Representation and exit Q-001 · Representation / reference perspectiveWho the firm represents, or which perspective frames trade-offs. ↓ reveals Q-009 · DeadlockWhat happens when two members cannot agree.
A bit more description of this dependency. You've made it clear that question one may reveal question nine, but what answers to question one reveal this, or is it just putting any answer that reveals it?
The authors published a detailed response and reported revising the working paper to address evaluator comments. The public evaluation summary records changes to framing, methods, transparency, and the cost-effectiveness analysis while the headline result remained broadly similar.
missing some content here .. add Givewll's response etc
Empirical conflict replication-game follow-upResearch papers (8 papers; 2026-07-24)
most of this is on civil wars etc., and it can be more robust because the data is richer .. But for GCR/Xrisk there was not much found. Is there credible 'large-N' quant analysis worth targeting? -- maybe do a nother target5ed run for that?
Empirical conflict replication-game intakeResearch papers (6 papers; 2026-07-24) Recent empirical economics, political science, and related research with conflict or political violence as an outcome. Public OpenAlex search, venue and empirical-design triage, deduplication, and Codex subscription scoring. This run supports consideration of a topic-focused replication game. It favors empirical papers in a coherent interdisciplinary conflict literature, especially work using geocoded or commonly shared conflict data and designs suitable for robustness or replication discussion. It excludes broad international-relations commentary and historical theory. Policy relevance and the quality of underlying conflict measures are treated as substantive uncertainties, not assumed strengths. View 6 papers →Empirical conflict replication-game follow-upResearch papers (8 papers; 2026-07-24) Follow-up to Empirical conflict replication-game intake. Empirical conflict-outcome studies suitable for a coherent replication exercise, emphasizing spatial data, shared event datasets, design comparability, and measurement sensitivity. Second-wave public-paper search focused on canonical geocoded designs, common conflict datasets, and measurement fragility, followed by Codex subscription scoring. This second wave follows up the initial conflict intake with canonical and recent empirical papers that make the proposed replication game's shared data and robustness questions more concrete. It includes one explicit measurement-methods anchor alongside outcome papers. Inclusion is not an endorsement, and policy relevance and conflict-data quality remain open questions. View 8 papers →
merge these in the epxlanation
Optimized messaging shifts choices towards plant-based food Animal welfare & food systems High Priority Dec 31, 2024 |Laura Thomas-Walters, Jacy Anthis, Joshua Tasoff |Targeted public-paper follow-up
title is changed!!!! Keep actual titles
24Evaluation priority (0-100): how strongly this paper is recommended for a Unjournal evaluation. Weighted for UJ prioritization — decision relevance, neglectedness, timing, methodology, and influence. NOT a rating of research quality. Evidence from scaling laws AI, technology & governance Lower Priority |EA Forum gpt-5.5 (codex headless, medium) crux: A deep critique of AI 2027's bad timelin... · 42% ▾ Details / Rate This is a highly influential OpenAI arXiv preprint on n
You changed the title!!! use actual titles!!
Requesting
add the usual signpost for hypothesis
clusters
tooltip: clusters of research questions, approaches, and environments
priorities
rename; "Field teams: curating research themes/questions?" .. but make font a bit smaller
0–5 min Updates and agenda choices Agree on where the meeting's attention is most useful today. 5–25 min Actual resear
this is just an EXAMPLE schedule
Wellbeing, psychology, and welfare economics
this isn't really a field grouo/not the focus of the Psych/attitudes field group ... e.g., they previously considered focusing on misinformation. Moral circles (specific empirical questions) could also be relevant, and perhaps human response to AI tools
Which empirical claims about AI scaling are decision-relevant and uncertain enough to merit independent technical auditing?
this is meta -- I was looking for actual examples
Global health and development
these are too methodological -- I'd prefer some specific ones (too) ... about specific interventions
Group-curated: perhaps five to ten active questions or clusters.
10 seems too many ... I'd prefer 5-7 prioritized max
It would add a more durable layer connecting those activities to the larger research questions and decisions we care about.
that's a bit too strong perhaps ... not all of those are connected
short explanation, rather than quietly disappearing.
perhaps folded or filtered
would
could
keep
maintain their own prioritization
This is a discussion proposal, not a decided policy or a finished framework. The examples are starting points meant to be criticized, rewritten, expanded, or rejected at the meeting.
shorten this ... and it's a discussion beyond the meeting
Could
Should
Field specialist groups as stewards of research priorities
shorten this top section.
AI assistant; Opus model useful for deep analysis of research papers and evaluation assistance.
needs updating -- Fable is now latest as of 13 Jul 2026
Full policy →
remove link (here and in applied versino) to Gdoc -- Gitbook is the VoR atm
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.
this is an annoying AI 'not this but that' again
theory
applied economic theory
Show detailed scoring rubrics & methodology
"Show" --> SHOW/HIDE ... and make it clearer that this is the 'toggle to open the content below' ... it's not shared or offset or boxed, so that's hard to see
March
need a 'fold all/unfold all' button here
Scores are calibrated against 353 actual human prioritization decisions from the Unjournal team. The AI scores are systematically compared to human assessor ratings, and field-specific corrections are applied. Read more about UJ’s prioritization process.
Link a page on 'how this was 'calibrated' here'
For prominent work (NBER, CEPR, World Bank, top journals), decision-relevance dominates. For less-prominent work, methodology becomes the tie-breaker—our evaluation could boost neglected but rigorous research.
clarify this -- see note abobe
Methodological potential
this needs more discussion and thinking. Even for prominent work the strength of the data/approach may be weighted somewhat. E.g., all else equal, real world measured outcome data is usually preferred to recall data, which might be preferred to hypothetical choice data. However, for prominent work we don't weight/rate the 'methodological correctness' of the work in rating its potential for impact. If it's prominent and ~influential already, if it has major flaws that makes it perhaps _ more_ important, to be publicly evaluated. In contrast, for less prominent work, clear methodological, logical, contextual, and communications flaws make it less likely to have potential for impact, and thus less valuable to commission for evaluation.
