The mailing went out on April 22, 2021. An unrelated ICRC “door-drop” campaign followed around May 25, and a TV spot ran in German-speaking Switzerland in the same period. This triggered the preregistration’s Case I contingency: the narrow window (gifts through May 31) is the principal sample for hypothesis tests, with the broad three-month window as descriptive robustness. The narrow flag keeps all letters and simply does not count gifts arriving after May 31.

3  Main results: The impact of impact information

information

anipulation strength and relevance: the cost-per-output treatments must have been meaningful, salient, and representative of what charities actually do and would

Our closest arm-level comparison holds the CHF 150 ask fixed. The control text asks for CHF 150 and mention

scheme with a large unit size substantially increased average giving.

not yet reviewed by the authors.

Jan Schmitz University of Münster

Our closest comparison keeps the monetary ask fixed at CHF 150:

that a well-designed unit-donation scheme can increase giving, especially with a large unit size. Their full treatment does more than report a unit cost: it reframes the decision as choosing physical units and can restrict choices to a unit grid

It does not show that private philanthropy can replace government aid, or identify how to build political support for official development assistance

Know of high-impact research we should consider for evaluation — or other work this tool should be doing? Leave an email if you're open to follow-up discussion; if we later introduce compensation for useful contributions, earlier contributors will be grandfathered in.

Since the workshop: a company-reported production run at 22,000 litres and a new hydrolysate review bear directly on the scale and media questions. Dated updates and their limits.

9, 2026. These developments

kind of food people would buy

generally expected

wide

differed over how far particular substitutions could go,

Rate/discuss (Team) Rate/discuss (Public)

Quick rate

Unjournal commission an evaluation? (-- Strong No … ++ Strong Yes)For aggregation and sorting, the five quick ratings translate to percentile-equivalent scores: -- = 10, - = 30, ~ = 50, + = 70, and ++ = 90. These are deliberately spaced away from 0 and 100..

More filters and source controls

Show/hide scoring rubrics & methodology

About this prioritization tool

You are not asked to become a modeler; you are asked to make your expertise modelable. The

visible CM_01 point estimates

Optional uncertainty range (10th and 90th percentiles)

Can you give an A1 cost estimate?

Jump to another question

Full definitions and accounting boundaries

at least 2,000 metric tonnes per plant per calendar year (2 kt/year).

Qualifying

The authors have made the atlas, methods, data, and replication code easy to inspect. We found some research indexing and light public discussion, but no major institution or policy report using the measures yet.

Several plausible routes could lower cultivated meat production costs.

Several plausible routes could lower cultivated meat production costs. The harder question is whether they can work together in a reliable process at commercial scale. Our current reading of the evidence favors testing complete, internally consistent production scenarios before treating lower ingredient prices or higher cell densities as a new cost forecast.

Key Takeaways

Embedded in media discussion

Framing note: Recent industry consolidation — including several company closures in 2024–2025 — is one set of data points, not a verdict on any particular TEA's methodology. Company outcomes depend on funding conditions, market timing, and management decisions that are largely independent of the underlying cost trajectory. The workshop aimed to evaluate the technical evidence on its merits.

PF

Public working summary Workshop held May 8, 2026 Last reviewed July 20, 2026 S2 was off-record and is excluded. Its section below gives only the announced session scope.

Informal Pre-Session Wed May 6, 2026 · 11:00am–12:00pm ET · Zoom · recorded Casual orientation for participants with limited May 8 availability — mostly preparation and broad discussion. Walkthrough of the beliefs form and cost model. David Manheim (Technion/ALTER) and Mirjam Capuder (University of Maribor) participated. It covered introductions, a walkthrough of the interactive cost model dashboard, and early framing questions about key modeling uncertainties — including Manheim's perspective on expert elicitation methodology and how to structure belief-updating around contested TEA assumptions. Substantive points from this session are incorporated where relevant in the S1 and S3 notes below.

These map onto the workshop's structure: cell-line engineering → Crux 1; hydrolysate substitution → Crux 2; albumin and recombinant proteins → beliefs-form question E4; "what share of a hybrid product is cultivated?" → Crux 4.

Regulatory milestones are not directly comparable across jurisdictions. The FDA's February 2026 inventory lists five completed US cultured-cell food consultations, including three completed in 2025; market authorization, inspection, and product scope can still differ.

it

Evaluation 2 of this forecast

in response to this and other evidence.

