Research record

HAG decision log

Published: 5 August 2026

Sections of this document

Pathway status
Candidate heuristic policy

This log records consequential HAG-guided decisions across the xResearch publication program. It distinguishes retrieval from application and preserves human authority over interpretation and model updates.

Decision ID	Scope	Model	Retrieved heuristic	Disposition	Decision and reason
HAG-001	Program evidence standard	Personal HAG	Research brief evidence search	Applied	Exact-mechanism searches, original-source checks, reported effect sizes, conditions, and criticism govern evidence use.
HAG-002	Program claim standard	Personal HAG	Evidence brief prioritization	Applied	A null or unsupported headline is a valid result; prose may not manufacture a defensible claim.
HAG-003	Pilot workflow	Personal HAG	Heuristic-to-Experiment Feedback Loop	Applied	Editorial rules are tested on representative pairs before portfolio-wide rollout.
HAG-004	Pilot workflow	Personal HAG	Rapid Small-Batch Market Validation	Adapted	Retained the small-batch pilot logic; rejected market, sales, and unit-economics criteria as inapplicable to scholarship.
HAG-005	Public framing	Writing Model	Frame Gaps Through Field Self-Assessment	Applied	Public papers derive gaps from the field's definitions, evidence, and self-criticism rather than attacking the field or defending the company.
HAG-006	Titles and openings	Writing Model	Foreground the paper's reframing move	Applied	Titles, openings, and early turns must expose the intellectual move rather than merely announce a topic.
HAG-007	Paper/essay pairing	Writing Model	Preserve Core, Reframe for Audience	Applied	The analytic result stays fixed while load order, context, examples, and consequences change for the essay.
HAG-008	Reuse	Writing Model	Borrow Transparently, Reuse Explicitly	Applied	Borrowed structures retain their source distinctions; textual and intellectual reuse is disclosed where it occurs.
HAG-009	Displays	Personal HAG	Make visuals claim-driven and explicit	Applied	Display form follows the claim; thresholds and transformations that shape interpretation appear in the caption or note.
HAG-010	Evidence-store tables	Personal HAG	Table as specification or evidence store	Applied	Verification tables remain complete and locatable without being narrated as headline findings.
HAG-011	Agentic disagreements	Personal HAG	Split persistent two-camp terms into testable objects	Applied	Agent disagreement prompts definition and discriminating evidence, not majority voting.
HAG-012	Editorial ownership	Writing Model	Align Framing and Form to Thesis	Applied	Thesis, architecture, and final synthesis remain centralized; specialist agents propose bounded analyses.
HAG-013	Human checkpoints	Writing Model	Post-edit verification note	Applied	Each substantive batch reports what changed, what remained, and the one decision requiring human verification.
HAG-014	Corpus execution order	Personal HAG	Disclose Context at Point of Need	Applied	Dependencies and reused constructs are surfaced where readers need them; inventory and reconciliation precede glossary lock.
HAG-015	Conclusions	Writing Model	Frame Conclusions by Problem and Stakeholder	Applied	Conclusions size the paper against the unfinished problem and end on the affected population or decision-maker appropriate to the contribution.
HAG-016	Series front matter	Personal HAG	Audience-Driven Argument Reframing	Applied	A small common metadata spine orients every reader; papers foreground method and evidence while essays foreground interpretation and consequence.
HAG-017	Series front matter	Personal HAG	Disclose Reuse Inline	Applied	Companion relationships do not imply silent duplication; reused intellectual or textual material is named where readers encounter it.
HAG-018	Series front matter	Writing Model	Context-Aware Textual Recalibration	Applied	Canonical metadata is stable, while document architecture and load order remain genre- and audience-specific.
HAG-019	Portfolio positioning	Writing Model	Frame Contributions Within Larger Program	Applied	Each paper states its bounded contribution and dependency within the larger xResearch program without presenting the program as proof of the paper.
HAG-020	HHITL evidence search	Personal HAG	Anchor Claims, Then Broaden Stakes	Applied	The paper first tests its foundational mechanism against primary evidence; wider implications appear only after that basis is established.
