R03 Verification Cycle D — Gemini correction round (CCKE v1.3)

Date: 2026-09-01 | Verifier: Claude (Crossref API + web two-check)

Summary table

Detail

1. S8 — FAILED (fabricated hybrid citation)

• DOI 10.1016/j.landusepol.2022.106531 (Crossref) = Polson, M., Bodwitch, H., Biber, E., Butsic, V., & Grantham, T. (2023). After legalization: Cannabis, environmental compliance, and agricultural futures. Land Use Policy, 126, 106531. Note: no Petersen-Rockney; includes Grantham; vol 126 not 123.
• The printed link 10.1016/j.landusepol.2022.106351 = Grigorescu et al. 2022, Bucharest COVID sub-urban farming, LUP 122:106351 (confirms prior cycle’s finding).
• A real paper titled “Shifting geographies of legal cannabis production in California” DOES exist: Dillis, C. et al. (2021). Land Use Policy, 105, 105369. DOI 10.1016/j.landusepol.2021.105369 (first author Dillis, NOT Polson).
• Gemini’s entry welds the Dillis 2021 title/topic onto a Polson-team author list with the wrong volume and two candidate DOIs, neither matching the stated tuple. Correction inherits full burden: FAILED.
• Correct tuples to offer: (a) Dillis et al. 2021, LUP 105:105369; (b) Polson et al. 2023, LUP 126:106531 (“After legalization…”).

2. S10 — FAILED (fabricated title/issue/pages on a real DOI)

• Crossref: DOI 10.1016/j.drugpo.2014.01.001 = Chouvy, P.-A., & Afsahi, K. (2014). Hashish revival in Morocco. International Journal of Drug Policy, 25(3), 416-423.
• No paper titled “Hashish clubs and the changing nature of the Moroccan kif economy” exists in Crossref (search returned only the real Chouvy/Afsahi corpus: Hashish revival 2014; GeoJournal 2022 terroir; IJDP 2024 property rights; Belgeo 2020; Afsahi Tempo Social 2017).
• Verdict: FAILED. Correct tuple: Chouvy & Afsahi 2014, IJDP 25(3):416-423, DOI 10.1016/j.drugpo.2014.01.001.

3. S4 — PARTIAL (wrong author list; prior corpus record is the right one)

• Crossref: DOI 10.5751/ACE-01134-130102 = Gabriel, M. W., Diller, L. V., Dumbacher, J. P., Wengert, G. M., Higley, J. M., Poppenga, R. H., & Mendia, S. (2018). Exposure to rodenticides in Northern Spotted and Barred Owls on remote forest lands in northwestern California: evidence of food web contamination. Avian Conservation and Ecology, 13(1), art. 2.
• Gemini’s “Franklin, A. B., Carlson, P. C., Rex, A., …” author list belongs to a DIFFERENT rodenticide-owl literature (Franklin et al. work on golden eagles/owls elsewhere) — it is not this paper. Article number is 2, not 10.
• Prior cycle’s record (“Gabriel et al. 2018, ACE 13(1):art.2”) is CORRECT; Gemini’s correction is wrong.

4. CC017 — VERIFIED

• The ACE paper (ace-eco.org/vol13/iss1/art2) reports 70% of Northern Spotted Owls and 40% of Barred Owls exposed to one or more anticoagulant rodenticides. Claim magnitudes exact. YES.

5. CC003 trace — PARTIAL

• Real origin: CDFA CalCannabis SRIA (Economic Impact Analysis of CalCannabis Cultivation Licensing Program, prepared by ERA Economics/UC Davis team, submitted Jan 2018; work done 2017) — estimated ~13.5 million lb annual CA production with “more than four-fifths” (~80%+) leaving the state (widely reported Jan 2018, e.g. NBC News citing CDFA). SRIA at dof.ca.gov (Cultivation_SRIA_CDFA_1-5-2018.pdf).
• So “13-15M lb / 80-90% illicit, from 2016-2017 economic forecasting” is directionally right on source and vintage; the 15M upper bound and 90% upper bound exceed the SRIA figures (later ERA 2024 Outlook revised DOWN to ~11.4M lb unlicensed, ~9M lb trafficked out).
• Gemini’s specific mechanism claim (“assumed CA supplied entire US illicit demand”) was NOT confirmed — the SRIA derived production estimates and netted out in-state consumption; no evidence found that it assumed CA = 100% of US illicit supply. That characterization is an unsupported gloss. PARTIAL.

