, tacit contribution, Method identity, nor causality.
Ordinary observation and artifacts	Overt occurrence evidence and result traces.	Observation supplies only what its access, scheme, and window support.

The claim matrix, decision-relevant sampling, contradiction-by-scope rule, conditional further checks, and combined stop are the bounded DPF synthesis. Reopen when a current evidence Method changes one of those moves, or when representative uses show that the extra burden does not improve account scope or the receiving decision.

RMP.10:12 - Relations

Use RMP.2 when research-design selection is still open and RMP.3 for the operationalization of a selected design; RMP.4 preserves the actual material. Use compatible supplied results directly. RMP.5 can criticize the resulting account under its actual evidence relation; RMP.9 receives a separate difficulty in the research Method itself.
A.3.1.MR supplies the ordinary floor and specialist-exit question; RMP.10 does not replace it.
A.15.1 governs any claimed Work occurrence; A.10 governs evidence paths and provenance; A.3.1 alone governs Method identification.
C.39.RP may add returned accounts provisionally to a repertoire. A.3.1.QM may qualify an account without changing its status.
Later description, support, trial, or architecture patterns consume only the scoped account, evidence limit, and next use their entry conditions admit.
RMP.11 addresses a different question: why and how a Method architecture differentiated. Use it for that explanation and any needed causal-use support assessment; evidence reconstruction here supplies no causal conclusion by itself.

RMP.10:End

RMP.11 - Explain Method History with Discriminating Evidence

Status: Stable

Type: Method pattern

Primary working result: a differentiation account with a dated sequence, named pressure/response/variation/selection links, serious rivals, diagnostic observations, and graded claims. When a receiving decision needs causal reliance, RMP.11 also returns one separate C.28 causal-use support result; the downstream architecture decision remains a third episteme.

RMP.11:0 - Use This When

Use this pattern when an identified Method, independently established family inclusion, or named relations among Methods are treated as natural or inevitable and a present decision depends on how that differentiation arose. Typical questions concern why a variant appeared, why one contribution separated from another, why a provider or profession retained a form, or why a relation changed under pressure.

Begin by bounding the exact Method, established family inclusion, or relation change to explain and the decision that will use the account. Decide immediately whether the receiver needs only a dated descriptive account or intends to rely on a causal conclusion.

The first useful result can be a descriptive stop: a dated sequence with source and evidence limits. When causal reliance is live, name the causal-use question, claim kind, causality-ladder rung, and required support before constructing the explanation.

Do not use a project-local grouping label as a professional-family fact. Do not use RMP.11 for decorative history, chronology, popularity ranking, or present architecture choice alone.

RMP.11:0.1 - Working Distinctions

Name used here	Meaning
explanandum	One identified Method, established family inclusion, or named Method relation and the differentiation that the present decision needs explained.
differentiation account	An episteme containing dated description, link claims, rivals, diagnostic observations, grades, limits, and reopen conditions. It does not select today’s architecture.
descriptive sequence	Dated source editions, Work evidence, variants, disappearances, and decision points with no causal-use claim.
link claim	A proposed relation among pressure or environmental change, affected actor or carrier, response or variation, selection/retention, and Method or relation change.
rival	A materially different account such as constraint, provider/tool availability, regulation, authority, diffusion/fashion, recording artifact, redesign, or survivorship bias. Rivals may coincide.
causal-use support result	A separate `C.28` result that states the question, claim kind, rung, actual support-component refs, threat screen, verdict, supported/unsupported uses, limits/window, and reopen.
non-causal observation or hypothesis	A separately named descriptive result that may justify a bounded probe without licensing a causal claim.
downstream architecture decision	A separate authorized choice that cites only support allowed by the causal verdict and keeps its own criteria, authority, stops, and burden.

