There are some signs of leniency too (anonymous evaluators rate about 5 points lower than named ones on the same paper).
You mean signs of leniency for "political reasons". However, this should be caveated too, because people have the option to choose anonymity. We did not run a randomized trial encouraging or requiring anonymity, which would be an interesting thing to do.
Evaluators' 90% intervals are too narrow for the disagreement we see, by a factor of about two. A co-evaluator's best guess lands inside your 90% interval about half the time; if everyone's intervals were realistic it would be about 70%. Nearly a quarter of evaluator pairs give intervals that don't overlap at all.
It's better explained here, but it's still a little bit of an overstatement. You need to define this more clearly. The intervals they gave were not intervals to try to predict the other evaluator. That's not how we phrased it. This is interesting, but you're making it too much of a normative statement. I would soften this and explain more in a tooltip.
A single evaluation carries roughly half signal, half noise on overall and methods; much less on the other criteria. Averaging our usual two evaluations makes the paper-level number noticeably more reliable, but for "advancing knowledge" and "open science" even the average is mostly noise.
This 'difference in differences' loosely speaking, is rather informative. I'd like to follow this up. Initially, this sort of suggests that the open science and advanced knowledge ratings are not giving us much information, or maybe people are not taking them seriously. But I might like to drill down on this more. If we weight in a meaningful way by people's stated level of confidence or what is implied about this from the upper and lower bounds they give, does this persist? My speculation is that some people are just not taking this rating very seriously, while others are doing so rather carefully. If there was some way to capture that, it could be helpful.
In terms of the "advancing knowledge," I see why it might be the case that people are not taking it very seriously, as we sort of explicitly said we were de-emphasizing that to some extent in favor of impact. I don't quite understand what is meant by "averaging makes it more reliable" for this measure, but for other things, it's mostly noise even after averaging. How, if you're averaging it, do you independently know that it's noise? - What's the basis for comparison?
ntraclass correlationThe share of total rating variance that is due to real differences between papers, rather than to evaluators disagreeing about the same paper. 0.5 means half and half.
This tooltip is good, but I still want a more granular explanation with reference to specific formulas and numbers, in a way that one could deeply understand and justify a particular inter-class correlation value, what it means, and under what situations it tells you something meaningful. Not saying that has to be here, but you should have something linked to this that's a really good explanation of it, which is referred to here.
Krippendorff's alpha
We never defined that or linked to what Krippendorff's alpha is, what it means, what justifies it, etc.
much less on the other criteria
Less what? Much less signal or much less noise? I guess you mean much less signal? Even here, a tooltip should explain what metric you're using.
On the overall rating, the median gap is 9.5 points
That seems fairly high, but I need to think more about what to make of this: whether it's so large as to not seem reasonable, whether it suggests people are using the scales differently, etc.
Data: The Unjournal's public ratings export (as of 16 October 2026), one 2026 evaluation added by hand from the published evaluation summary, journal outcomes from OpenAlex and our database, and LLM ratings from the LLM evaluation project.
How many evaluations/evaluation packages were recorded? Did we capture all of them, or are some of the data still missing?
not yet The Unjournal's priorities or decisions.
Italics, not bold, please.
ut has not yet checked every number or sentence.
Tooltip or mention that we're tracking this now with the Hypothesis tool. We should keep track of which parts have been human-considered and adjusted, if we can do so in some way that reports it to readers of this page without being overly cluttered.
numbers mean,
Mention Ssomething about descriptive patterns and relationships too. These could suggest systematic "biases" (evaluators using scales differently, using the ratings in ways that they were not intended, being unfavorable to certain research for idiosyncratic or personal reasons) ... But that will always be hard to tease out without clear experimental variation.
we should change.
"what we should change" is vague without a sense of what the goals are.
realistic
The word "realistic" is a bit vague here.
114 evaluations of 59 papers
Hyperlink to these and to the more raw data/data dashboards/presentations.
Main findings in plain terms The data, and problems in it How evaluators use the scales How much do evaluators agree? Are the 90% intervals realistic? The two journal-tier questions What the overall rating is built from Which questions earn their place How LLM ratings compare Possible changes (AI-generated suggestions) Open questions Sources and how this was made
This is good, but it takes up a lot of vertical space. Maybe have it fold and then also just put it on the side as a sidebar navigation.