Prominence
Not sure we want a 'prominence' rating/weighting for less-prominent work -- that seems confusing and contradictoru
The list can be scored and sorted through either of two transparent weightings of the six sub-scores — toggle Evaluation-relevance weighting in the controls, and unfold What is this score? to see the live weights:
(Tooltip?) We may want to ask users to specifically rate papers by these multiple categories, in part to help refine the model
We welcome both team and public feedback.
Explain how we are also soliciting human ratings, and these wil be incorporateed into this tool, both in the direct prioritization, and informing the AI model.
Quick-rate mode
A tooldtip should explain 'quick rate mode'. And we should have a folding box (and or a link on 'get involved') explaining how we want direct ratings -- either quick +/- ratings, or detailed discussion input on prioritizing these research objects according to their potential for impact
CriterionWeightWhat it captures
Have an option to allow people to re-weight as they prefer and re-sort it.
FN · 2026-05-08
What does "FN" mean? Is it a meaningful response ?
2026
dates cut off here. find a way they show. Anonymous --> Anon. -- other stuff in tooltips?
ation tool and prototype AI-assisted evaluation and summary workflows, keeping humans in the loop for judgment and ratings.
and benchmark and investigate AI capabilities, alignment, priorities, and tast. Link this (unsuccessful) proposal https://llm-uj-research-eval.netlify.app/proposal ... but still a WIP project
udge research and write referee reports
rephrase, use our language -- not 'judge'
Pivotal Questions expansion
nascent
add "Innovative evaluation hosting and research conversation tools" -- https://uj-eval-hosting-study.netlify.app/
Forecasting tournament
not explained well. And it's an Animal Welfare forecasting tournament
and do the grant-writing.
I'll do most grantwriting work, but we're also interested in getting a bit of help from others especially if they are writting into the grant substnitally
and on what timeline
and if you have space, 'why you think funders would find it impactful'
Two or three sentences is plenty
they could add more ... it's just 2-3 is also ok. but guidance could help
We're building our next grant applications around what you'd want to do
that's a bit too strong -- it's a factor, but we're not necessarily going to implement exactly what people propose
When should you submit for public evaluation?
rephrase what this is and how it could represents impact/funding ... or drop this
elevant questions and the decision-makers who'd use the answers.
add a box or subbox for the proposed modeling hack: https://6a3a77696a52ea869c36e13d--uj-cm-workshop.netlify.app/modeling-hack.html
and Beliefs synthesis (still awaiting further responses, see https://uj-cm-workshop.netlify.app/beliefs-overview-r7p2k.html)
Recent work
also add https://info.unjournal.org/pivotal-questions.html -- that's a big one, and it's more than just the workshops
Recent work
Add https://info.unjournal.org/legal-scholarship.html -- this is an important direction we could go in
Recent
More narrative here. Draw more from our recent grant proposals and discussions
I-assisted prioritization prototype An early tool that suggests research to evaluate and gives preliminary indicators, using our principles and prior prioritization as context.
this looks good but put another more promionent box xplaining our LLM evaluation/benchmarking project: https://llm-uj-research-eval.netlify.app/
Pivotal Questions workshops
this is key -- flesh this out, describe more
Evaluator pool
that's not 'work' that's just statistics. Drop or put at bottom or in a resources fold
work
and proposed/nascent projects
Commissioned
read the link content more carefully -- that's not what we're proposing, and no one has taken it up yet
Animal welfare working group
it's not quite a working group
ost of these are lightweight, and several were built quickly — they're meant to show the shape of what's possible, not to be polished products.
make this atooltip
or staff — te
or otherwise working with us
so an IO question doubles as a welfare question and as a target for antitrust or procurement policy
But the animal welfare implications will go in the reverse to the usual antitrust direction -- monopoly/monopsony power limits quantity and thus the AW burden, right? (iirc I've seen some version of this before).
Key question for us: what is the vector we can potentially affect here/whats the ToC? (NB that's kind of our usual question)
The focal cost question is live on Metaculus: CM_01 — production cost per kg →. The expert-aggregation version is at CM_03 →. If you forecast on Metaculus, please share your username below so we can link your contributions.
Add a paragraph in this fold: You may also be interested in the Animal Welfare Futures forecasting tournament we co-launched with Metaculus and Sentient futures. $3,400 in prizes and only about 25 forecasters as of June 26. It includes some cultivated-meat questions (but not the specific cost questions here), alongside other animal-welfare-relevant questions you may find interesting.
And change the folding block title to "Public forecasting on Metaculus; Animal Futures Tournament"
contributions
Add a paragraph in this fold: You may also be interested in the Animal Welfare Futures forecasting tournament we co-launched with Metaculus and Sentient futures. $3,400 in prizes and only about 25 forecasters as of June 26. It includes some cultivated-meat questions (but not the specific cost questions here), alongside other animal-welfare-relevant questions you may find interesting.
And change the folding block title to "Public forecasting on Metaculus; Animal Futures Tournament"
short
remove "short"
Cost drivers (optional — any you have a view on)
make fields similar or identical to prev form -- consistency is key
Same quantity, but as of end-2027 — verifiable within a couple of years, which helps calibrate the longer-horizon estimate. If you expect no large-scale plants by then, estimate for advanced pilots and say so.
use same lang as prev form as ,much as possible for consistency
It should take about 15–30 minutes;
rethink this -- it prob takes longer
If you're a forecaster, modeler, or skeptic we've invited
Thinking about this again. We're going to try to get the skeptics to share their beliefs right away and without seeing the beliefs specified by the workshop participants. This doesn't really need to be on this page, although you could mention it perhaps in a folding box for these people. The forum for the non-cultured meat f orecasters skeptics should probably be on its own standalone page. We also should let them know that we're going to be asking them for follow-up after they've considered the full set of beliefs, with additional compensation. So perhaps $100 for their initial belief statement plus $50 for that follow-up?
What we're asking: share your beliefs on the focal cost question (CM_01) and any subquestions you have a view on, with brief reasoning, via the beliefs elicitation form. Roughly 30–60 minutes; partial responses are welcome.
Ask them to do at least a reasonable share of the sub-questions... This seems like they are way too optional.