RMP.11:1 - Problem Frame

An account of professional Method change relates variants recorded in successive sources to changes in institutions, tools, regulation and local Work. It examines deliberate redesign, diffusion, fashion, retention and loss. Questions about variants and selection help develop competing accounts. Process-tracing research supplies methods for examining the sequence within a case, rival explanations and diagnostic observations.

For causal reliance on either contribution, use C.28. A plausible sequence can support description and hypothesis while leaving an interventional claim unsupported. The pattern therefore separates differentiation, causal support, and current choice.

RMP.11:2 - Problem

Chronology is often narrated as cause: a pressure appears, a practice changes, and the account says “therefore”. Search frequency or institutional prominence is treated as adoption evidence. Surviving versions hide abandoned variants. A project-local label is promoted into a professional family.

The downstream decision then consumes a causal premise that no identified result supports, or a useful descriptive account is discarded because it cannot prove causality.

RMP.11:3 - Forces

Force	Tension
Useful history	Dated differentiation can expose alternatives and assumptions, while decorative chronology adds no decision value.
Diagnostic depth	Within-case observations can discriminate rivals, while narrative confidence does not supply identification or estimation.
Evolutionary language	Variation, selection, retention, and loss organize inquiry, while analogy can overstate causal mechanism.
Causal cost	Some decisions need causal reliance, while many can stop with description or a non-causal trial hypothesis.
Present authority	History can inform a choice, while it neither selects today’s architecture nor grants decision authority.

RMP.11:4 - Solution

RMP.4 supplies the surviving records and their access, reconstruction and loss limits. RMP.7 supplies a corpus boundary and qualified source-dependence account where that contribution is already available. The historical construction below then explains the particular Method differentiation: identify competing accounts, derive the observations that separate them, compare diagnostic weight conditional on the sources already used, and carry a supported historical result into its receiving question. A corpus synthesis or chronological list does not replace that construction. C.28 still governs a separately claimed causal use.

Recover dated differentiation and rivals first. Apply the full C.28 causal-use assessment only when another decision needs a causal conclusion, and return every result with its own identity and use limit.

RMP.11:4.1 - Pattern-Use Unfolding

Bound the explanandum. Identify one Method under A.3.1, an independently established family inclusion, or named Method relations and the differentiation to explain. A local grouping can supply candidates and questions, not the family fact.
Name causal use or choose a descriptive stop. State the receiver’s exact question. If causal reliance is needed, name CausalUseClaimKind, target CausalityLadderRung, comparator or intervention where applicable, and the link claims. Otherwise return dated description without causal vocabulary.
Establish the dated descriptive sequence. Recover source editions, Work evidence, variant appearances and disappearances, abandoned forms, and decision points. A timeline is an index, not an explanation.
Propose explicit link claims. Name the pressure or environmental change, affected actor or carrier, response or variation, selection or retention process, and resulting Method or relation change. Use mechanism only in a source-local sense unless an A.6.1 U.Mechanism is separately identified.
Generate serious rivals. Consider demand or constraint, provider/tool, regulation, authority, diffusion/fashion, recording-artifact, deliberate-redesign, and survivorship accounts where they are plausible for this history. Explain which alternatives compete and which can coexist; do not invent a demand account merely to fill a category.
Derive and collect diagnostic observations. Before choosing observations that are easy to collect, state which dated records, version differences, decisions, Work evidence, interviews, abandoned variants, negative cases, or cross-setting contrasts are expected under each link and rival. Preserve provenance and source dependence. Weigh the observations through :4.4 before grading the resulting claims; provenance alone does not establish diagnostic strength.
Run the complete C.28 boundary when needed. Issue one identifiable support result with question ref, claim kind, rung, actual evidence-path/data-regime and specialist-result refs, common-threat-screen ref, verdict, supported/unsupported causal uses, limits/window, and reopen. Raw observations or source names are not support-component results. A missing identification, bound, or estimate required for this named causal-use question, or an unresolved live threat to that use, lowers the verdict. Do not commission an estimate merely because another kind of causal question would need one.
Return distinguishable results. Grade differentiation claims as observed, source-supported, inferred, expert-estimated, contradicted, or missing and retain rivals. Keep the causal result separate. A downstream choice cites only causal reliance allowed by the verdict and retains its own authority. An unsupported or undecided result may not serve as positive causal evidence; a separately named non-causal observation or hypothesis may still support a bounded trial.