6. CC008 — PARTIAL

• License half: DCC-derived counts show active cultivation licenses 8,493 (end 2021) → 4,805 (end 2024), a 43% drop (Cannabis Benchmarks/DCC; Cal Poly Humboldt Cannabis Studies Lab tracks counts vs max canopy). Prior corpus figures (~10,700 in 2022 → <5,200 by 2025) imply >50% over that specific window. “Over 50% 2019-2024” is not precisely matched by any single verified series — plausible but the stated window/magnitude pair is unconfirmed. PARTIAL on exact framing.
• Canopy half: corroborated in direction — SB 833/DCC rules allowed converting multiple small licenses into one large license; 2023→2024 licenses fell 18% while licensed production rose 11.8% (ERA Economics California Cannabis Market Outlook 2024, cdn.cannabis.ca.gov). Recommended citation for this half: ERA Economics 2024 Outlook + Cal Poly Humboldt Cannabis Studies Lab license/canopy series.

7. CC011 — PARTIAL

• StatCan NCEA-derived household spending: 2024 licensed non-medical $6.071B vs unlicensed $2.654B → legal ≈ 70% of recreational spending; 2023: $5.811B vs $2.811B → ≈ 67%.
• “~70-80%” overstates the top: verified range for 2023-2024 is ~67-70%. Source exists and direction is right; magnitude band is inflated. PARTIAL — correct to “approximately two-thirds to 70%”.

8. ACLU 2020 — VERIFIED

• ACLU (2020). A Tale of Two Countries: Racially Targeted Arrests in the Era of Marijuana Reform (Edwards & Greytak, April 2020) exists; covers 2010-2018 FBI/UCR data; finds Black people 3.64x more likely to be arrested for possession nationally, disparities persisting in every state INCLUDING legalized states, and larger disparities in 31 states in 2018 vs 2010. Supports the persistence claim. Caveat stands: advocacy-organization analysis of official arrest data; note institutional position in provenance field. YES.

9. S13 UNODC Morocco 2003 — VERIFIED

• UNODC/Govt of Morocco Cannabis Survey 2003 (unodc.org exec summary): 134,000 ha cultivated in the Rif; resin production ~3,060 t (UNODC WDR 2006 restates 3,060; some UNODC docs cite 3,080). “~3,070 t” sits within the source’s own reported range — acceptable with a footnote that UNODC’s own documents vary between 3,060 and 3,080 t. YES.

10. CC025 premise — OVERSTATED

• UNODC Morocco/Afghanistan surveys were indeed satellite + ground survey hybrids (identity holds).
• BUT the claim “North American illicit production structurally hidden from satellite” is contradicted for outdoor/greenhouse cultivation: Butsic & Brenner (2016, Environmental Research Letters 11:044023, “Cannabis agriculture in California…”) and the UC Berkeley Cannabis Research Center digitized thousands of grow sites (4,428 sites in 60 Humboldt watersheds) from <1 m resolution Google Earth satellite/aerial imagery, distinguishing outdoor plots and greenhouses; later CRC work extended this (“The emergence of cannabis agriculture frontiers as environmental threats,” Butsic et al. 2018, ERL 13:124017).
• Defensible narrowing: North American illicit production lacks any SYSTEMATIC national satellite monitoring program (unlike UNODC state-partnered surveys), and indoor cultivation is invisible to imagery — but small-plot outdoor/greenhouse cultivation IS detectable and HAS been mapped academically. CC025 should be rewritten to “no institutional satellite survey program,” not “structurally hidden from satellite.” PARTIAL/FAILED as worded.

Counts

• VERIFIED: 3 (items 4, 8, 9)
• PARTIAL: 4 (items 3, 5, 6, 7) + item 10 partial/overstated
• FAILED: 2 identity fabrications (items 1, 2)