THE HEARING: SECOND-ROUND REVIEW RESPONSE REGISTER

All original reviews are preserved without editing. This register consolidates their recommendations and states what was implemented, retained, deferred or still needs verification. Reviews are AI simulations; content-only readings are not usability trials.

R1 TECHNICAL/COPY EDITOR

- 1 Implemented: action/rule/advantage scope and explicit missing-link instruction; no need to abandon a legitimate support hypothesis.
- 2 Implemented: affected option and inferred-baseline prompt replace unconditional before/after.
- 3 Implemented: complication and claim-specific limit stated before drafting; shared-paragraph promise corrected.
- 4 Implemented: competing account concerns the same observation/link; complementary later steps remain legitimate complications.
- 5 Implemented: one shared response pair across both questions; question reflection is not another essay.
- 6 Implemented: neutral comparison labels and named event/period; no supplied assumption of different outcomes.
- 7 Implemented: local card/table/source ID key and shared-source distinction.
- 8 Implemented: minimum source-opening check labelled honestly; human sign-off includes used records in both cases.
- 9 Implemented: precise IPU summary and mixed legal-announcement badges.
- 10 Implemented: incumbent covers person, party and coalition; glossary entry added.
- 11 Implemented: people-versus-incidents source caveat rewritten without an unexplained omitted figure.

R2 EDUCATION SPECIALIST

- 1 Implemented acronym expansions and a student-visible fictional reasoning model; a new identification exercise is deferred to avoid adding another phase.
- 2 Implemented bounded-step guidance and an action/rule/advantage scope; unsupported voter effects remain explicit in source notes.
- 3 Implemented student-facing human criteria, observed/inferred distinction and claim-specific notes. Optional source attachments remain optional; a person judges the warrant.
- 4 Implemented distinct-observation criterion and shared-source warning; both supporting IDs remain required.
- 5 Implemented source limits tied to the particular inference, while retaining exact-total audit details.
- 6 Implemented a competing account of the same observation, with compatible later-step work distinguished.
- 7 Implemented contrasting expected-pattern cue and an invented example; no automatic causal score.
- 8 Implemented event/period and neutral comparison prompts. Additional scratch cells deferred; no new essay.
- 9 Implemented consistent 7/9/7/12/20-minute phase model across game/help/guide. Timing remains unmeasured; early evidence/extended completion remain accommodation options.
- 10 Implemented a brief private pair-thinking turn in the guide. Individual assessment still requires separate ownership, not two credits for one shared record.
- 11 Implemented compact human-quality criteria available beside drafting; participation remains distinct from mastery.

R3 ELECTORAL-AUTHORITARIANISM SCHOLAR

- 1 Implemented: Serbia required option starter names pending Constitutional Court route; non-turnover is not the option.
- 2 Implemented: organization/backing opening names campaigning and coordination, without claiming seat totals measure mobilization.
- 3 Implemented: no forced baseline; RU2 explicitly labelled as 2023 background through its card-specific note.
- 4 Implemented: explanations may work together, and unestablished links must be named; reported backing retained.
- 5 Implemented: RU1 survey does not adjudicate registration grounds or compare applicants.
- 6 Implemented: RS2 continued contestation is qualified by the same observers' incumbent-advantage/pressure findings.
- 7 Implemented: content, reach, exposure and persuasion distinguished on RS3.
- 8 Implemented: allocation does not establish unfair competition; exact decomposition not invented. Optional new calculation exercise deferred.
- 9 Implemented: comparable event/unit/period and a limiting difference; no controlled causal-effect claim.
- 10 Implemented: expected patterns under accounts that may both contribute; source limits and comparison retained.

R4 COMPARATIVE-METHODS SCHOLAR

- 1 Implemented: main question/opening stay visible during comparison; event/period and failed-analogy boundary explicit.
- 2 Implemented a stricter competing-account prompt. Complementary later steps remain allowed as complications; they alone do not satisfy the rival task.
- 3 Implemented: MW3 direct-judgment/provenance caveat and MW5 implementation limit precede the full document caveats.
- 4 Implemented: MW6 headline says higher reported share. Original qualified body and numerical cautions retained.