RMP.11:4.2 - Record the Results

Result	Required content
differentiation account	Explanandum, dated sequence, source editions and Work evidence, link claims, rivals, diagnostic comparisons with their assumptions and source dependencies, grades, limits, and reopen.
optional causal-use support result	Question ref, claim kind, rung, actual support-component refs, threat-screen ref, verdict, supported/unsupported uses, limits/evidence window, and reopen.
optional non-causal observation or hypothesis	Separate identity, descriptive basis, bounded design use, inferences this result does not support, and reopen.
downstream handoff	The result and its allowed use supplied to another decision, which retains its own authority.

RMP.11:4.3 - What Changes in Practice

Practitioners can use history without turning it into inevitability. A dated account can reveal forgotten variants and serious rivals; a causal stop can prevent unsupported reliance; and a current architecture decision can still authorize a reversible probe from a separate non-causal observation.

RMP.11:4.4 - Weigh Historical Evidence between the Actual Accounts

A reliable source can contain a fact that all serious accounts predict. Such a fact helps establish the sequence without favoring one explanation. Conversely, a potentially revealing statement may have little weight because its date, authorship or relation to the events is uncertain. Keep these questions separate from the claim grades in :4.1.

Make the comparison specific. For the same finding, ask how expected it would be under each serious account, given the evidence already incorporated. Explain the difference. Compatibility with one account alone is insufficient. If two causes can operate together, compare their proposed contributions or a combined account rather than forcing an exclusive choice.
Examine how the finding became available. Distinguish an event from a later assertion about it. Ask what the source could observe, why it was recorded, and how it was selected and preserved. A missing record weighs against an account only to the extent that the proposed event should have produced an accessible surviving record.
Use additional information, not additional copies. A brochure, an interview repeating it and an article citing that interview may transmit one assertion. Assess the later item conditional on what is already known; do not count it as independent corroboration. Separately established timing or Work evidence can add information even when it appears in the same archive.
Return a conditional comparison. State which finding favors which account and which assumptions carry that judgement. Preserve an indistinguishable result when the accounts predict the same observations or source uncertainty defeats discrimination. Check whether the preference survives plausible changes in disputed coding, source selection, dependence or the weight given to a key finding. If it does not, make that fragility part of the result. Use numerical analysis only with a justified observation model; do not invent probabilities or turn named test types into automatic grades.
Reconsider the relevant evidence when the account changes. A newly discovered actor, date or mechanism may alter what earlier findings would mean. Compare the revised account with the relevant prior corpus as well as the new item, keeping source dependence visible. Do not attach the old conclusion to a newly worded hypothesis without this reconsideration.
Return only the supported use. The comparison can identify a promising next inquiry or a bounded historical hypothesis. When a receiver needs causal reliance, supply this contribution and its limits to C.28; the comparison alone is not its complete support result. If a revised premise matters to an existing repertoire or choice, use its evidence dependencies to identify the affected C.39.RP, A.3.1.QM or C.32.MWA.PP result and reassess that reliance. A change to historical attribution need not invalidate an independently qualified current Method.

RMP.11:5 - Archetypal Grounding

RMP.11:5.1 - Descriptive Stop

This is a constructed design-office history, not a report about a real organization. The office must reconcile names and source editions after drawing signoff differentiated. For this case, independent A.3.1 identification and A.3.1.MV variant qualification have already established common Method M-Signoff-0 and local variants M-Signoff-E and M-Signoff-M. All return approve/reject for a drawing, but the electrical variant requires clearance and harness-revision checks, while the mechanical variant requires load-case and fit checks; each withholds approval when its required check is absent. Those reusable differences are not inferred from separate form names. No professional-family or composite-Method claim is made.