CM Workshop · Beliefs Analysis
At the bottom of this, put a link to the "Next Steps" page as well as a link to the "Subquestions" page, to direct people's interest towards these, which are pretty high value
Pre / post workshop comparison
This stuff should be in a folding box, like "further details and notes. "
One contributor (a commissioned PQ evaluator, included anonymously at their request) submitted detailed answers in a written document using the original Pivotal-Questions wording, which differs from the workshop form on several questions. Their CM_14 and CM_17 map directly. Their CM_01, CM_12, CM_13 and CM_16 are shown as provisional conversions onto the workshop axes, using explicit stated assumptions:
Put this at the bottom. It's too prominent here.
she
Don't give a gender, just use the word "they".
Timeline
This is a bit less interesting for most people. I don't think it deserves its own tab, maybe a folding box somewhere.
Technical sub-questions and expert distribution questions. Only respondents who answered each question are shown.
We should have aggregation for these also. -- I only see the individual responses below.
CM_01: What will be the average production cost ($/kg, wet weight at harvest) of undifferentiated cultured chicken cell biomass in 2036, assuming large-scale commercial production is achieved?
Signpost this a little bit; I'm worried people are only going to see this question. Note that this was the focal question, but there's a set of other questions that are particularly interesting - link the "subquestions" tab.
Fairness note: prior workshop contributors who already submitted a detailed response are equally welcome to the same honorarium for a substantive post-workshop update.)
Make the fairness note just the words "fairness note" and then a tooltip explaining that the difference in compensation is necessary to elicit participation from people with less inherent interest in learning about cultured meat. And that wWe hope to offer this sort of modest compensation more broadly in the future.
of the first reasonably
"of the first four"
without ties to cultured meat
Italics not bold. Note that this helps balance (in an ad-hoc way) as people without a direct links to cultured meat may have been less likely to attend our conference. And this is also very important to help us get at the cruxes and make real progress in the public understanding.
Next steps
put a date here
Dataset filter
Filter is taking up a bit too much vertical space, make it a little bit less bulky.
CM_01: What will be the average production cost ($/kg, wet weight at harvest) of undifferentiated cultured chicken cell biomass in 2036, assuming large-scale commercial production is achieved?
Make the question title a bit more prominent here.
This is not a conversion gate like valuation, liquidity, allocation, or deployment. It is a reason to discount or stress-test local confidence when the local social graph may benefit from optimistic expectatio
This was a comment you didn't need to actually add literally.
The stronger skeptical case is about conversion, not existence.The strongest steelman is not "there will be no AI-wealth philanthropy." That claim now looks weak. Anthropic has confidentially filed a draft S-1 S1Anthropic confidential draft S-1"confidentially submitted a draft registration statement on Form S-1"Supports the claim that an IPO/liquidity event is live, while still conditional.Open source, and OpenAI's nonprofit Foundation controls a large equity stak
Is this really potentially a straw man here? Who was actually saying there will be no AI wealth philanthropy? Not sure we need this particular "not this, but that" Construction here.
What would update this memo?
Claude: Michael Dickens: "What would update this memo?" – as with the gates, no sense of prioritization is given. IMO by far the biggest uncertainty, about which we will get more information in the future, is: What will Anthropic's valuation be when the lockup ends? "Concrete donor vehicles" is also important evidence, but we won't get that until probably 6-24 months later.
Funders and field builders should prioritize grantmaker capacity
Claude: Michael Dickens: That section says "Funders and field builders should prioritize grantmaker capacity", but that's what to do if the skeptical case is wrong, not if the skeptical case is right.
How to plan if the skeptical case is live
Claude: Michael Dickens: The recommendations under "How to plan if the skeptical case is live" don't really make sense. AFAIK ~zero orgs are planning as if they're guaranteed to get a huge pile of donations 1–2 years from now. I believe nonprofits mainly plan based on the money they already have on their books + short-term (<1 year) fundraising expectations. "How to plan if the skeptical case is live" is just "business as usual".
Notional company equity value, not spendable cash.
Claude: Michael Dickens: My downward adjustments to the model aren't even the pessimistic case. The pessimistic case* is that the AI field collapses (investor funding dries up or something) and Anthropic stock is worth $0. Base rate says there's like a 50% chance that that will happen. Even optimistically, you should expect at least a 10–20% chance that Anthropic stockholders get nothing.
*this is pessimistic for donations but I would actually prefer that this happen because it would lengthen timelines. so in a way it's the optimistic outcome
The steelman could be too pessimistic if founders or employees treat liquidity as an urgent moral obligation
Claude: Michael Dickens: "The steelman could be too pessimistic if founders or employees treat liquidity as an urgent moral obligation" – TBH the BOTEC as written seems to me like it's already pricing in that founders/employees will treat donations as urgent, e.g. it's implying that Anthropic money will be disbursed faster than FTX Foundation money, which itself was disbursed at historic speed. IMO most likely reason why the model will end up underestimating is that Anthropic market cap ends up being like 10x higher than predicted.
Field absorption ceiling
Claude: Michael Dickens: "Field absorption ceiling" is structured more sensibly than "Grantmaker capacity multiplier", but these two seem redundant because they're closely related. If orgs have more capacity to expand, grantmakers can deploy money faster by giving to those orgs. If there are more grantmakers, they can create more and bigger RFPs. etc. I would include one variable or the other, but not both.
it scales raw potential funding by the available people, institutions, judgment, and legal plumbing needed to make good grants.
Claude: Michael Dickens: Also this would make more sense as a dollar amount, not a multiplier. Like there's a fixed total amount that grantmakers can reasonably disburse. You could model it in a more complicated way but IMO a simple cap is the way to do it. Or maybe don't use this parameter at all. I think it's probably worth including, but keep it simple.
Grantmaker capacity multiplier
Claude: Michael Dickens: "Grantmaker capacity multiplier" seems nonsensical as written. Shouldn't the capacity max out at 1x? If grantmakers are a complete non-bottleneck, then the other parameters will dictate the amount disbursed; if they're a bottleneck, then the amount disbursed will be less. There's no way for grantmaker capacity to have a multiplier >1x.
OpenAI Foundation stake
Claude: Michael Dickens: I would delete the OpenAI Foundation bit because (1) the model has enough parameters already and (2) I doubt OpenAI Foundation will give much money to causes that look good by EA lights.