S2-derived claim,

rigorous

asitka et al. (2024): ~$6/lb under continuous production; Humbird (2021):

S2 was off-record and is excluded. Its section below gives only the announced session scope.

Tool updated: Aug 28, 2026Normal cadence: papers twice weekly; feedback hourlyPaper discovery, scoring, matching, and the full dashboard rebuild normally run on Monday and Thursday. New rating and suggestion submissions are checked up to hourly. The dated chips show the actual latest completed updates.Last paper scan: Aug 31, 2026Last AI scoring: Aug 31, 2026Crux/PQ match update: Jul 23, 2026Human feedback: 17 current ratings across 13 papersPublic anonymous aggregate refreshed Sep 1, 2026. Each rater's latest rating per paper counts once. Discussion and a display name are included only after explicit public-sharing consent; emails, rating keys, and submission ids are never included.Earlier API cost: at least ~$4.06 reconstructed

Normal cadence: papers twice weekly; feedback hourly

A public economics column lists the review among its sources. This suggests some circulation outside the World Bank, although the article does not assess the review’s methods or health-policy conclusions in detail.

Mean absolute score movement?

?

movement?

Agreement alpha?

ecent AI-impact/governance papers Ordinary rerun

AI score stability metrics, with paper cohorts and variation sources shown side by side AI stability metric

17.2 points

cautious

rivate review found that one run treated prior public scrutiny as substantial while the other called it unclear, materially changing the estimated value of a new evaluation

identified publicly.

One same-wording execution pair differed by 30 points.

Prior human-rated comparison papers were less stable on both dimensions: ordinary rerun movement averaged 8.5 points and the wording effect averaged 5.8 points. 2/6 (33%) had wording effects larger than ordinary rerun movement; wording-mean alpha was 0.69.

Recent AI-impact/governance papers: ordinary same-wording reruns differed by 4.1 points on average, while the repeat-averaged wording effect was 1.1 points. 0/6 (0%) had a wording effect larger than their ordinary rerun variation, and wording means produced no broad-tier changes. Prior human-rated comparison papers were less stable on both dimensions: ordinary rerun movement averaged 8.5 points and the wording effect averaged 5.8 points. 2/6 (33%) had wor

Recent AI-impact/governance papers: ordinary same-wording reruns differed by 4.1 points on average, while the repeat-averaged wording effect was 1.1 points. 0/6 (0%) had a wording effect larger than their ordinary rerun variation, and wording means produced no broad-tier changes.

ordinary same-wording reruns differed by 4.1 points on average, while the repeat-averaged wording effect was 1.1 points.

Six recent AI impact/governance papers and six papers with prior Unjournal team ratings

026-08-31T23:54:58.113454+00:00

Held-out check: a third baseline execution is kept outside the balanced headline analysis and used only as a robustness check.

What is being tested: every stability number below describes AI model scores. “Previously human-rated papers” names a paper cohort selected because earlier Unjournal ratings are available; those ratings do not enter the stability calculations. Human and AI scores are compared separately in the amber reference panel.

mber reference panel.

paper-level human ratings.

calibration

Average the two runs within each reviewed, equivalent wording, then compare those wording means. This reduces ordinary rerun noise in the wording comparison.

Run each wording twice with the paper, rubric, model, settings, and output schema fixed. Differences within a wording estimate ordinary run-to-run variation.

prompt variants

A shared first pass: Browse research that may inform policy, funding, or further evaluation. Scores rank the value of attention or evaluation; they are not quality grades.