In the 2034 AI-lab guidelines question, two and three labs are tied as the most likely answers (22% each). The crowd spreads probability across several outcomes. This is a long-horizon indicator rather than a near-term welfare outcome.
this needs a bit more explanation. Guidelines for what?
t gives 1% to approval of Nueva Pescanova’s Canary Islands octopus farm before June 2027.
that's a separate issue, I don't see the link; fix this
The octopus farm is 'bad' from an AW pov. Plant-based hitting 10% would be good. Tooltip -- what's the current share of PB products, for reference?
he crowd
these should have tooltips mentioneing the number of predictors (number of people predicting)
Can someone trace a claim assessment back to the evaluator’s reasoning, and forward to the test of that concern?
good but tooltips should remind us of things like qhat a 'claim assesment' is
sted checkAccuracy overstates forecasting; predictors measured during regrowth; use earlier data.The claim and the evaluator’s reasoning
Not sure what this 'graph' is meant do, but I see a lot of value in making a 'methodological claim' an object ... e.g., "when using ML approaches to build a prediction model, using predictors that are causally affected by the outcomes leads to overstated precision and recall" ... with a reference to the analytical research explaining and justifying htis issue. Then the 'graph' would involve another object or edge considering whether that critique applies here. The advantages of this include systematic methodological improvements, broader understanding of 'how important certain methodogical critiques (and further methods research) are', better understanding of the domain and moderators of methodological issues
Blocked and forward tests; back-dated and dropped predictors; each with its accuracy.
again, these are multiple checks. Should we be describing and situating each separately?
Accuracy overstates forecasting
"overstates forecasting' is vague, this is a bit too shorthand (tooltip?) And why (or why not) make 'suggested check' it's own obkect
50/100 from each evaluator, with 90% intervals; Evaluation 2’s interval revised from 30–80 to 40–60.
NB, there were also a set of more specific ratings
“Low confidence” in 215 Mha, with reasons. Nothing numerical on the accuracy claim.
ideally IMO such claims would indeed be quantified
measure
what is 'the measure' and what does that object look like? who is it attributed to, etc?
87.9% accuracy under the paper’s test design. Separately, 215 Mha of regeneration potential.
those are 2 claims -- shouldn't they be 2 separate entirels?
a research graph
define and link a 'research graph'. A 'research graph' of what and why? What is it a 'research graph' intended for, and how does it apply to this situation ?
77.8% within the period falls to 71.7% going forward, and accuracy also falls in new regions
I'm not sure "how large that drop is" ... and what to infer from it. There will typically be some drop when going to a new context, because of generalizability, no?
There’s no formal power or equivalence test. Neighbouring pixels aren’t independent, so millions of points aren’t millions of independent observations.
I get this a little, but it's underexplained
Soil characteristics, which the evaluators also listed, stay in every model, including the forward one.
were these claimed to be "leaky"?
Other predictors could carry the same information, so dropping three doesn’t rule out leakage through others.
Right, and I'm suspicious that "If they didn't notice these leaky predictors, they may have neglected others" (and we also may have neglected them
And A2 should forecast about as well as it does within its own period, if nothing else changes over time.
I don't understand this sentence. What is "It's own period"
Drop productivity, burned area and road density (scored on same-period held-out points)
Clarify: I guess this is dropping the "leaky" variables in question from the model and fitting it on the same authors' data. Intuitively, that if this changes the prediction accuracy on 'in time' data so little, this is evidence against the idea that the model accuracy was substantially overstated/overfit.
But I'd still like to see a comparison of the leaky vs non-leaky models on out-of-time data.
By 'predicted on out of time data' I mean the model must be reapplied to predict outcome data not seen by the authors at all (ideally before it even exists) and where there are no 'outcome-contaminated' predictors being used
No change.
no change from what? an identical result to a tenth of a percentage seems worth rechecking.
Forward test: train on 2000–2012, predict 2012–2024, MapBiomas labels
but I'm confused as to how this relates to what the original authors did
Result
'result' for what? accuracy in a 50/50 test set?
Same period, but independent MapBiomas labels instead of the paper’s
but is that outcome similar in kind?
Spatial blocks, whole Neotropics
needs more explanation or tooltip
eproduces well on random points in Colombia
Be clear: is this data outside of what the model was fit on, or just a subset of it?
or an interval for each claim.
Tooltip: Although we'd often prefer this, and will consider eliciting it in future. We're doing this in a more targeted way in our https://info.unjournal.org/pivotal-questions.html project
“Low confidence in 215 Mha” is a judgment about one claim, in the evaluator’s own words.
not really a number -- didn't the evaluators also give quantified uncertainty for this?