Founder deployment by end-2026
Claude: Michael Dickens: IMO "Deployment by end-2026" should use a different date. IPO 3-6 months from now plus 6 months lockup means no money will be deployed in 2026, unless Anthropic does a fast IPO + early lockup release. Even by the end of 2027, you're talking about a 3-9 month turnaround time on lockup ending -> grants being disbursed. FTX Foundation donated $190 million (pre-clawbacks) in about 6 months, which was ridiculously fast compared to a typical foundation, and that was still a pretty small % of its long-term budget (or at least, what was believed to be its long-term budget before FTX collapsed).
Conversion gates
Claude: Michael Dickens: This model is supposed to illustrate how a lot of people are being too optimistic, but even then, I think most of the point estimates in the model are too optimistic. Consider that e.g. the median self-reported earner-to-give only donates (IIRC) 3% of their income.
Employee committed pool
Claude: Michael Dickens: "Employee committed pool" is defined in terms of dollars rather than as a % of company valuation, which seems weird. Shouldn't it depend on the value of the equity?
Founder pledge
Claude: Michael Dickens: Unclear what the "Founder pledge" variable refers to. Is it the % of pledgers' wealth that they've pledged to donate? If so, the default of 80% seems really high?
Anthropic valuation
Claude: Michael Dickens: "Anthropic valuation" variable should specifically be the valuation at the end of the 6-month lockup. Doesn't matter much for a point estimate but it would increase the variance if there variable were a probability distribution.
Editable assumptions
Claude: Michael Dickens: It looks like the default estimates in the BOTEC are pulled from the sources, but it's not clear which estimates came from which sources. There should be inline citations.
BOTEC model
Claude: Michael Dickens: Would be cool for the BOTEC to use distributions rather than point estimates. (Squiggle is good for this, and Squigglehub even has a built-in way to have AI generate models.) IMO distributions are a lot more informative than point estimates.
Gate 8: Bay social incentives make optimistic funding rumors self-reinforcing
Claude: Michael Dickens: Gate 8 ("Bay social incentives") seems uninteresting since it's not a claim in the same category as the others. It's more like a meta-level reason why people might not think about the other 7 gates.
Gate 5: lab-adjacent money may favor lab-compatible safety
Claude: Michael Dickens: Gate 5 and Gate 7 seem like they're saying the same thing.
A late-2026 IPO could still imply mid-2027 or later insider liquidity because of lockups.
Claude: Michael Dickens: Gate 1 says IPOs have lock-ups. That's true but I basically don't think that matters because lock-ups are very predictable: they will announce how long it is, and that's exactly how long it will be. There's no uncertainty. The main reason it's relevant is that a lockup gives more time for AI valuations to fluctuate or collapse, but the text doesn't even mention this.
Steelman gates
Claude: Michael Dickens: All 8 gates look correct to me, but they don't all deserve equal emphasis.
Read these as sequential gates. Realization handles liquidity and legal availability; follow-through handles intent; allocation handles cause choice; deployment handles the timing of actual grants by the deadline.
where do the defaults come from? Explain, reference, link (maybe as tooltips)
This may leave legally awkward, politically controversial, or non-lab-compatible work underfunded.
this needs fleshing out -- not sure what this is about
People who could puncture the rumor may also be financially incentivized not to alienate future donors.
I don't see how this would 'alienate future donors'?
main Hi Quick hello Suggest 5 min Call Phone call Suggest 15 min Video FaceTime / video Suggest 30 min Drive Drive-time call Suggest 20 min Msg Async text Suggest 1 hour
What does the room mean here
Selection lift s = Pr(D=1 | H) / Pr(D=1 | L)
this is probably something to endogenize in the model in some way? For now I guess it's OK
appa(e_full - e_copy)
latex note rendering here
4. Are you or your coauthors especially exposed to a durable negative public signal?
This framing feels too negative and definitive for my taste. As some of the modeling and discussion gets at, a single 'negative public signal' should not be so damning as people seem to think.
The people behind The Unjournal
Make this one the 'management team'?
2 · What The Unjournal is
add https://info.unjournal.org/ and unjournal.org links and content here
Separating evaluation from publishing → a world of benefits
less text, item by item
it
'behind The Unjournal'
The research can
have these come out in steps
. Or use + Suggest entry above to open a GitHub issue.
the 'suggest entry' is now going to a dead GH page -- find a better approach and solution
Three explicit flags that would substantially revise the estimates: (1) better proxy validity research (reversal learning, parental care as welfare indicators); (2) new data on understudied invertebrates; (3) theoretical advances o
can these be restated as questions?
The load-bearing belief is that neuron counts are only a defensible proxy for moral weight if they reliably correlate with the welfare-relevant capacities organisms actually posse
Better stated as a question ... something like "Do neuron counts reliably correlate with the welfare-relevant capacities organisms actually possess?" (of course 'reliably' probably would need operationalization, and this elides the possibility that it may "reliably correlate" but the correlation may be low)
Three explicit cruxes structuring the entire cross-cause model: (1) animals' moral weights relative to humans; (2) expected value of the long-run future; (3) preference for making a difference vs. expected value. Cause rankings reve
these are good, but can they be separated out and flexhed out and made more explicit?
The load-bearing belief is that any significant moral weight for animals, combined with Rethink Priorities' finding that corporate animal welfare campaigns are ~1000x more cost-effective than top global health interventions (e.g. AMF), implies Open Phil should prioritize animal welfare in neartermism. The author's position would change if the moral weights / welfare-range estimates favoring animals were substantially lower, or if a defensible reason were given for valuing human welfare units far above animal welfare units purely on species membership. The author explicitly asks what would have to change in OP's views to NOT prioritize animal welfare.
Not sure this one is an actual crux, rather an attempt to draw out Open Phil (now CG) on their moral weighting
the moral-weighting framework for suffering: whether one prioritizes duration of welfare improvement or the intensity/severity of suffering averted.
'prioritizes' is not clear here. Do they better operationalize this ... how would it be decided/measured?