Forum Post Cruxes and Pivotal Questions Explorer

Living reviewFilter to papers surfaced by one or more expert-maintained living literature reviews. A paper keeps this provenance even if it originally entered through another academic source. All Reviews ▾ AI Accountability ReviewBridging BoundariesExistential CrunchLauren Policy: Migration Literature ReviewMonitoring Gene DrivesNew Things Under the SunRegression to the MeatReskilledScaling in Human SocietiesThe Care GapThe Patentist Review / UJ overlapFind review posts that cite papers The Unjournal has already evaluated, has in its pipeline, or previously considered. This is intended as an engagement and follow-up queue. All Any UJ overlap UJ evaluated In pipeline / considered Targeted intakeRequested, topic-focused additions outside the broad recurring scan. Use this to include all records, exclude deliberately oversampled batches, or inspect only one targeted run. Include allExclude targeted intakeOnly targeted intakeOnly: AI impacts on global health and development (29)Only: Animal-welfare focused intake (19)Only: Empirical conflict replication-game intake (14)Only: Large-N GCR and existential-risk quantitative evidence (6)Soil-invertebrate crux sweep (6 cruxes) Crux/PQFilter to papers that map to one or more Unjournal Pivotal Questions or community cruxes (from the cruxes explorer). "PQ match only" restricts to papers mapped to a Pivotal Question. All Has crux/PQ match PQ match only RecencyFilter papers by how recently they were released or last updated. Useful for focusing on the newest research that may benefit most from timely evaluation. Any time Last 30 days Last 3 months Last 6 months Last year Last 2 years Last 5 years ModelThe AI model used to score each paper. New papers receive a standard GPT-5.5 assessment; stronger candidates receive a second, more thorough GPT-5.5 assessment with higher reasoning effort. "UJ historical" papers were prioritized by the Unjournal team (not AI). Papers marked WAIT are temporarily excluded from AI ranking pending a genuine model reassessment. All Models ▾ UJ historicalawaiting genuine model assessmentclaude-sonnet-4-6gpt-5.4gpt-5.4-minigpt-5.5 (codex headless, high)gpt-5.5 (codex headless, medium)gpt-5.5 (codex headless, xhigh)manual_public_followupopus (headless) Also show small-model scored (mini/haiku) Also show UJ historical Also show legal scholarship Targeted intake catalog · 74 records · 5 search families

Source text (unverified type) source noteText supplied by the discovery source; its status as a formal abstract has not been verified.

AI-assisted prioritization for The Unjournal evaluation

he AI premium is concentrated in frontier and intensive AI use, including closed-source models, paying or seasoned users, and long prompts, rather than casual or open-weight use.

interaction-and-communication skill content strongly associated with higher AI exposure, while analytical, scientific, and operations-control skills load more negatively.

predicts

The paper informs live decisions about AI labor-market adaptation, AI industrial policy, competition and market-concentration monitoring, financial-risk exposure to AI shocks, and prioritization of retraining or adjustment support across occupations. It i

causal welfare effects.

This is a high-value Unjournal candidate: it is an NBER working paper using unusually granular proprietary data on realized AI consumption to estimate how AI demand is priced across firms, sectors, countries, and occupations. The findings could inform decisions

AI score the weight of three quick ratings and updates it with each rater's latest, evidence-weighted human rating.

How human suggestions reach team review

r sufficiently strong, discussed public feedback—places the paper in a private SQLite-to-Coda queu

Public-safe

Source abstract

Quick-rate mode: for each paper, how strongly should The Unjournal commission an evaluation? (-- Strong No … ++ Strong Yes). Each click is saved to The Unjournal's server. Clicking another option for the same paper updates this browser's rating rather than adding another. Add your name or email once below; the detailed form will reuse it.