The explanandum is this local differentiation and its documented timing. The receiver needs to distinguish the introduction of the different checks from their later representation in separate electronic forms. It does not need a claim that one pressure caused the split or that either variant is better.

The case premises also admit ElectricalReviewer-E and MechanicalReviewer-M as Agents for their respective checking under A.13, and their 9 and 18 April occurrences WE7 and WM4 as Work under A.15.1. The Work enacts M-Signoff-E and M-Signoff-M, respectively; E1 and M1 are the instructions used, not the enacted Methods. The bounded source set contains these case facts:

Date and source	Recoverable fact and limit
15 January 2024, signoff instruction S3	One common signoff account covers both drawing types and checks that required evidence is present. S3 contains neither variant’s distinct test-and-withhold rule. A source omission alone does not prove that nobody already performed such checks.
4 March, exception log X2; 11 March, proposal P1	X2 records two returned drawings, one lacking an electrical-clearance check and one lacking the mechanical load case. P1 cites these exceptions and proposes the two domain-specific rules. It also retains the alternative of common signoff with expert escalation; the record does not show that the exceptions alone determined the choice.
2 April, approval A14 and instructions E1/M1	A14 approves the two local variants for their respective drawing types. E1 and M1 state the different required checks and failure returns. The common-with-escalation proposal is not selected in A14, but the deliberation minutes are missing.
9 and 18 April, signoff Work WE7/WM4 and their annotated results	ElectricalReviewer-E follows E1 in WE7 and withholds approval for missing clearance evidence; MechanicalReviewer-M follows M1 in WM4 and withholds it for a missing load case. The records support these two enactments before electronic-form separation, not population-wide use or the first-ever occurrences.
6 May, PLM migration note T2 and form edition F2	Separate electronic forms and approval-record names are introduced. T2 says they carry the existing E1/M1 rules. It supplies no new variant semantics and no evidence that the forms caused earlier Work.

Before inspecting the April records, the office has two serious explanations. R-demand says domain-specific omissions prompted the proposal and selection of different rules; it predicts a problem record, a response proposal, and distinct checks before the form migration. R-carrier says the apparent split is only a naming or recording effect of that migration; it predicts unchanged checks beneath the new names. R-carrier is initially plausible because the searchable PLM history starts with F2 and makes the split look like a May event.

Comparing E1/M1 and WE7/WM4 with S3 and T2 is diagnostic: different reusable checks and their enactments are recorded in April, before F2. This contradicts R-carrier’s strong claim that only names changed in May. It fits R-demand, but the available records do not settle why the office selected separate rules over common signoff with escalation, or how other pressures contributed. Missing deliberation evidence leaves that link unresolved here; a separate control office is not a universal requirement for within-case causal inquiry. A weaker carrier contribution to later retention remains possible and untested; rejecting the naming-only account does not prove its rival.

DA-Signoff-Differentiation-1 returns these grades and limits:

Claim	Grade and allowed use
The common and domain-specific instructions differ in the named required checks and stops.	Source-supported by S3/E1/M1; Method identity and variant qualification remain the separate case premises.
Separate rules were approved by 2 April and enacted in the two named April Work occurrences.	Source-supported by A14 and WE7/WM4. This is a documented bound, not an exact origin date or evidence of universal uptake.
The March exceptions were a stated reason for proposing the variants.	Source-supported by P1; that stated reason is not a causal-effect result.
The split was merely a May renaming with no earlier semantic difference.	Contradicted by the April instructions and Work records.
Demand caused selection over common signoff with escalation, or the carrier caused later retention.	Missing causal support; retain the distinct hypotheses and missing deliberation/continuation evidence.