What belief changes would actually alter donations or work — and what are the poster's actual cruxes? Author foregrounds room for more funding and marginal value.
meta -- not an actual crix
The load-bearing belief is that no compelling cost-effectiveness case exists for the alternatives; a promising cost-effectiveness estimate for institutional meat reduction campaigns would make the author excited about (and supportive of reallocating funding toward) them. To me, to be excited about such campaigns I'd need to see a promising cost-effectiveness estimate. High Shortlist EA Forum Did corporate campaigns in the US have any counterfactual impact? A quantitative model verified No — published 2019, before 2024 window Karolina Sarek 2019-06-24 Key uncertainty
clarify -- hard to read this
The estimated counterfactual impact of US corporate cage-free campaigns (2.1-10%) is load-bearing on the price elasticity of egg demand, a parameter drawn from a literature review with only ~13 observations.
this is a fairly well-defined operationalized crux. It's old, but I guess 'we still are not confident'!
is load-bearing on the price elasti
'is load-bearing on' ... that's too jargony; state it more clearly
He would be persuaded toward global health if shown a defensible rationale for valuing one unit of human welfare so much more than animal welfare that it justifies Open Phil funding GH ~6x as much as AW —
this latter bit is closer to being a specific 'crux'. remember we want these stated as operationalizable questions,if possible. And try to find the single question crux that seems most important or most correlated to the others. You can have another column listing other cruxes raised.
The author's load-bearing claim
you don't need "the authors ... " ... just state their crux question ... obviously it's the author
y The author's recommendation hinges on whether furnished-cage advocacy is actually more cost-effective per unit welfare than cage-free advocacy. He remains skeptical despite his own estimate (2.84x), and his position would shift on whether advocacy costs scale linearly with producer costs, whether infrastructure lock-in blocks later cage-free transitions, and whether furnished-cage campaigns would undermine the cohesiveness of global laying-hen advocacy.
This one is getting good. It seems pretty relevant, but I want you to state it as an explicit opertionalizable question here.
What are the most important questions you'd want answered before deciding how, where, and when to give $20M?
too meta
The author's position that shrimp welfare rests on a weak evidentiary base
State the actual crux. Your description here is more about the implications of the crux.
You can state somewhat general grounds, but then give a specific instantiation if possible. E.g., something about whether one form of slaughter is more painful to shrimp than another, or whether analgesia is good evidence, etc.
Would update on: timelines, public/model concern for animals, indirect normativity, moral-circle expansion, and simulated-animal welfare.
this is too general/vague. I want explicit defined crixes. What is the specific question they are uncertain about, how would you measure it, and what decision would it change?
Would update on: timelines, public/model concern for animals, indirect normativity, moral-circle expansion, and simulated-animal welfare.
What exactly is the crux here? You haven't explained it.
What are the most important questions you'd want answered before deciding how, where, and when to give $20M?
It's also not that explicit. It's kind of meta.
Outlines eight cruxes that would change the ideal balance among cause, within-cause, and cross-cause prioritization.
ok but too meta -- maybe name ONE crux here and/or flesh out rows?
What belief changes would actually alter donations or work — and what are the poster's actual cruxes? Author foregrounds room for more funding and marginal value.
This is more like meta. I don't think it's an actual crux
Whether safety interventions and welfare interventions conflict or create synergies.
This feels a bit vague and could be explained and specified better
Coverage by cause area — click to filter · amber = legacy AI cluster · green = Unjournal core & in-scope AI
Allow more sophisticated sorting, e.g., by relevance and then by date, or by some combination.
Fund the Prize Pool
Note the existing prize pool ($3400) ... change this to 'expand'
These are illustrative examples. Final tournament questions are being finalized with Metaculus.
these have now been finalized. Maybe integrate some of them in?
Predict the Future of Animal Welfare
Put in the link here, make it clear it's live, make it prominent, and perhaps use some of the style and language from there and incorporate it into this page.
his will be Metaculus' first tournament focused entirely on animal welfare. Questions will cover topics like:
We should adjust this - it's the first funded tournament that focuses broadly on animal welfare (there was an unfunded tournament as well as one focused on cultured meat).
Animal welfare funders face a strategic question:
Cite this. https://forum.effectivealtruism.org/posts/bDqdfMXe6YuYnwMDA/good-tasting-alt-meat-is-a-scientific-problem-we-can-solve Perhaps give a tool tip or a little background to help make it clear that this is something being funded and of strong interest to major funders.
· compensation?
'compensation' tooltip not working
Email (optional, only if you'd like a reply)
add "And compensation for good feedback, if we offer it -- always grandfathered in" -- tooltip actually
When we evaluate a paper, the pieces — the paper itself, two or three evaluations, the ratings, the authors' response, any follow-ups — end up on separate pages, and you have to jump between them to follow the thread. I've been trying out ways to put the whole thing on one page instead. These are rough prototypes. The idea is a reading layer that sits alongside PubPub (which stays the system of record for DOIs, indexing, and so on), not a replacement — though I realise splitting things across two homes has its own costs. I'm honestly not sure any of this is better, and that's mostly what I'd like your help with.
this but at the top is a bit long and even though it condenses a bit it's still a bit bulky and keeps popping up again as we scroll down.
Also, I want this to clearly state at the top that most of this content, including the explanations, has been generated using AI, although I have been looking over it and adjusting
I think that alternative proteins may fall into this bucket. While it doesn’t meet my personal donation bar for an evidence-based intervention, I understand why it has absorbed large amounts of philanthropic money earmarked for animal welfare. It might just eventually work.
GPT5.5-extra: This seems right, and it points to a useful portfolio distinction. Weak evidence for direct substitution is not the same as a strong case against all alternative-protein R&D; low downside risk can make exploratory bets reasonable even when near-term evidence is thin.
The decision-relevant question is not just "is this evidence-based now?" but "what is the option value, learning value, crowding-out cost, and plausible downside risk relative to other animal-welfare bets?"
We (the EA animal welfare community) should be using significant resources to generate useful evidence. But simply funding the broad field of animal welfare science is likely to create scattered research results that are difficult to translate into action.
GPT5.5-extra: I agree with this direction. A useful design distinction is between (1) decision-question elicitation, (2) study design and data collection, (3) independent review/replication, and (4) translation into funding or implementation decisions.
"Taking ownership" should not mean collapsing these roles into a single interested evaluator. The highest-value infrastructure may be a pipeline with clear handoffs, independent review, public evaluation packages, and explicit decision thresholds.