R5 EXPERT UNDERGRADUATE

- Scope and legitimate route: action/rule/advantage language and explicit inferred-link boundary added; plausible mechanism is not proof of vote effects.
- Provenance: MY4 badge/headline and RU1 source-chain note clarified.
- Complementary accounts: both may contribute; a rival must address the same observation/link and expected patterns can differ without mutual exclusivity.
- Revision: current versus earlier wording and a paragraph-opening action clarified; no automatic prose synchronization.
- Comparison: event/period, process and limiting difference made explicit; 30-45-word range retained.
- Adversarial weak completion: preserved structural checks, strengthened human-review criteria and changed completed status to say reasoning not assessed. Semantic auto-grading rejected.
- Verification limits: substantive replacement after the weak control test and tab cleanup were unavailable in that review. No substantive final-completion claim is made for R5.

R6 UX/ACCESSIBILITY

- 1 Implemented and browser-sampled: aria-pressed exposes decision state.
- 2 Implemented and browser-sampled: contextual card/source/question/missing-item accessible names; required-state labels clearer.
- 3 Implemented: earlier narrowing and actual current field labelled separately; no stale Now banner after later replacement. Event regressions exercise the sequence; full repeated-change browser path remains unverified.
- 4 Implemented: first actual missing requirement, preparation/refresh, explicit return route and accurate completed guidance.
- 5 Implemented: paragraph-location promise corrected and comparison offers a main-paragraph review control.
- Retest addition: found an older saved completion conflicting with current missing items. Fixed current UI/report status and preserved earlier timestamp; browser recheck and import regressions pass.
- Remaining layout/access checks: focused one-question layout, physical phone, full keyboard-only path and screen-reader audit deferred to appropriate users/devices.

R7 LAW/DEMOCRATIZATION SCHOLAR

- 1 Implemented: MW-D also mapped to MW3; direct-judgment retrieval caveat shown there.
- 2 Implemented: original MW3 wording rendered in separate paragraphs, with note separating finding/remedy/interpretation and denying unsupported personal culpability inference.
- 3 Implemented: reported 2020 winner satisfies both majority and plurality; strategic rule effects need further evidence.
- 4 Implemented: MW4 ticket is not tracked individual voter transfer.
- 5 Implemented: retrospective implementation does not independently verify all instances or explain compliance motives.
- 6 Implemented: MY1 conversion does not establish biased boundaries or explain geographical vote distribution.
- 7 Implemented: MY4 headline and mixed evidence label retain allegations rather than imply a legal finding.
- 8 Implemented: MY6 supports later reform capacity or complicates turnover claims; cannot cause earlier victory.
- 9 Implemented: explicit comparison event/year and neutral labels. No claim to have inspected original Malawi judgments.

R8 ORDINARY NOVICE STUDENT

- 1 Implemented entry-to-case shortcut, incumbent/seat/list/source vocabulary support and a concrete action/option task. A sensational recruitment hook was not required.
- 2 Reduced redundant directions and exposed the immediate missing action. Full one-question focus layout is deferred; all evidence stays reachable and no analytical phase removed.
- 3 Implemented and browser-checked: no empty-paragraph revision warning; action-specific labels and earlier/current wording distinguish actual changes.
- 4 Implemented: local ID key, previously-opened record badge, claim-specific limits and shared-source warning. No copyrighted original excerpts added without purpose/verification.
- 5 Implemented: truthful complete state, teacher hand-in boundary, readable-report/resume-file labels and optional work below primary saving.