Three kinds of number, each with its own uncertainty. None of them converts into another.
We present three things above: research ratings, judgments about the claims, and measured replication results". (the rest should be in a fold.
Share of sites that actually regrow: 32%
it's confusing what the slider is doing here. what is the 'thought experiment' or statistic this is affecting?
What’s real and what’s made up: the model’s hit rates are the ones Fouilloux reports for Colombia: it correctly flags 85.3% of real regrowth points and correctly rejects 92.0% of non-regrowth points
If we're using these rates of precuision and recall, I don't understand what the slider is doing. What are we talking about when we select "Share of sites tha tactly regrow"
the model’s hit rates are the ones Fouilloux reports for Colombia: it correctly flags 85.3% of real regrowth points and correctly rejects 92.0% of non-regrowth points. The share of regrowth in the landscape is the part you choose; the real share isn’t given here. Numbers are expected counts per 1,000 sites.
be a bit more succinct, extra details and discussion in a tooltip here
joint synthesis. I
hyperlink that as well
Recast the model as predicting future outcomes from conditions observed now.
this is underexplained
Low. Probably a lower bound on biophysical potential, because human factors held past regrowth back.
doublecheck, and provide exact quot ein tooltip
predictor timing
what do you mean 'predictor timing'? The leakage issue?
Why a good test on past data can still miss future regrowth more detail A model can classify past records well and still do worse in new places or later years. Land use changes, and nearby points tend to look alike, so a random test point is often a close neighbour of a training point
this needs a better explanation please. "Predicting results in sample" is not helpful or meaningful. But you are saying something else here, I think
(section 4).
anchor that section and link it
I haven’t checked that table myself.
this was AI's comment. AI -- please do check that table, thanks/
with our evaluation summary,
The Unjournal's evaluation summary
Williams et al. (2024)
hyperlink
The paper reports
hyperlink the paper here and a few other prominent places
a random-forest m
tooltip is good, but also give it a hyperlink to a relevant refernce or definition .. wikipedia or otherwise
them
"bears on each of them"
About 10 minutes to read.
what takes 10 minutes? Ths seems better suited for 'about this page' and a tooltip
research graph
put hyperlinks in this tooltip, to relevant discussion/examples of such graphs
three kinds of uncertainty
are these "3 kinds of uncertainty" discussed below? If so, a tooltip should mention this and maybe give a quick list
Anne’s
hyperlink it, and use her surname not first name
Both evaluators rated the paper 50/100 overall
hyperlink eval summary page
joint report
add hyperlink
This page shows how those pieces fit together, as material for Tuesday’s assessment exercise.
hyperlink this
One claim, two evaluations, one replication
Help think of a better title.
AI’s economic impacts and governance are priorities for The Unjournal. We’re looking for research where independent scrutiny could inform decisions about economic growth, jobs, development, power, and policy. This collection includes both economic evidence and governance proposals.
This is a pretty bland explanation. We should make a better case for why this is a priority. You can consider our existing language and discussion in doing this. At the moment, you open the "why" box, and it doesn't really tell you much of why, just some general stuff.
METR's time horizons (with its 2026 update) if capability measurement counts as governance evidence; otherwise the liability set (Weil, Trout, Gans on staged access). → METR time horizons, Overcoming Judgment-Proofness (Weil), When Does Regulation by Insurance Work? (Trout), Staged Access and Liability (Gans)
you are proposing these 'paired evaluations' -- but I don't think we have a sstem or structure for that
Working shortlist
New status? "Selected for Unjournal evaluation"?
AI interpretation of this input
that should be nested under the human comment, and thought about more carefully
Why this page exists. We are considering whether an independent evaluation of this paper would be useful. The synthesis score uses provisional weights, and the human sample is small. It is one input to the decision.
this takes up a bit too much space, and sounds a bit too AI in it's 'not this but that, this is only one input'. Improve it and put more into a tooltip
Human input and prioritization discussion
these 'human inputs' are taking up too much space, need to be better consolidated
Transcript, key equations and sources
I see the transcript, but I don't see the equations.
Worth asking for code Survey and elicitation data; FRI has released data from earlier tournaments, so a request is reasonable.
bold yellow background is too bright
attention research
explain 'attention research'
You made 8 Keep/Maybe/Drop picks or notes here before they were folded into the rating scale. 2 of those papers have no rating yet; the rating panel shows your earlier pick and note so you can convert them.