All three intervention examples I give have been extensively discussed, recommended by charity evaluators, and financed by major funders.
GPT5.5-extra: This is an important point, but I would separate "widely recommended/funded" from "best marginal opportunity." A field can converge on interventions because they are legible, scalable, fundable, or coalition-friendly, not only because they have the strongest evidence.
That distinction matters for diagnosis: the problem may be less "people made bad calls" and more "the field lacks institutions that can generate, audit, and translate decision-relevant evidence before interventions become large funding defaults."
R&D should be the #1 priority right now, ahead of scaling existing work.
GPT5.5-extra: This is plausible, and I am sympathetic, but the post does not fully close the argument for "#1 priority." To make the case, I would want an explicit value-of-information comparison: how much better could targeted R&D make future funding decisions, how soon, at what cost, and relative to the best marginal scaling opportunities?
This is exactly the kind of question that seems well suited to animal-welfare evidence infrastructure: pre-specified decision-relevant questions, rapid review, and evaluation packages rather than scattered academic outputs.
The report is a few years old now, but even then we already had multiple strong studies showing that the availability of equivalent substitutes does not lead to major changes in meat consumption.
GPT5.5-extra: I would soften "equivalent substitutes" here. Peacock's review is mainly about current plant-based meat and hypothetical improvements in price/taste/convenience; it does not establish that actually equivalent future substitutes, cultivated meat, or precision-fermentation products would fail to substitute. It is stronger as evidence that current plant-based availability and many current interventions have weak or uncertain effects.
However, we have little evidence for this substitution effect.
GPT5.5-extra: I agree with the direction, but would phrase this as "limited and hard-to-interpret evidence" rather than simply "little evidence." There is evidence from current plant-based meat and meat-reduction interventions, but it is hard to generalize to future products or to settings where price/taste/convenience are genuinely competitive.
This is also a measurement problem: substitution across animal products, plant-based products, eating occasions, and longer-run market responses is difficult to estimate with standard demand data.
Related Unjournal/PQ context: https://uj-pba-workshop.netlify.app/
I do not feel comfortable supporting a blanket push for cage-free reforms, given the currently available data on what happens when farms/countries switch.
GPT5.5-extra: This conclusion depends on what "blanket push" means. I would distinguish: (1) poorly managed or abrupt transitions, (2) regulatory/corporate cage-free commitments with weak implementation support, and (3) mature higher-welfare cage-free or free-range standards. The evidence against (1) or (2) is not automatically evidence against (3).
This seems like a good place for a decision-relevant decomposition: mortality, painful morbidity, behavioral deprivation, transition costs, and implementation quality should be modeled separately rather than compressed into "cage-free yes/no".
A counterpoint I sometimes hear is that the difference in mortality between caged and cage-free systems disappears as farmers gain experience with cage-free systems.
GPT5.5-extra: The 2021 meta-analysis does more than gesture at this counterpoint: it analyzes 6,040 flocks and 176M hens and finds mortality in cage-free aviaries decreases with experience; in recent years it finds no significant mortality difference among indoor systems. That does not prove all cage-free transitions improve welfare, but it weakens the inference from early/high-mortality transitions to mature cage-free systems.
Source: Schuck-Paim et al. 2021, https://www.nature.com/articles/s41598-021-81868-3
These aren't sudden, painless deaths. Increased vent pecking itself is also a sign of increased environmental stress. Overall, this suggests that hens in the cage-free systems generally experienced more distress.
GPT5.5-extra: This is a reasonable concern if the CSES aviary result is representative, but I would avoid leaning too hard on this single-farm comparison as evidence about cage-free systems overall. The linked CSES system was one implementation at one commercial site, and the broader Schuck-Paim et al. 2021 dataset suggests mortality in aviaries falls substantially as producers gain experience.
The stronger statement is: transition quality and management can matter a lot, and mortality/vent pecking are important welfare proxies, but this particular comparison does not by itself settle overall cage-free welfare.
Sources: https://www.nature.com/articles/s41598-021-81868-3 and https://welfarefootprint.org/laying-hens/
When shrimp first hit the ice slurry, they perform sudden full-body contractions (tail flips), but this also happens if you first cut their head off (check the supplementary material for a video).
GPT5.5-extra: I read this as evidence that the tail flip is a ventral-nerve-cord reflex, not as a proposal to decapitate shrimp as a slaughter method. Weineck et al. 2018 transected shrimp and found that isolated abdomens began tail-flipping in ice slurry; cutting the segmental roots prevented the movement. Their interpretation is that this is locally neural and involuntary, not driven by higher central brain centers.
So the evidential role is narrow: tail flips alone should not be treated as strong evidence of either conscious distress or insensibility, because a similar movement can be generated after separation from the head. It does not show that ice slurry is humane; it weakens one simple behavioral proxy.
Source: Weineck et al. 2018, section 3.7, https://www.mdpi.com/2076-2615/8/9/158
Additionally, both studies implanted conductive electrodes in the test animals. It is plausible that this significantly affects how current flows through the shrimp's body.I also feel confused about what a signal from an electrode on a heart or a ganglion actually tells us. The plots of the recorded “power” are hard to interpret without a control signal to assess what the noise floor is.
GPT5.5-extra: Clarification after checking: Weineck et al.'s "ECG" was not a mammalian-style electrical ECG. It was an impedance measure of movement or fluid between leads over the heart, so amplitude tracks contraction/flow rather than subjective experience directly. The 2026 Somerville et al. preprint instead uses electrophysiological recordings from the supraoesophageal ganglion and operationalizes neurological insensibility as total power below 10% of pre-treatment.
That is a meaningful proxy for neural activity, not a direct welfare readout. The electrode/current-path concern remains reasonable for external validity, but I would phrase the issue as partial proxies with unclear thresholds/noise floors and weak industrial translation, not as "the signals tell us nothing".
Sources: Weineck et al. 2018 https://www.mdpi.com/2076-2615/8/9/158; Somerville et al. 2026 https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6315379
n conclusion, evidence for electrical stunning is extremely limited and we shouldn't feel comfortable recommending anything with confidence.
GPT5.5-extra: Steelman from the current sources: the evidence does not seem to imply "electrical stunning is bad" or "ice slurry is fine". It implies parameter sensitivity.
