# Assessor-only manual review of the nine-task expansion This is an asset-authoring review of catalogue 1.1, not an experimental report. No worker outputs, model calls, trials, or empirical scores were generated. The three original public packets and their individual guides remain unchanged. The shared scoring convention is reused: twelve public requirements per task, with Cn mapping to Rn and no extra scoring obligations. The index count of twelve remains an unfrozen development choice. | New task | Distinct reasoning focus | Direct arithmetic/feasibility check | Source sufficiency and scoring boundary | |---|---|---|---| | DEV-SYN-02 | Corrected survey aggregation, subgroup changes, and respondent selection | Peak 90%→80%; off-peak 40%→70%; pooled 110/150→111/150, +0.7 percentage points rounded; response 40%→60% | All counts, invitation totals, and design limits are supplied. Subgroup invitation counts and causal effects remain unknown; neither is a required invented answer. | | DEV-SYN-03 | Separating evidence for date, role, funding, and authorship | 1912 commission precedes 1914 installation and 1961 restoration; donations total 100 crowns, largest share 45.0% | Authorship can remain unresolved, so no unseen artist record is needed. Sources explicitly determine date precedence without resolving all ambiguous wording. | | DEV-SYN-04 | Material balance, unit conversion, denominator choice, and claim review | 1,200−240=960; 1,000−150=850; accepted destinations sum correctly; yields 60%→70% incoming basis; landfill reduction 50/120=41.7% | Rejected destinations and eventual reuse are absent by design. Correct completion is a qualified review, not a fabricated site-wide diversion rate. | | DEV-CON-02 | Blind intake, version cutoffs, role separation, conflicts, and appeals | Local 18:00 at UTC+02 is 16:00 UTC; Q2's repair is one minute late; Q4 appeal elapsed 47h30m fits 48h | Roles, counting conventions, clock offset, and four case facts are supplied. No literary judgment or appeal outcome is secretly required. | | DEV-CON-03 | Segment-specific access, revocation versus expiry, and derivative scope | Day 18 precedes day 30 by 12 days; a day-12 revocation precedes the current day despite an old grant through day 20 | The public-quote path exists; other cases may remain pending/refused under supplied directives. No legal doctrine, actual private record, or deletion guarantee is needed. | | DEV-CON-04 | Credit-lot eligibility, duplicate effects, expiry-preserving restoration, and display correction | Recorded 7 may mean usable 2; duplicate T7 keeps 6; expired 3 restored on day 11 adds usable 0; display correction targets 6, not a ledger increase to 8 | Complete independent case ledgers are supplied. No cash valuation, discretionary credit award, or unseen transaction is required. | | DEV-PLAN-02 | Group coverage with room/instructor constraints and an infeasibility bound | At least 3+2+2=7 room slots exceed original 6; a four-slot two-room witness exists. Cost 36+24+20+25+15+15=135, headroom 5 | The extension is already authorized in the packet. A failed first schedule alone is not the requested impossibility argument; no hidden extra room or chosen ordering is required. | | DEV-PLAN-03 | Grade-specific daily conservation, receipt peaks, and reserves | Fresh closes 50,100,150; recovered closes 0,0,100; receipt peaks stay within capacity. C would make 1,200>700. Total charge 150, headroom 5 | Whole offers, use grades, opening stock, reserve, and cost rules suffice. No outside water-quality assumptions or dumped stock are needed. | | DEV-PLAN-04 | Integer programme selection with quotas, overheads, and budget/time tradeoffs | A/B/D/E film time 145 +15 turnover +15 break=175, fees 80. E→C gives 180 minutes but fees 95>90 | All film attributes and full costs are supplied. Any valid ordering is accepted; no audience preference, actual contract, or unique-order proof is required. | Each individual guide includes the complete C1–C12 mapping and a direct reasoning explanation or feasibility witness. These keys describe supported facts and acceptable alternatives, not full participant deliverables. Presentation preferences, unknown outcomes, external knowledge, and optional extra calculations cannot become hidden requirements. Mechanical ID/link checking supplements this manual review but does not establish task validity through model execution or empirical evaluation.