The receiving source-reconciliation decision can now keep E1 and M1 as different variant descriptions, keep S3 as the common predecessor account, and link F2 as their later form without dating Method differentiation from form creation. It leaves the exact first occurrence and reasons for selection unresolved. No C.28 result is created because no causal reliance is taken. Reopen the relevant date, relation, or link claim if an earlier instruction, the missing deliberation record, or contrary Work evidence appears; do not choose today’s architecture from this history alone.

RMP.11:5.2 - Causal-Use Branch and Non-Causal C.32.MWA.PP Consumption

The EC-417 release scenario records a cadence/evidence mismatch near several reopenings. Four result identities remain separate:

DA-EC417-CadenceDifferentiation-1 is a partial descriptive account: the retained co-occurrences do not fill the dates, variants, and selection history needed for a complete differentiation account;
CUR-EC417-CadenceEffect-1 is the causal-use support result;
DC-EC417-CadenceMismatch-1 is a non-causal timing/co-occurrence hypothesis; and
AD-EC417-B2-Trial-1 is the receiving C.32.MWA.PP architecture decision.

The causal-use question CUQ-EC417-CadenceEffect-1 asks: “Would entering provisional-evidence reconciliation at D-21, rather than waiting for signed evidence at D-8, reduce mismatch-related reopenings in EC-417-like releases for this team, supplier, and change class?” Its causalUseClaimKind is causalEffectClaim; its target rung is interventionalActionRung.

The actual support-component refs are evidencePathRefs=[EP-EC417-BundleSequence-20, EP-EC417-RigAvailability-20, EP-EC417-QuarterlyCadence-2] and empiricalDataRegimeRefs=[EDR-EC417-NaturalReleaseHistory]. No identification or estimate result exists.

Common threat screen CTS-EC417-CadenceEffect-1 records:

Threat-screen field	Result
`causalUseQuestionRef`	`CUQ-EC417-CadenceEffect-1`
`interventionWellDefinedOrConsistency`	`liveThreat`
`temporalOrdering`	`clear`
`exchangeabilityOrConfounding`	`liveThreat`
`positivityOrOverlap`	`liveThreat`
`interferenceOrSpillover`	`liveThreat`
`selectionCensoringOrMissingness`	`liveThreat`
`measurementErrorOrConstructShift`	`liveThreat`
`transportToTarget`	`liveThreat`
`routedThreatRefs`	`[]`
`resultingSupportBoundary`	`unsupported`

Every live threat lowers the causal-support verdict because no specialist result closes it. The C.28 verdict is unsupported. supportedUse: no interventional causal reliance; use this result only as a stop against that reliance. unsupportedUse: claiming that cadence mismatch caused the reopenings or that B2 will reduce them. The evidence window is the named twenty releases and two earlier cases for this team, supplier, and change class. Reopen only on a governed comparison or replayable identification or bound that varies reconciliation timing while rig, approver capacity, outcome definition, and evidence access are controlled or explicitly modeled.

DC-EC417-CadenceMismatch-1 separately retains only the observed timing/co-occurrence as a trial hypothesis and design constraint. AD-EC417-B2-Trial-1 consumes that non-causal result, capacity, confidentiality, authority, and reversibility to select only a bounded B2 trial. SafetyReviewer-17 performs W-SafetyEvidenceDecision-17 under ASG-SafetyReview-17 and AUTH-SafetyEvidence-17 for the evidence-condition result; ReleaseDecider-17 separately performs W-ReleaseDecision-17 under ASG-ReleaseDecision-17 and AUTH-ReleaseDecision-17 for branch entry and release disposition. The architecture decision does not cite CUR-EC417-CadenceEffect-1 as positive causal evidence.

RMP.11:5.3 - A Dated Proposal Has More Weight than Repeated Promotion

Consider a constructed extension of the signoff history. Method and variant identification remain the independent premises of :5.1. The next historical inquiry must choose between H-response, in which repeated domain-specific omissions contributed to selecting separate rules, and H-tool-only, in which a provider’s form package induced the split and the omissions were invoked afterwards. Tool support may also have helped retain a split selected for other reasons, so these are not the only possible contributions.