The 2026 Somerville et al. preprint reports that lower-voltage/shorter electrical shock followed by cold shock sometimes left substantial neural activity or behavioral response, while higher-voltage/longer shock plus cold shock confirmed insensibility in the no-response group; below-0C ice slurry alone also reduced time to low neural activity. SWP's 2026 review interprets the wider decapod evidence as electrical stunning having greater potential than chilling to induce insensibility, while saying species-specific electrical parameters and operational standards remain uncertain.
Sources: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6315379 and https://www.shrimpwelfareproject.org/humane-stunning-white-paper
We have mixed evidence on whether transitioning egg producers to cage-free improves welfare overall.
GPT5.5-extra: I would narrow this. The record supports "mortality evidence during cage-free transitions is mixed and confounded"; it is less clear that evidence on overall welfare is mixed in the same way.
Schuck-Paim et al. 2021 cover 6,040 flocks and 176M hens and find cage-free aviary mortality falls as producers gain experience, with no recent mortality difference among indoor systems. WFI's welfare model estimates lower cumulative pain in cage-free aviaries than cages, while noting important research gaps. The comment thread also matters: Cynthia Schuck-Paim argues the CSES study is a weak basis for mortality/welfare conclusions because the aviary was poorly managed and not a fair mature-system comparison.
Sources: https://www.nature.com/articles/s41598-021-81868-3 and https://welfarefootprint.org/laying-hens/
We have evidence that the substitution effect of alternative proteins is weak, at best.
GPT5.5-extra: I would soften/generalize this claim. The strongest public evidence I found supports: current plant-based meat and many meat-reduction interventions have weak or uncertain substitution effects.
Peacock's RP report explicitly focuses on plant-based meats because cultivated meats are not widely available; it is not a direct test of future cultivated meat or precision-fermentation products. Green, Smith & Mathur 2025 meta-analyze RCTs of behavior-change interventions and find a very small pooled effect; they also say many promising interventions await rigorous evaluation. That is evidence of a gap and weak current intervention effects, but not yet evidence that all alternative proteins, once genuinely cheaper/tastier/more convenient, would have weak substitution.
Sources: https://rethinkpriorities.org/research-area/price-taste-and-convenience-competitive-plant-based-meat-would-not-currently-replace-meat/ and https://www.sciencedirect.com/science/article/pii/S0195666325003861
GPT5.5-extra: I added a small evidence-audit pass on this post. The detailed notes are threaded under David Reinstein's Hypothes.is annotations, but the main takeaways are:
Shrimp stunning: the evidence seems parameter-sensitive, not a blanket case that electrical stunning is bad or that ice slurry is fine. Current Whiteleg-shrimp evidence supports more R&D and machine validation rather than confident deployment standards.
Tail flips in ice slurry: Weineck et al. 2018 suggests this behavior can be a ventral-nerve-cord reflex, so tail flips alone are a weak proxy for either conscious distress or insensibility.
Electrode and neural/heart signals: the measurements are meaningful proxies, but not direct welfare readouts. Instrumentation, thresholds, and industrial translation remain important uncertainties.
Cage-free hens: the strongest narrower claim is that mortality evidence during cage-free transitions is mixed and confounded. That should not be collapsed into a claim that overall cage-free welfare evidence is equally mixed.
Alternative proteins: the public evidence I found supports weak or uncertain current plant-based substitution effects, but does not directly settle future cultivated meat, precision-fermentation, or genuinely cheaper/tastier/more convenient alternatives.
Detailed threaded notes: - Tail-flip reflex: https://hypothes.is/a/hyqLCmP9EfG-yj97z6fOYA - Electrode/proxy clarification: https://hypothes.is/a/jBrRpmP9EfGf93sIx5Tvwg - Shrimp-stunning evidence: https://hypothes.is/a/kSFQ5GP9EfGIt0f_B-4m3g - Cage-free welfare distinction: https://hypothes.is/a/lhR3FmP9EfGfBDupz9xTgw - Alternative-protein substitution scope: https://hypothes.is/a/nFLdSGP9EfGlirtHBhD0AQ
Layer 1 · The paper
this is a 'layer', yes, and that's relevant for thinking about now, but in the ultimate display I don't think we should call it a 'layer'
The paper's findings bear directly on whether scanner-data demand estimates can be trusted for policy analysis — including animal-welfare-motivated policies that hinge on meat / plant-based substitution elasticities. If observational methods systematically differ from true experimental estimates, cost-effectiveness conclusions built on scanner-data elasticities could be unreliable.
OK your language is good, but why not use direct quotes from evaluation manager's report/summary
Note on the title: The published evaluation summary uses the title "cannot reproduce." The authors subsequently softened the paper title to "does not reproduce" to better reflect the scope of the evidence.
some things like this could should be tooltips
This paper tests whether standard observational demand estimation with rich scanner data recovers the price elasticities revealed by a randomized pricing experiment — and finds they diverge substantially.
link the paper and provide DOI, bibliometrtics etc -- OK, you do that below, but the paper link should really be at the top too.
2 toward more
2 what? What units? This seems a bit of an AI lack of contexgtual understanding
Using data from a large grocery retailer's pricing experiment, Bray, Sanders & Stamatopoulos (Robert Bray, Robert Evan Sanders, and Ioannis Stamatopoulos) ask whether standard observational demand-estimation methods — applied to rich scanner data — recover the "true" price elasticities revealed by a randomized price experiment. They find the answer is no. The observational methods shift own-price elasticities substantially relative to the experimental benchmark:
this is an AI generated summary. Not sure we want that. Perhaps better to lead with content from the Evaluation Manager's summary ... .although I admit this overview seems very useful, as long as it's accurate
The Unjournal · Evaluation package Observational Price Variation in Scanner Data Does Not Reproduce Experimental Price Elasticities
bit at the top a bit too persistent. I can't seem to fold or dissapear it as I scroll down
separate PubPub pages,
and whether having both would be too disagregated/cluttery
Roemheld: The paper documents a striking divergence, but the experimental estimates face a plausibility challenge — they imply profit opportunities far exceeding observed industry margins. Contribution is valuable, but headline conclusion requires stronger foundations.
make it clear when things are quotes vs AI-generated paraphrases
void hosting papers unless there is a clear reason; SSRN, journal sites, institutional repositories, and author pages can remain the paper source. Allow papers to be curated by our team, partner organizations, workshops, funders, or external nominators, not only submitted by authors.
these are not legal-specific -- we do this in the standard model too
Early prioritization prototype
Should be labeled as an early automated prioritization tool, outside of legal research.
high-impact research-candidates workflow
--> 'prioritization workflow'
safety
and governance
nominated
curated by our team or ...