HAG-021	HHITL claim calibration	Personal HAG	Calibrate Claims to Evidentiary Reach	Applied	Naming novelty, component prior art, configuration distinctiveness, observed use, and demonstrated benefit are treated as separate claims with separate evidence burdens.
HAG-022	HHITL opening	Writing Model	Foreground the Reframed Contribution	Applied	The title, abstract, and first few hundred words expose the distinction between human review in a loop and inspectable human judgment shaping AI reasoning.
HAG-023	HHITL definition	Writing Model	Surface and Canonicalize Core Constructs	Applied	HHITL receives one early canonical definition; later sections test its components and boundaries rather than repeatedly redefining it.
HAG-024	HHITL prior art	Writing Model	Contextualize Before Citing	Applied	Each borrowed lineage is introduced as an answer to a defined problem, then traced through origin, adaptation, and present operational use.
HAG-025	HHITL conclusion	Writing Model	Frame Conclusions Around the Larger Problem	Applied	The close names the evaluation and governance program still required and the faculty whose judgment the mechanism is intended to keep consequential.
HAG-026	Pilot selection	Personal HAG	Earned, Bounded Claims	Applied	WP-01 and WP-09 form the smallest batch that tests a mature evidence argument and the central contested construct without implying corpus-wide effects.
HAG-027	Pilot adaptation	Personal HAG	Preserve core, adapt presentation	Applied	The pilot changes load order, apparatus, and audience consequence while freezing the verified evidential core.
HAG-028	Pilot conclusions	Personal HAG	Scale Argument To Problem	Applied	A successful edit still closes on the unresolved educational or validity problem rather than on completion of the document.
HAG-029	Scale-up architecture	Writing Model	Context-Responsive Text Adaptation	Applied	WP-05 follows the pilot as the first architecture test; genre- and layout-specific form may vary while the core result remains fixed.
HAG-030	WP-09 public framing	Personal HAG	Frame By Problem, Not Product	Applied	The revision leads with the educational measurement problem and treats the mechanism as a construct to be validated, not as the source of the paper’s importance.
HAG-031	WP-09 and Essay 16 adaptation	Personal HAG	Preserve core, adapt presentation	Applied	The verified prior-art findings remain fixed while audit language becomes field-facing and the essay loads the practical consequence first.
HAG-032	WP-09 evidence boundary	Personal HAG	Separate Method Claims from Stakes	Applied	Implemented status, construct validity, and possible educational use are kept as separate statements; wider faculty and learner stakes appear after the evidence argument.
HAG-033	WP-09 display	Personal HAG	Make visuals claim-driven and explicit	Applied	The inference ladder distinguishes the observable record, process interpretation, and outcome inference; its caption states that arrows are evidential tests rather than a product dataflow.
HAG-034	WP-09 argumentative turn	Writing Model	Foreground the Reframed Contribution	Applied	The three-layer learning distinction appears in the abstract and opening, before the literature is reviewed.
HAG-035	WP-09 and Essay 16 conclusions	Writing Model	Frame Conclusions Around the Larger Problem	Applied	Both pieces end on the unresolved validation problem and what faculty and learners would need from the evidence, not on document completion.
HAG-036	WP-01 preservation control	Personal HAG	Preserve core, adapt presentation	Applied	The revised pair preserves every verified study estimate, condition and evidential judgment; intervention was limited to public metadata, construct precision, signposting, display notes and conclusion architecture.
HAG-037	WP-01 and Essay 01 genre relationship	Personal HAG	Audience-Driven Re-sequencing	Applied	The paper retains the review method and full evidence apparatus; the essay keeps the classroom scene and converts the same withdrawal evidence into a faculty-facing decision standard.
HAG-038	WP-01 and Essay 01 display decision	Personal HAG	Present estimates in tables when verification is the reader's task	Applied	The two existing semantic tables remain the paper's displays and receive propositional titles, source status and inference boundaries. A new cycle figure was rejected as duplicative because the essay already carries the sequence in accessible text.