Does everyone see this, or is this just me-specific?
Rate for evaluation priority
Once I open this sidebar bit, it's hard to close it. I'm not sure how to close it.
October 2026. We are choosing a handful of these papers to commission for public evaluation; nothing is decided yet. Ratings, picks and comments here feed that choice.
Mentioning that we're aiming to decide and move forward before the latter part of October
Why this area. AI governance
It's not just AI governance. I don't think it's accurate to call all the papers below simply "AI governance."
Why this area. AI governance is one of The Unjournal's stated focus areas, but we have published only one evaluation in it so far. Policy is moving fast (pacing proposals, export controls, the EU AI Act, state liability law), and the research feeding these debates is influential yet rarely gets independent expert scrutiny. We focus on governance and the economics of AI policy, not technical AI safety.
This is getting a bit text-heavy. Use folds and tooltips more. ... Remove the "published only one evaluation in it so far." That's a bit negative and not entirely accurate if we think broadly
Rate this paper for evaluation priority
having this only on the rhs is weird -- probably give it full horizontal space
and it overlaps the FRI paper. Note the lab authorship.
overlaps it in what way? I thought that was more about forecswters, and this seems more like a theory model?
Where the paper sits in the working list: shortlist, getting attention now, alternates, found on the dashboard, or older papers kept because they are still in use. Attention
make it easier to click these to sort by them?. And make 'attention' sort hte default.
Candidate papers on AI governance and the economics of AI policy. Working notes, not Unjournal decisions or evaluations.
Give a little more context here. Explain why we're focusing on this and what stage this is at. Make it clear how this relates to the prioritization database too (without cluttering -- the latter cound be a tooltip)
can
"can" is a bit vague as an empirical claim
Latest completed updates Paper scan: Oct 5, 2026 AI scoring: Oct 5, 2026 Crux/PQ matches: Oct 5, 2026 Public feedback aggregate: Oct 5, 2026 Normal cadence Paper discovery, scoring, matching, and dashboard rebuilds normally run Monday and Thursday. Feedback aggregates refresh separately from paper discovery; see the feedback timestamp above. Feedback 25 current ratings across 20 papers. Each person's latest rating per paper counts once; discussion and display names appear only with explicit public-sharing consent. Data context 57 evaluated papers and 129 in the Unjournal pipeline. Earlier reconstructed API cost: at least ~$4.06. Search:
Make it easier to clear all the filters.
▾ Open details, rate or discuss
Make it easier for me to just click on 'rate for prioritization' rather than having to scroll through
(experimental)
not really 'experimental' more like 'preliminary'
Discussion visibility:
Maybe we should make "public visible" the default option, and people can opt in for more confidentiality.
Is the crux description wrong? A better question phrasing or PQ mapping? Tell us here — it saves directly to The Unjournal, no GitHub needed.
Add box and adjust the interface a bit -- encourage people to provide feedback along with their rating of the crux, not just whether it was described incorrectly.
reviewed
What does "reviewed" mean here? Who reviewed it?
Dark mode Reset filters Share URL Quick-rate: ON + Suggest a crux Download CSV Filters (1) Reset Showing 8 of 281 cruxes Quick-rate mode on. On any row, click ▼▼ (strong downvote) · ▼ · ▲ · ▲▲ (strong upvote) to rate how valuable / decision-relevant the crux is. Votes save instantly (anonymous, no account).
Explain the "rate the croaks" bit - and make that a bit more prominent, including the "quick rate" part.
Coverage by cause area — click to filter · amber = legacy AI cluster · green = Unjournal core & in-scope AI
Put the table below. Make it clearer that you can click on each column to sort by that column. make this general across pages like this.
"Date" is this the date the paper/object was last made public, or updated?
If possible, use tooltips to explain what each column means and how it was generated. For example Op/.Clarity it's probably something about the clarity of operationalization, but most people won't understand this.
needs review research paper targeted run: AI governance and the economics of AI policy: priority papers
The yellow color of some of these boxes takes a bit too much attention, and it also overlaps with the color used by the hypothesis tool. Use a more muted color or just gray.
Brookings
Use the column space a bit better in this table. In particular, the second column seems to be a bit too wide given the size of the content. Also, I know that the label "targeted: AI governance blah blah blah" is very long, but that could be split across two lines.