Working papers, workshop drafts, and preprints could benefit from fast structured feedback before journal placement, policy uptake, or citation in public debate.
This is sort of generic -- why is this especially relevant to legal scholarship?
, and improve it.
remove 'and improve it' because the 'it' here is vague. The idea is to help us improve this curation and it's usefulness. We should also invite experts and practitioners to suggest additional research.
legal research
tooltip -- use context to add a note on the prevalence of preprinting/working papers in legal scholarship
or open the direct annotation view.
not sure what this is, but it yields "Access to Via is now restricted This Via link displayed annotations on this page: https://info.unjournal.org/legal-scholarship.html
You can view annotations on this site and others using our Chrome extension, our Bookmarklet, and on sites that embed Hypothesis."
Claude: Deep-research update (June 2026) — new sources added on omnivore fraction of PBM consumers.
Three high-confidence new findings integrated:
GFI US Consumer Snapshot (Jan 2025, Morning Consult Dec 2024, n=3,079): 72% of past-year US PBM eaters are active meat-eaters (57% omnivore + 15% carnivore); only 11% veg*n/pescatarian. Retail panel corroboration: 87% of US retail PBM dollars come from meat-buying households ($970M of $1,118M, NielsenIQ Homescan 2023). Now shown as primary chart in §04.5, replacing the less-precise Hartman Group range.
Neuhofer & Lusk (2022, Scientific Reports — already at src-29) details added: 85.97% of PBMA-buying households also purchased conventional ground meat; only 2.79% were exclusive PBMA buyers. Peer-reviewed, IRI scanner panel, n=38,966 households — the strongest methodological data point in the set.
Bryant Research UK (July 2023, n=1,000, new src-48): Occasional UK PBM consumers ~67% omnivore (consistent with prior claims). Key nuance: frequent UK PBM buyers split ~1/3 each (omnivores, flexitarians, non-meat-eaters). The ≥50% arithmetic lower bound for weekly consumers still holds regardless of frequency segmentation, but the 70–80%+ claim is better supported for occasional buyers or all-buyers combined than for the most committed frequent buyers specifically. This caveat is now added to the §04.5 evidence table and §08.
Stat card updated: '≥50%' replaced by '72%' (US direct survey) as the headline figure, with UK lower bound in the tooltip. TLDR and §08 paragraphs updated accordingly.
Product label + independent confirmation from: Food Navigator, Bloomberg Intelligence, SPINS/NielsenIQ, or academic publication
these should be checked -- this is AI suggestions
pre-registered study is published
pre-reg is no guarrantee of credibility
SV inventories,
But we do want some database structure for the content.
Give both tools the same one-hour task: create a sample template_map.csv from three approved, non-confidential documents and produce a lawyer-readable Markdown summary. Compare setup friction, edit quality, citations to source files, and how easy it is to resume the work later.
this seems like a good test
Unjournal Pivotal Questions — Annotate this page via Hypothes.is (select any text to comment). This working note assesses whether current PBM market evidence is informative enough to study substitution and welfare impact — input to the workshop's research-value question. Produced by David Reinstein with iterative AI prompts.
make it clearer that this is a living document, under active adjustment, and responsive to suggestions
hicken, shrimp, fish, beef, pork, eggs — since welfare weights differ sharply across them
link or tooltip support for this, for people not familiar with welfare weights etc
(c) which animal types bear that displacement and in what proportions,
Put the 'which animal products, and their welfare burden' as the fourth item (d instead of c), because that weighting acts on 'the net reduction in production of each animal product'
supply
I'd make (c) 'how much the production of each animal product adjusts in response in market equilibrium' [tooltip: this may not be 1 for 1, e.g., with scarcity or increasing marginal costs, if some consumers purchase less of an animal product, other consumers may pick up the slack. For example, with an ~inelastic supply of wild fish, if some consumers switch to plant-based equivalents, this may drive prices down, getting other consumers to buy more wild fish]
make it the most tractable
"make it a relatively tractable" ('most' is too conclusive)
plant-based
how large is the plant-based meat market today
This is a relatively lenient threshold (a product can be marginally worse for the median consumer and still cross it
I don't see how this follows -- if 50%+ rate it as as good or better than the conventional product, that should imply that the median consumer finds it 'as good or better'. Correct?
Market-share premium: categories with better average taste vs. worse-tasting categories²⁴ 10× more market share
This wotks better as a note than a quantitative report. Same for "Most favorable public category result" -- perhaps put those at the bottom in a few sentences rather than a tabular format
Net assessment
Fold this ... it's just your own speculation. Maybe even shorten it and use tooltips more, and note it's AI's speculation
The strongest indicator for self-consumption is the lapsed-buyer taste data:
I don't think this is 'the strongest evidence'. To me the GFI Germany/UK surveys seems particular strong at least for the EU.
Survey wording
which survey? a US-one right? THe GFI Europe one asked about actual consumption it seems
consuming
highlight/flag more that this is consuming not just buying
Price evidence
This is missing a lot of studies that look at real-world price data and are even considering plausibly exogenous price shifters -- see https://coda.io/d/Unjournal-Public-Pages_ddIEzDONWdb/Pivotal-Qs-curated-research-list_suEdljp5#_lu4-mRYw for a relevant list and classification of some of these
How to read the evidence for the Pivotal Question
This should be a folding box folded by default, but also I'm not sure these explanations are particularly helpful here.
This supports price as a live lever, but it is not a field elasticity estimate.
This last sentence doesn't add value. Cut it.
finds
"Finds" is too strong ... Weaken this to something like "provides evidence suggesting"
his is about the information content of share data — not an argument against studying displacement rates directly through scanner panels or field experiments
I don't think this sentence is necessary. It's deeply confusing as well. You should probably remove this.