HAG-039	WP-01 learning and intervention claims	Writing Model	Calibrate Claims to Evidence	Applied	Guardrails are reported as preventing harm in one pivotal trial, not as producing a learning gain; independent performance, transfer and durability are explicitly separated.
HAG-040	Essay 01 reuse	Writing Model	Borrow Transparently, Reuse Explicitly	Applied	Front matter identifies the companion relationship and the distinctive assistance-and-withdrawal loop is disclosed at its point of reuse as an untested synthesis rather than a new result.
HAG-041	WP-01 opening summary and Essay 01 definition	Writing Model	One paragraph, one claim	Applied	Repetitive direct answers were compressed into four bounded signposts, and the essay's performance, independent-performance, transfer and durability distinctions were separated into one explicit conceptual unit.
HAG-042	WP-01 and Essay 01 conclusions	Writing Model	Reframe the Conclusion	Applied	The paper ends on the staged empirical standard beyond assisted fluency; the essay preserves its Monday-quiz return and names the multi-site, delayed-transfer evidence still missing.
HAG-043	HHITL canonical definition	Writing Model	Surface and Canonicalize Core Constructs	Applied	One definition of HHITL and one definition of a provenance-bearing expert heuristic will appear in the abstract and opening; later sections test their components rather than redefining them.
HAG-044	HHITL contribution boundary	Personal HAG	Frame gaps from the field's own logic	Applied	The configuration-level residual is derived only after expert systems, CTA, HITL, mixed initiative, cognitive forcing, learner models, and heuristic-inference research are stated at full strength.
HAG-045	HHITL claim ladder	Personal HAG	Separate Method Claims from Stakes	Applied	Naming, component precedent, proposed configuration, implementation, process trace, assisted performance, complementarity, learning, and authorship are treated as different claims with different evidence.
HAG-046	HHITL public framing	Personal HAG	Frame By Problem, Not Product	Applied	The paper opens with the absence of inspectable, accountable human judgment in many AI workflows; the xResearch use case appears only after the field and construct are established.
HAG-047	HHITL evaluation destination	Writing Model	Align Framing and Form to Thesis	Applied	The argument culminates in human-alone, model-alone, generic human+AI, and HHITL comparisons plus direct falsifiers, so the paper's form enacts its claim of testability.
HAG-048	HHITL evidence language	Writing Model	Calibrate Claims to Evidence	Applied	Present wording is limited to a governed design pattern, a candidate process trace, and an untested configuration; no effect, complementarity, learning, authorship, or superiority claim is inferred.
HAG-049	HHITL portfolio role	Writing Model	Frame Contributions Within Larger Program	Applied	HS-WP-2026-10 connects the expertise work in WP-02, the learning-process construct in WP-09, and the library-wide methods note while remaining a standalone field paper.
HAG-050	Essay 09 evidence boundary	Personal HAG	Evidence brief prioritization	Applied	The strongest defensible result is a component-grounded faculty-handoff proposal; the complete relay is stated as untested, and private testimony without documented publication permission was removed.
HAG-051	Essay 09 source use	Personal HAG	Research brief evidence search	Applied	The essay retains only exact-mechanism evidence from the finalized tutoring, oral-assessment and integrity reviews, including reported estimates, conditions and inference limits.
HAG-052	Essay 09 argumentative turn	Writing Model	Foreground the Reframed Contribution	Applied	The opening moves within the first few hundred words from faculty replacement to the distinct question of who governs evidence, withdrawal and instructional response after AI-supported preparation.
HAG-053	Essay 09 adaptation and reuse	Writing Model	Context-Responsive Text Adaptation	Applied	The essay remains a cross-paper faculty-handoff extension rather than a second Essay 01; the withdrawal criterion is reused and disclosed where the handoff question begins.
HAG-054	Essay 09 conclusion	Writing Model	Reframe the Conclusion	Applied	The close returns to the Thursday/Friday scene while positioning the relay inside an unfinished multisite evaluation agenda for faculty and teaching-and-learning leaders.