More filters and source controls DisciplineThe academic field or sub-discipline of the paper, as classified by the AI model. Used to filter by research methodology and domain expertise. All Fields ▾ AI & Data ScienceAnimal WelfareDevelopment & AgriculturalEconomicsEnvironmental & ClimateLabor, Education & HealthOtherPhilosophy & EthicsPolicy & GovernancePolitical Science & LawPsychology & BehavioralStats, Finance & Methods SourceHow the paper entered this dashboard. Source is a discovery route, not a publication venue, endorsement, or quality judgment.Unjournal database: imported from The Unjournal's existing evaluation and prioritization workflow.Targeted public-paper follow-up: an independently public paper or DOI found during a requested topic-focused search. It does not mean the author supplied or endorsed the score.Academic feeds: NBER, CEPR, and RePEc working-paper feeds; arXiv and SSRN preprints; OpenAlex and Semantic Scholar indexes.EA Forum: research linked from EA Forum posts.Research organizations: Anthropic, DeepMind, and selected AI governance or safety organizations.Legal sources: legal-scholarship searches covering OpenAlex Law, law reviews, and the Institute for Law & AI. All Sources ▾ AI Governance (arXiv)AI Safety OrgsAnthropic ResearchDeepMind ResearchEA ForumNBEROpenAlexRePEcSSRNSemantic ScholarTargeted OpenAlex intakeTargeted public-paper follow-upUnjournal databasearXiv Targeted intakeRequested, topic-focused additions outside the broad recurring scan. Use this to include all records, exclude deliberately oversampled batches, or inspect only one targeted run. Include allExclude targeted intakeOnly targeted intakeOnly: AI, global health, and development curation (3)Only: Animal-welfare focused intake (19)Only: Empirical conflict replication-game intake (14)Only: Large-N GCR and existential-risk quantitative evidence (6)Only: Transformative AI, global health, and wellbeing (16)Soil-invertebrate crux sweep (6 cruxes) Crux/PQFilter to papers that map to one or more Unjournal Pivotal Questions or community cruxes (from the cruxes explorer). "PQ match only" restricts to papers mapped to a Pivotal Question. All Has crux/PQ match PQ match only RecencyFilter papers by how recently they were released or last updated. Useful for focusing on the newest research that may benefit most from timely evaluation. Any time Last 30 days Last 3 months Last 6 months Last year Last 2 years Last 5 years ModelThe AI model used to score each paper. Papers are scored in tiers: a fast model (gpt-5.4-mini) screens all candidates, a stronger model (gpt-5.4) re-scores the top papers, and the most capable model (gpt-5.4-pro) provides detailed analysis of the highest-ranked papers. "UJ historical" papers were prioritized by the Unjournal team (not AI). All Models ▾ UJ historicalclaude-sonnet-4-6escalatedgpt-5.4gpt-5.4-minigpt-5.5gpt-5.5 (codex headless, high)gpt-5.5 (codex headless, medium)gpt-5.5 (codex headless, xhigh)manual_public_followupopus (headless) Also show small-model scored (mini/haiku) Also show UJ historical Also show legal scholarship Targeted intake and prioritization requests - 64 records across 5 paper search families and 1 crux search family This shared catalog records topic-focused searches outside the broad recurring scans. Initial and follow-up passes are combined below as one search family. A targeted label records how a record was found; it is not an endorsement, quality judgment, prevalence estimate, or completed Unjournal prioritization decision.Large-N GCR and existential-risk quantitative evidenceResearch papers (6 papers; 2026-07-30) Credible large-N, cross-country, historical, forecasting, survey, or other quantitative evidence directly bearing on global catastrophic or existential risks. Public OpenAlex search, quantitative-design triage, deduplication, and prioritization scoring. This is a separate follow-up to the empirical-conflict search. It asks whether global catastrophic and existential-risk questions have credible quantitative literatures with enough observations or repeated judgments to support evaluation or replication. It excludes primarily conceptual argument, technical engineering, and ordinary civil-conflict studies without a direct catastrophic-risk connection. Inclusion is not an endorsement or a completed Unjournal team decision. View 6 records →Transformative AI, global health, and wellbeingResearch papers (16 papers; 2026-07-24) Economic, social-science, and policy research relevant to global health and wellbeing under transformative AI, especially implications for lower-income countries. Search-pass details (2)Transformative AI, global health, and wellbeing follow-up: Second-wave public-paper search focused on first-run gaps, followed by deduplication and Codex subscription scoring. This second wave follows up the initial requested intake and deliberately targets gaps it left: public finance and tax-base resilience, distribution and social protection, AI market power, worker attitudes, and developing-economy structural transformation. It uses only independently public paper or institutional publication pages. Inclusion is not an endorsement or a completed Unjournal team decision.Transformative AI, global health, and wellbeing: Public OpenAlex search, theme-fit triage, deduplication, and Codex subscription scoring. The search follows the linked Coefficient Giving request for proposals but is an independent Unjournal intake exercise. It emphasizes non-catastrophic transformative-AI scenarios, labor markets, fiscal capacity, social protection, biomedical and health-system bottlenecks, development strategy, and aid allocation. Inclusion does not imply endorsement by Coefficient Giving or an Unjournal team decision. Coefficient Giving request for proposals. View 16 records →

Median with 80% model interval

Named public commitments

A steelman against the Bay-Area assumption that very large AI-lab-adjacent philanthropy will quickly become usable funding for AI safety and EA cause areas.

The reasoning goes that if there is always a high level of background risk to humanity, then we should expect to go extinct soon anyway, which means the importance of avoiding any one particular risk is not as valuable as it may seem. For more details see the full report here.