A newly recovered internal decision memo compares separate rules with common escalation, rejects escalation because the named specialist would not be available, and recommends separation. Its authenticated March date precedes the earliest presently documented provider contact. This combination is more expected under H-response than H-tool-only: it records a problem-specific choice and reason before the currently known tool opportunity. The weight depends on the memo’s authenticity, whether it recorded an actual decision rather than advocacy, and whether earlier provider influence remains unrecorded.

A budget approval refers to that memo. It supplies evidence that resources were authorized, but its repeated rationale is not a second independent explanation. A later brochure and three interviews all repeat the same launch narrative; they show circulation of that story, not four independent confirmations. No trial of common escalation has been found, but trial records were not routinely preserved. That absence adds little against an account in which such a trial occurred.

The investigator therefore gives H-response greater relative support than H-tool-only under the stated evidence and assumptions, while retaining possible provider influence and a mixed account. This selects a focused search for earlier contact and actual decision records. It supplies neither a numerical probability nor a claim about the effect of imposing separate rules today.

Suppose the next archive yields an authenticated February provider proposal. The earlier temporal advantage disappears. The investigator reassesses the memo and budget in light of that contact, and compares a mixed account with tool-only influence and response to omissions. The memo still records its stated reason; it no longer independently shows that this reason preceded any known provider opportunity. Revised historical attribution returns to a present repertoire or decision only where that result relied on the attribution. Existing direct evidence that a local variant works within its qualified scope remains available.

RMP.11:6 - Bias-Annotation

Recurring bias	Likely drift	Repair
chronology bias	Earlier and later events become cause and effect.	Return the descriptive sequence first and open C.28 only for a named causal use.
survival bias	Current variants hide abandoned or excluded alternatives.	Seek abandoned variants, negative cases, and source gaps before grading links.
fashion-metric bias	publication count, search frequency, or institutional promotion becomes adoption or effectiveness evidence.	Treat it as a possible diffusion indicator with scope and rivals, not a causal result.
family-label bias	A local grouping becomes a professional family explanandum.	Require independent family membership or explain named Methods and relations instead.
history-authority bias	A differentiation account chooses today’s architecture.	Keep current decision criteria, authority, causal boundary, and trial stops in the downstream pattern.

RMP.11:7 - Conformance Checklist

The explanandum is an identified Method, established family inclusion, or named Method relation change.
The receiving decision and exact causal use or descriptive stop are explicit.
The dated sequence precedes explanation and keeps source editions, Work evidence, variants, disappearances, and decision points recoverable.
Every link claim names pressure/change, affected actor or carrier, response/variation, selection/retention, and resulting Method or relation change.
Serious rivals and diagnostic observations are derived before convenient collection.
Diagnostic weight compares the actual rivals under stated prior evidence, source-generation conditions and dependencies; source repetition is not independent support.
Differentiation claims carry grades and limits, and changed accounts prompt reconsideration of the relevant earlier evidence.
Any causal-use result contains the complete current C.28 identity, support, threat, verdict, use, window, and reopen fields.
Raw observations and source names are not presented as specialist support results.
Differentiation, causal support, non-causal hypothesis, and downstream architecture decision remain separate.
An unsupported or undecided result supplies no positive causal premise.

RMP.11:8 - Common Anti-Patterns and How to Avoid Them

Anti-pattern	Repair
“The pressure came first, so it caused the Method change.”	State a link claim, rivals, diagnostic observations, and the C.28 result required by the receiving use.
“The timeline is the explanation.”	Treat it as an index and test explicit links and rivals.
“Process tracing proves the causal effect.”	Use it for within-case diagnosis; apply C.28 for the named causal use and lower the verdict when identification, a bound, or estimation required for that use is missing.
“Unsupported means the history is useless.”	Retain descriptive sequence and a separately named non-causal hypothesis within their use boundaries.
“The history tells us what to implement.”	Send bounded results to a separate authorized architecture decision.