HAG-055	HHITL Essay 17 genre adaptation	Writing Model	Preserve Core, Reframe for Audience	Applied	The essay retains the paper's prior-art and evidence boundaries while replacing its review sequence with a faculty-facing contrast between approval and consequential judgment.
HAG-056	HHITL Essay 17 opening	Writing Model	Place the argumentative turn early	Applied	The human-presence versus human-judgment distinction and unevaluated status appear within the opening movement, before the literature is explained.
HAG-057	HHITL Essay 17 claim boundary	Writing Model	State the claim at the evidence boundary	Applied	The practical value is framed as a more inspectable question for faculty and research design, not as a demonstrated improvement in learning or decision quality.
HAG-058	HHITL Essay 17 construct	Writing Model	Foreground and Canonize Core Constructs	Applied	HHITL receives one stable definition near the start; later sections unpack the verbs and evidentiary distinctions rather than redefining the term.
HAG-059	HHITL Essay 17 reuse	Writing Model	Make Reuse and Dependency Explicit	Applied	The essay identifies its derivation from HS-WP-2026-10 in front matter and discloses the adapted comparison figure at the point of reuse.
HAG-060	HHITL Essay 17 conclusion	Writing Model	Frame Conclusions by Problem and Stakeholder	Applied	The ending identifies the evidence that faculty and teaching-and-learning leaders would need before adopting the category as more than a useful design question.
HAG-061	Methods note reuse disclosure	Writing Model	Borrow Transparently, Reuse Explicitly	Applied	The note identifies the human sources of the heuristic model, names transferred constructs in their source traditions, and discloses reuse where it shapes the publication workflow.
HAG-062	Methods note agency language	Writing Model	Specify who is doing what	Applied	Active role statements distinguish human decisions from HAG retrieval, AI-assisted research and drafting, external evidence, automated checks, and independent review.
HAG-063	Methods note argumentative move	Writing Model	Foreground the Paper's Core Move	Applied	The note opens as an accountability map rather than a technology inventory: process visibility can support attribution without proving authorship, quality, originality, or learning.
HAG-064	Methods note audience and boundary	Writing Model	Contextualize Without Altering Core	Applied	The same verified workflow is translated into a public disclosure while private source texts and proprietary implementation details remain outside the document.
HAG-065	WP-03 field framing	Personal HAG	Frame Gaps Through Field Self-Logic	Applied	The revision begins from educational measurement's own distinction between an elicited performance, an interpretable score, and a learning outcome; oral assessment is stated at full strength before its evidentiary conditions are derived.
HAG-066	WP-03 thesis architecture	Writing Model	Foreground the paper's reframing move	Applied	The phrase "reasoning, not recall" is clarified near the opening as a task-design contrast rather than a cognitive dichotomy, and the construct definitions precede empirical results.
HAG-067	Essays 03 and 11 genre relationship	Personal HAG	Audience-Driven Re-sequencing	Applied	Essay 03 loads the instructional problem first; Essay 11 loads the measurement chain first. They preserve the same evidence without becoming paraphrases of one another.
HAG-068	WP-03 family display decision	Personal HAG	Method-First, Stakes-Last Framing	Applied	A semantic evidence-map table retains study-specific estimands, and a conceptual flow separates prompt stages. A cross-study effect chart is rejected because the outcomes and metrics are not commensurable.
HAG-069	WP-03 family claim boundary	Personal HAG	Earned, Bounded Claims	Applied	Oral tasks are described as eliciting construct-relevant performances under defined conditions; neither speech, higher marks, self-explanation, nor scorer agreement is treated as proof of learning or validity.
HAG-070	WP-03 family conclusions	Writing Model	Frame Conclusions Around the Larger Problem	Applied	Each close states what is supported, converts remaining uncertainty into discriminating tests, and ends on the faculty measurement problem rather than a permission register or document recap.
HAG-071	WP-02 family evidence search	Personal HAG	Research brief evidence search	Applied	The revision preserves study-specific estimates and conditions, checks them against the complete source ledger, and treats missing intervals or effect sizes as missing rather than reconstructing them.