5 why this matters6

Newest independent response: a low-cost counterpoint

One additional written estimate is visible but not pooled

Andrew

DisciplineThe academic field or sub-discipline of the paper, as classified by the AI model. Used to filter by research methodology and domain expertise. All Fields ▾ AI & Data ScienceAnimal WelfareDevelopment & AgriculturalEconomicsEnvironmental & ClimateLabor, Education & HealthOtherPhilosophy & EthicsPolicy & GovernancePolitical Science & LawPsychology & BehavioralStats, Finance & Methods SourceHow the paper entered this dashboard. Source is a discovery route, not a publication venue, endorsement, or quality judgment.Unjournal database: imported from The Unjournal's existing evaluation and prioritization workflow.Targeted public-paper follow-up: an independently public paper or DOI found during a requested topic-focused search. It does not mean the author supplied or endorsed the score.Academic feeds: NBER, CEPR, and RePEc working-paper feeds; arXiv and SSRN preprints; OpenAlex and Semantic Scholar indexes.EA Forum: research linked from EA Forum posts.Research organizations: Anthropic, DeepMind, and selected AI governance or safety organizations.Legal sources: legal-scholarship searches covering OpenAlex Law, law reviews, and the Institute for Law & AI. All Sources ▾ AI Governance (arXiv)AI Safety OrgsANIMAL_LAWANIMAL_LAW_REVIEWAnthropic ResearchDeepMind ResearchEA ForumLaw & AI InstituteLaw reviewNBERNEP_LAWOpenAlexOpenAlex LawRePEcSSRNSSRN_LAWSemantic ScholarTargeted OpenAlex intakeTargeted public-paper follow-upUJ seedUnjournal databasearXiv Targeted intakeRequested, topic-focused additions outside the broad recurring scan. Use this to include all records, exclude deliberately oversampled batches, or inspect only one targeted run. Include allExclude targeted intakeOnly targeted intakeOnly: AI, global health, and development curation (3)Only: Animal-welfare focused intake (19)Only: Empirical conflict replication-game intake (14)Only: Large-N GCR and existential-risk quantitative evidence (6)Only: Transformative AI, global health, and wellbeing (16)Soil-invertebrate crux sweep (6 cruxes) Crux/PQFilter to papers that map to one or more Unjournal Pivotal Questions or community cruxes (from the cruxes explorer). "PQ match only" restricts to papers mapped to a Pivotal Question. All Has crux/PQ match PQ match only RecencyFilter papers by how recently they were released or last updated. Useful for focusing on the newest research that may benefit most from timely evaluation. Any time Last 30 days Last 3 months Last 6 months Last year Last 2 years Last 5 years

5. Three problems that a better estimate of the exchange rate won't fix

3. What this is based on

vignette calibration

Preliminary — do not read Unreviewed working draft, 30 July 2026. Not for circulation, citation, or quotation.

2. Bottom line

The decision this is about

A funder comparing a psychotherapy

Targeted intake provenance

Optimized messaging shifts choices towards plant-based foodsource titleExact title resolved from the linked paper's bibliographic metadata; discovery-page anchor text and model output are not used as titles.

Strong near-term candidate for animal-welfare evaluation because it tests behaviorally relevant messaging around plant-b...

2 ratings

Mar 23, 2026

Decision Relevance

abstract provenanceSource-supplied abstract wording; HTML entities and whitespace normalized. Not independently compared with the paper PDF.

Rate on a 0–100 percentile scale relative to all papers in our database (Strong Yes=top quintile, Strong No=bottom). Try to form your opinion before reading the AI discussion above. Discuss why you gave that rating.

2 ratings

Email

Rate (Team) Rate (Public) View Paper

A documented elicitation limitation.

displayed submittable slider defaults, so some responses may be anchored or untouched.

Direct validation of the linear WELLBY against better-specified alternatives — per Benjamin, not yet done.

would answer the question a

What would change this answer

evidence

Do if you have budget or control an instrument

Flag explicitly when a conclusion depends on the location of the neutral point.

Use directly measured life satisfaction wherever a trial reports it, in preference to a converted mental-health measure.

nd publish sensitivity bounds

1. The decision this is about

Ratios move even when coefficients do not

instruments

SDs inherit the population you measured them in A standard deviation is a property of a sample, not of a person.

Mechanically, comparisons will tend to favour interventions run in more heterogeneous populations.

2. Bottom line

The reason for the SD step is data, not theory.

income-doubling to lives-saved, lives-saved to DALYs — then backing out the remaining si

its conversion

Founders Pledge raised these questions with us through our Pivotal Questions programme;

3. What this is based on

cheap fixes are known

ny single global conversion factor therefore systematically mis-weights domains

Disability weights top out around 0.3, while depression and anxiety show losses above a full life-satisfaction point.