RMP.11:9 - Consequences

Professional Method history becomes decision-usable without being made causally stronger than its evidence. Forgotten variants and rivals remain available, while current architecture choices can distinguish causal reliance from a reversible probe based on non-causal observations.

The cost is explicit result separation and threat screening. Some elegant origin stories will end as descriptive accounts, and some causal questions will remain unsupported until a stronger design or bound exists.

RMP.11:10 - Rationale

Differentiation history, causal support, and present choice have different subjects and truth conditions. A dated sequence can be accurate while its causal explanation is unsupported; an unsupported causal claim can coexist with a useful non-causal design constraint; and neither supplies present authority.

The pattern connects questions about variant generation, transmission, fashion, retention and loss. Evolutionary analogies and search-frequency examples can suggest hypotheses and rivals; their causal use depends on diagnostic observations. Process-tracing sources strengthen that inquiry. C.28 states what support a causal use needs, keeping a plausible narrative separate from identification or estimation.

RMP.11:11 - SoTA-Echoing

Source	Retained contribution	Use boundary
Stacey et al., Methods as a form of engineering knowledge	Attention to Method knowledge, variation, loss, and engineering context.	Conceptual history supplies no project causal verdict or current architecture authority.
Collier, Understanding Process Tracing, with 2019 addendum	The 2011 foundation supplies within-case sequencing and diagnostic observations; the addendum develops relations among rivals and moves beyond rigid necessity/sufficiency test classes.	This historical foundation is supplemented by the later inferential treatment below; it does not replace FPF causal-use assessment.
Fairfield and Charman, Explicit Bayesian Analysis for Process Tracing and A Dialogue with the Data	Compare specified alternatives, condition on prior evidence, retain source dependence, and reconsider existing evidence when hypotheses develop.	These foundations discipline qualitative comparison; numerical probabilities require warranted assumptions. They supply no complete causal-use verdict or architecture decision.
Fully specified Bayes factors for hypothesis testing and sensitivity analysis in process tracing, v2, 2026 preprint	Model-based evidence weighting and sensitivity analysis sharpen the question of how a conclusion depends on coding, selection and weight.	Its binomial observation model and one-rival comparisons require substantive justification; clustered source dependence limits the claimed guarantees. This optional numerical development is not a default historical protocol.
Mahoney, The Logic of Process Tracing Tests	Weak, necessary-condition, and strongly discriminating test logic.	Grades observations; it does not make the whole chain certain.
Current FPF `C.28`	Causal-use question, rung, support components, common threat screen, verdict, use boundary, limits, and reopen.	C.28 states evidence support only; downstream choice and authority remain separate.

Reopen when a current source changes the diagnostic logic or causal-use rules, when a representative history cannot keep differentiation and causal support separate, or when a downstream use repeatedly needs a missing result field.

RMP.11:12 - Relations

RMP.4 supplies inspectable source relations and limits; RMP.7 supplies warranted corpus contributions without multiplying dependent reports. RMP.5 can examine the historical claim under its Method, and RMP.8 returns an affected knowledge claim to its actual receiver.
A.3.1 governs Method identity; G.5 governs established family or selector claims; a local grouping label supplies neither.
A.10 governs source and evidence paths; A.10.1 helps locate actual dependencies when a premise changes; C.27 governs temporal-claim adequacy; C.28 and C.28.CM govern causal-use support and construction of causal inquiry.
C.39.RP may retain variants and lineages without inheriting causal truth.
C.32.MWA.PP may consume the differentiation account, causal-use verdict, or separately named non-causal observation only within each result’s use boundary and keeps its own authority.
Cultural continuation and professional-history uses remain separate from one current project architecture decision.