HAG-072	WP-02 expert-claim boundary	Personal HAG	Distinguish enacted from espoused	Applied	Elicited expert accounts are framed as provenance-bearing hypotheses; observation, held-out decisions and learner traces are separate tests rather than confirmations inferred from an articulate account.
HAG-073	WP-02 family field framing	Writing Model	Frame Gaps Through Field Self-Assessment	Applied	The papers state cognitive-task-analysis and expertise research at full strength, then derive reliability, criterion-validity, transfer and durability gaps from those fields’ own methods and limitations.
HAG-074	Essays 02 and 10 genre adaptation	Writing Model	Preserve Core, Recast Context	Applied	Essay 02 loads the faculty classroom scene; Essay 10 loads the uploaded-mind metaphor. Both preserve the working papers’ evidence boundaries while serving different public arguments.
HAG-075	WP-02 family display decision	Personal HAG	Structured Tradeoff & Qualification Framework	Applied	A mobile conceptual sequence, an elicitation-method matrix and an interpretation-burden table expose comparison criteria and non-claims. The recorded reasoning pathway to evidence integration and bias-aware synthesis reinforced source comparability and qualification; a heterogeneous effect chart and any product flow were rejected.
HAG-076	WP-02 family conclusion	Writing Model	Forward-Looking Generalized Conclusions	Applied	The paper ends with held-out decision, independent-application, transfer and durability tests; the essays end where evidence stops rather than converting a design commitment into an effect claim.
HAG-077	WP-04 family evidence boundaries	Personal HAG	Robust-evidence appraisal	Applied	Product quality, contribution, individual learning and judgment governance retain separate units, evidence standards and non-claims; proxy outcomes and cross-level inferences are qualified at the study where they arise.
HAG-078	WP-04 thesis architecture	Writing Model	Foreground paper’s reframing move	Applied	The paper now opens with its central measurement move: one group grade is commonly asked to answer four different questions, and the literature is organized around that distinction.
HAG-079	Essays 04 and 12 genre adaptation	Writing Model	Preserve Core, Reframe for Audience	Applied	Essay 04 loads the faculty problem of attributable learning; Essay 12 loads the governance problem created by AI-supported group work. Both preserve the paper’s evidence boundaries without becoming summaries of it or of each other.
HAG-080	WP-04 family display decision	Personal HAG	Weighted Bias-Aware Evidence Synthesis	Adapted	Study estimates remain attached to their original designs and outcomes; a quantitative cross-study chart was rejected. GROUP-01 is rendered as a four-row semantic argument matrix in the paper and Essay 12, with source, purpose and non-validation notes.
HAG-081	WP-04 family conclusions	Writing Model	Frame Conclusions by Problem and Stakeholder	Applied	The paper distinguishes faculty assessment design from institutional evidence governance; both essays close by naming the empirical boundary rather than granting or withholding permission to claim an effect.
HAG-082	WP-05 family headline claim	Personal HAG	Evidence brief prioritization	Applied	The positive skills/application result and the negative but nonrobust knowledge result remain adjacent; the absence of a direct AI-cast or topology effect is reported as a valid finding rather than repaired into a supportive headline.
HAG-083	WP-05 mechanism transfer	Personal HAG	Triangulate evidence and transfer mechanistic inference	Applied	Evidence from worked examples, comparison, calibrated assistance and role rehearsal supports bounded mechanism hypotheses; it is not transferred as proof of a case label, interface, cast or arrangement.
HAG-084	WP-05 thesis architecture	Writing Model	Foreground the paper’s reframing move	Applied	The abstract and opening distinguish the work learners perform inside cases from the case label or presentation, organizing the review around mechanism, outcome and condition.
HAG-085	Essays 05 and 13 genre adaptation	Writing Model	Preserve Core, Reframe for Audience	Applied	Essay 05 begins with an engaging discussion that fails to transfer; Essay 13 begins with the allocation of reasoning responsibility. Both preserve the paper’s evidence without becoming summaries or effect claims.