The likely error is directional, not symmetric. Compression and ceiling effects both push toward overstating the wellbeing gains from mental-health interventions relative to physical ones. A conversion factor that is wrong in a known direction should not be treated as noise.

han a near-1:1 point estimate implies

Do not describe it as established.

and

The SD-to-SD step is defensible as a working assumption, and it is nobody's preferred method — including its originators'. Keep using it; report it as an assumption; run sensitivity analysis. Do not describe it as established.

The question funders ask us is not "is subjective wellbeing valid." It is: can I keep using this step, and how wrong might it make my ranking?

Founders Pledge's moral weights use it. Happier Lives Institute's cost-effectiveness work generates it. Anyone reading either inherits it.

Almost every such conversion in current use

one standard deviation of improvement on a depression or anxiety instrument as one standard deviation of improvement in life satisfaction

means

Jul 2, 2026

Identify the fragile assumptionIn your preferred bottom-up TEA, which single assumption would break the headline result if it were wrong by one order of magnitude? Which interactions …

workpapers

traceable draft calculations

Stop and obtain specialist review

The PDF output is a draft field packet, not a filled official IRS PDF.

PDF/XLSX extraction and an AJ Bell-specific parser are not yet automated.

QEF, mark-to-market, purging elections, foreign-tax-credit allocation, and notional accumulation-unit determinations are not implemented.

The calculator refuses to guess a missing rate or cumulative interest factor.

Reset to evaluation lens Current to

Job-matching and recommendation systems affect labor-market frictions, unemployment duration, and match quality, all of which have major welfare and policy consequences.

Decision relevance

Crux/PQ match update:

API cost: not shown in this export

Human ratings visible here:

Adjust these sliders to create your own priority score, then sort by Custom weights. This changes only your browser view; it does not change The Unjournal scorer.

Human ratings visible here:

The useful question now is not whether the old paper should be revived unchanged. It is what, if anything, is worth testing under current technology and policy.

gument in an Essex Economics discussion paper and an unfinished theory project

The strongest warning comes from take-up: Ofcom reports 532,000 UK social-tariff customers in June 2025, only 8.6% of a conservative proxy for eligible households.

so "free redistribution" is too strong. The more defensible claim is narrower: this could create additional purchasing power for some low-income consumers with less public expe

lower pric

The OpportunityCard, ten years on

::: {.callout-note collapse=“true”} ## Mo

Calibration anchors are a small set of real example papers that teach The Unjournal's AI prioritization scorer what “value of evaluation” looks like in each area — including the boundary cases where a good paper is not a good candidate. Each anchor's rating is the team's real prioritization rating; the proposed lesson is an AI inference about the calibration takeaway. Use this page to confirm or correct each rating and lesson, suggest new anchors, and track the discussion.

Research training

Foie gras (luxury/unstructured)

How to read each column: the conventional product's retail price; the cultivated cost the model delivers to retail (your biomass slider + scaffold for cuts + markup); their ratio R; and the estimated share of that product's own market cultivated would capture (not a share of all meat). Share-bar colour: green > 30%, amber 8–30%, red < 8%.

One caveat to this species-by-species framing (our own intuition, not in his model): especially early on, a cultivated “chicken” or “shrimp” product may be received as its own distinct food rather than competing head-to-head only with the conventional product it imitates, so cross-category substitution could matter more than a per-species contest implies.

tated expert

What drives the answer.

On agreements: we draw on the same source literature and read it similarly. Humbird’s 2021 pessimism was driven mostly by amino-acid/media cost, which is not a hard thermodynamic constraint and which Pasitka’s 2024 empirical work (hydrolysate at $0.63/L) pushed down sharply. Both of us read this as suggesting cultivated meat likely lands a few-fold above conventional meat — not at parity, but not orders of magnitude off either.

$/kg →

Cost

Share colour: green > 30%, amber 8–30%, red < 8%. Ordering across species is driven almost entirely by the price ratio R (cost fixed, conventional price varies), plus the per-tier authenticity offset — reproducing Pablo's inversion: cheap chicken and pork resist, expensive beef, seafood, and luxury foie gras are penetrable.

may capture little chicken or pork share

validates out-of-sample against plant-based milk

and its 89% mainstream split,

it fills the half we skip;

an ad hoc dependence structure

weak