R9 WORKING GEN Z STUDENT

- 1 Implemented quicker case entry and concrete task language; full contract retained.
- 2 Added local ID/incumbent/plurality and case acronym help; existing searchable definitions retained. A pervasive tooltip layer is deferred pending device/access testing.
- 3 Student-visible fictional guide distinguishes source limit, rival and missing link; no historical answer supplied.
- 4 Positive observations remain in previews/full cards with claim-specific limits beside the relevant source record; no implication that all evidence is unusable.
- 5 Implemented affected-option prompt without invented baseline.
- 6 Implemented worked expected-pattern example and short historical request.
- 7 Main question/opening/chain and case actor key remain available in comparison; equal advantage strength is not assumed.
- 8 Human-quality review is separated from inclusion checks; fixed word ranges and citations retained.

R10 EDUCATIONAL CREATOR LENS

- 1 One-card preview route deferred: it would add another route needing review and timing. The full evidence activity is not presented as a 40-second game.
- 2 Implemented a direct entry-to-case action; the three-space attention constraint and open evidence stay visible.
- 3 Implemented progressive full-task disclosure in casual mode; all requirements remain before finishing.
- 4 Implemented precise producer/source badges and claim-specific top notes; exact totals, provenance and links retained.
- 5 State-aware next actions and action-specific decision labels implemented. A new focused question layout is deferred rather than risk introducing navigation regressions.
- The accurate demo outline is preserved in the full review. No social posting, engagement forecast or real influencer endorsement.

R11 E-LEARNING/COURSE DESIGN

- 01 Implemented student-visible human criteria, fictional reasoning/discrimination and comparison example; two observations remain distinct.
- 02 Implemented Malawi umbrella question covering reopening and later winner, while asking the learner to choose one step.
- 03 Implemented Serbia rule/allocation starters and inferred-baseline prompt. A new allocation lesson is deferred; exact 129-seat decomposition is not invented.
- 04 Implemented observation-specific source limits while retaining full document details.
- 05 Implemented honest minimal-access label, shared-source relation and used-source human confirmation. Clicking is not treated as reading.
- 06 Implemented action-specific apply labels, current/earlier wording and paragraph handoff.
- 07 Implemented required/optional field labels and semantics; evidence-role menu is a best-fit optional use, not a factual classification.
- 08 Implemented contextual destinations and decision-state semantics. Existing pair cues retained; full assistive-technology certification remains unverified.
- 09 Implemented local-completion/hand-in boundary, clear report vs resume labels and conditional closing briefing.
- 10 Participation policy remains teacher-controlled and separate from mastery; no automatic score, confidence grade or speed reward.

R12 TECHNICAL PRECISION

- 1 Implemented complication criterion at first drafting and in opening contract.
- 2 Implemented truthful minimum source-opening label and full used-source human review.
- 3 Implemented affected option without an unsupported baseline.
- 4 Implemented corrected shared-paragraph promise.
- 5 Implemented one shared rival/request pair across the two questions.
- 6 Implemented local ID key and valid case-ID decision validation; card/table/source IDs can support a reason, but paragraph gates still require the selected card IDs.
- 7 Implemented precise practice answer-choice/decision inclusion label, without implying correctness.
- 8 Implemented action-specific hearing recording/mutation labels and human paragraph review.
- 9 Implemented truthful finish/report/resume/handoff language; readable preview retains current and earlier attempts. Actual completed browser downloads remain unverified.

ADDITIONAL ENGINEERING FIX FOUND DURING INFORMED RETEST
- A stale filed timestamp no longer overrides missing current requirements. Earlier completion is retained as previousFiledAt; UI and report consistently say draft until current checks pass. Valid imports remain complete.

VERIFICATION
- 198 new and 128 existing assertions pass (326 total).
- All 24 historical card texts and all active/legacy answer keys preserved. Main 60-85 and comparison 30-45 word ranges and all analytical requirements remain.
- Informed browser check passed sampled labels, pressed states, empty-paragraph guidance, comparison copy and the saved-completion fix. No full new replay or human learning/timing evidence.