HAG-086	WP-05 family display decision	Personal HAG	Integrate and Prioritize Diverse Evidence	Applied	CASE-01 keeps mechanism, evidence, condition and non-claim together; CASE-02 uses generic role and information arrangements only as opportunities. A cross-study effect chart and any comparative topology encoding were rejected because outcomes are not commensurable and topology effects have not been estimated.
HAG-087	WP-05 family claim calibration	Writing Model	Calibrate Claims to Evidence	Applied	Case labels, vividness, expert embodiment, cast size and named topologies are not described as demonstrated causes; application, knowledge, transfer and confidence retain separate evidence states.
HAG-088	WP-05 family conclusions	Writing Model	Forward-Looking Conclusions	Applied	The paper closes on component comparisons and delayed unaided transfer; both essays return to their opening case or topology distinction and end on what a comparative study must establish.
HAG-089	WP-06 field framing	Personal HAG	Frame gaps from the field's own logic	Applied	The revision begins from educational measurement's distinction among artifact quality, provenance and independent competence; detector, oral-assessment and redesign evidence are stated at full strength before their inferential limits are drawn.
HAG-090	WP-06 display decision	Personal HAG	Make visuals claim-driven and explicit	Applied	INT-01 remains a narrow semantic inference table; INT-02 stores current accreditor requirements and boundaries without implying effectiveness. A detector league table and universal threshold are rejected because the studies do not share a stable population, task, tool state or estimand.
HAG-091	WP-06 paper architecture	Writing Model	Foreground the paper's reframing move	Applied	The paper makes the three-inference distinction before reviewing detection and organizes each evidence block as claim, study-specific finding, interpretation and boundary.
HAG-092	Essay 06 genre architecture	Writing Model	One paragraph, one claim	Applied	The essay retains its public faculty-facing arc while separating detector fallibility, fairness, competence evidence and misconduct procedure into distinct argumentative units rather than compressing the working paper.
HAG-093	WP-06 family conclusions	Writing Model	Transform-and-Extend Conclusions	Applied	The paper and essay do not merely recap findings: they convert the inference distinction into a forward research and governance agenda and end on the unresolved evidence needed by faculty and teaching-and-learning leaders.
HAG-094	WP-08 field framing	Personal HAG	Frame Gaps Through Field Logic	Applied	Phase A states agency, delegation, contribution, trace analysis and authority-classification precedents at full strength, then locates the remaining question in episode-level attribution of consequential decision rights.
HAG-095	WP-08B adverse-result spine	Personal HAG	Foreground the Finding Only the Reframing Could Produce	Applied	The missing independent reference is the organizing result, not a caveat; prototype execution and secondary output behavior cannot substitute for the preregistered primary attribution endpoint.
HAG-096	WP-08 paper and essay architecture	Personal HAG	Audience-Driven Re-sequencing	Applied	The papers carry the construct review, preregistered phase chain and validation burden; the essay loads the faculty assessment problem while preserving the same prior-art and non-estimable evidence state.
HAG-097	WP-08 public display decision	Personal HAG	Make visuals claim-driven and explicit	Applied	GOV-01 separates governance from operative contribution in a conceptual matrix; VAL-01 renders ordered evidentiary burdens. Prototype panels and participant-derived values are not displayed because they could be mistaken for construct evidence.
HAG-098	WP-08 family conclusions	Personal HAG	Size the Contribution Against the Problem	Applied	Phase A ends with the unvalidated measurement proposal, Phase B1 with the missing-reference study that must come next, and the essay with the evidence needed before faculty can infer who governed the work.
HAG-099	WP-07 evidence verification	Personal HAG	Research brief evidence search	Applied	Every central study and 2026 publication-status claim was checked against a primary publisher, proceedings or preprint record; effect estimates remain attached to their original task, model, metric and conditions.
HAG-100	WP-07 headline boundary	Personal HAG	Evidence brief prioritization	Applied	Evidence-first scoring is stated as established prior art, while the absence of a peer-reviewed isolated hard-gate effect is retained as the valid bounded finding rather than repaired into a novelty or efficacy claim.
HAG-101	WP-07 field framing	Writing Model	Frame Gaps Through Field Self-Assessment	Applied	Human norming, cue bottlenecks, rule systems and LLM pipelines are represented at full strength; the remaining gap is derived from their bundled designs and validity limits, not from an external critique of the field.
HAG-102	WP-07 paper and essay architecture	Writing Model	Foreground the Reframed Contribution	Applied	The paper opens with the hidden inference inside a score and makes the evidence-chain reframing visible in the title, abstract and first figure. Essay 07 loads the explanation problem; Essay 14 loads the scoring-protocol problem.
HAG-103	WP-07 claim and conclusion discipline	Writing Model	Calibrate Claims to Evidence	Applied	Auditability is the immediate construct claim; recoverability, agreement, validity and fairness remain separate. The paper and essays close at the controlled-study boundary rather than turning an implemented mechanism into a demonstrated effect.
HAG-104	Portfolio synthesis genre	Writing Model	Preserve Core, Reframe for Audience	Applied	The synthesis retains each paper's fixed evidence status and converts it into faculty, teaching-and-learning, and research decisions without treating the essays as additional confirming studies.
HAG-105	Portfolio synthesis claim boundary	Writing Model	State the claim at the evidence boundary	Applied	Architectural coherence, source coverage, implementation, validity, learning, transfer and product effect remain separate; the synthesis claims a research program and evidence architecture, not demonstrated HeuriSight outcomes.
HAG-106	Portfolio audience consequences	Writing Model	Re-consequence stable findings for each reader	Applied	Stable findings are translated into distinct decisions for faculty, teaching-and-learning leaders, assessment/governance teams and researchers rather than one generic recommendation.
HAG-107	Portfolio display architecture	Writing Model	Frame Contributions Within Larger Program	Applied	A dependency diagram and construct/evidence table locate each paper inside the larger program; arrows indicate intellectual dependency, never effect, validation or product dataflow.
HAG-108	Portfolio evidence synthesis	Writing Model	Surface Hidden Preconditions in Mature Debates	Applied	The cross-paper synthesis identifies the untested preconditions behind familiar claims—independent performance, attributable evidence, valid observation, stronger-solo comparison and transfer—after representing each source field at full strength.
HAG-109	Portfolio and Essay 15 conclusion	Writing Model	Forward-Looking Conclusions	Applied	Both pieces close on the staged validation program and the decisions its evidence would enable, not on the number of documents, citations, features or completed edits.
HAG-110	Portfolio source reuse	Writing Model	Make Reuse and Dependency Explicit	Applied	Repeated canonical sources are disclosed as bibliographic dependencies at the point of reuse; the overlap matrix is explicitly not evidence of agreement, replication, quality or effect.
HAG-111	Essay 15 argumentative move	Writing Model	Foreground the paper’s reframing move	Applied	The essay now opens with the shift from abundant artifacts to scarce trustworthy evidence and organizes the corpus as one inference chain rather than a paper-by-paper feature catalogue.

Pathway status

The retrieved heuristics above were checked for recorded reasoning-pathway links. Most had no applicable pathway at the time of retrieval. For HAG-075, a recorded pathway connected the tradeoff framework to evidence integration and bias-aware synthesis; those linked heuristics were applied only to source comparability, weighting and qualification. For HAG-077, the recorded precursor emphasized detecting and labeling claims that outrun their observable evidence; it informed the proxy and cross-level inference checks in Family C. For HAG-086, the recorded pathway again connected structured tradeoff, evidence integration and bias-aware synthesis; it governed outcome comparability and the decision not to encode unestimated topology effects. HAG-089 through HAG-103 and HAG-110 through HAG-111 had no recorded pathway links at retrieval. Unrelated downstream branches were not used. No unrecorded reasoning chain is implied. Exact graph relations and internal record identifiers are not part of the public methods disclosure; the template-identifier column is retained in this working copy for audit and is omitted from the published version.

Candidate heuristic policy

New editorial rules discovered during revision remain in program memory until they recur across independent decisions, can be stated as general human-authored rules, survive review, and receive explicit human approval for HAG ingestion.