Reasoning behind the criterion ratings
make it clear these are AI generated prioritization ratings -- that shoul dbe made clear everywhere
by setting concrete safety requirements for continued development and deployment, setting a speed limit on capabilities growth, monitoring high-stakes AI R&D experiments and building the option to shut them down if needed, requiring that AI R&D take place in secure and isolated environments (e.g.,air-gapped networks), establishing international incident-sharing, clarifying how deterrence would apply around an intelligence explosion so as to avoid unexpected escalation, securing international agreements to pace progress if needed without fear of falling behind, and developing verification tools in advance for such agreement
They're listing a variety of tools the government or the labs could take on. I say it should close quote. They're suggesting that these will have measurable impacts on reducing the catastrophic risks from AI . They give quite a laundry list here. Evaluators could consider whether these ideas indeed do have strong potential in terms of both feasibility and likely impact. However, given that it's such a laundry list, it makes it hard to do a the intensive expert evaluations were the thing that we considered
Preliminary evidence suggests that this feedback loop could radically accelerate AI progress, overcoming frictions such as diminishing returns to research labor and hard-to-automate tasks.
Here's another definitive claim that could be investigated, at least if the paper here is presenting or making claims about this evidence. However, the word "could" softens this in a sort of vague way.
ome experts think that AI R&D could be fully automated within the next few years.
"some experts think" - is that a pretty vague statement? Are the authors making a claim about their own beliefs or the evidence about this ?
Anthropic reported AI systems completing 26% of internal AI R&D work with only high-level supervision, compared to just 1% five months earlier.
Here's one substantive claim - is this paper, presumably, standing by it. That's something that could be brought under some scrutiny as to whether those numbers are actually meaningful AND CORRECT
Jazz on the Green: Annika SkooghCrediton Arts Centre · CreditonDoors listed at 18:30. Jazz with Sri Lankan food; check the organiser for the set time.
add a link so people can put it on their calendar
Jazz, within reach.
Give this a different name. Try to use my voice more. This looks kind of PR/AI … Just "Live jazz in the Exeter area" is fine. Mention me as the creator a little more prominently and link some of my stuff, including my YouTube and my band listing
NoteLearn: Understanding Uncertainty & Distributions
I think this is too much content for a fold. Turn it into a tooltip as well as a link to a different page that has this fleshed out in detail - maybe that page already exists and it's just a matter of linking it
New to Cultured Meat?
Maybe combine this or align this with the new to this model bit?
NoteNew to this model? Start with the Simplest Model → — a shorter version focusing on some key levers with line-of-sight explanations. You can carry your settings over to this Advanced Model when you’re ready.
This one is short. It could probably just be a tooltip
NoteModel version and September 17 update This page runs the current model (engine 2026-09-17.1). See the September 2026 review and correction log. For comparisons with beliefs recorded at the May 8 workshop, use the hosted workshop-era model or its tagged source. The archived model is preserved as it was deployed and includes errors corrected here. Two supplied research critiques prompted optional median-preserving priors, explicit growth-factor price ranges, and structural comparisons for media use, GF dosage, and maturity dependence. Baseline numerical assumptions are retained. New controls include provenance tooltips and links to the discussion responses, justification, and remaining questions. These additions are AI-implemented scenario tests, pending expert review. NoteNew to this model? Start with the Simplest Model → — a shorter version focusing on some key levers with line-of-sight explanations. You can carry your settings over to this Advanced Model when you’re ready. ►Audio overviews(AI-generated · note)How Cultured Meat is Made ~11 minDownload MP3View/annotate scriptThe Cost Model Explained ~10 minDownload MP3View/annotate script WarningImportant: Model Status & Limitations This model is largely AI-generated and, as of September 2026, has not been independently validated. It is provided to fix ideas, make assumptions inspectable, and support discussion and comparison. Do not treat the outputs as authoritative cost estimates. Several limitations are decision-relevant, not cosmetic: the single latent-maturity factor is ad hoc; source dollars are not yet normalized to one real-dollar year; wet-biomass water content is not standardized; and the model does not separate the probability of reaching commercial scale from cost conditional on reaching it. The sensitivity chart is a dollar-swing ranking, not a variance decomposition. We are addressing these incrementally. Read the full critique and our responses → Limits/Critique We especially welcome expert review of: growth factor quantities and prices, bioreactor CAPEX ranges, and the maturity correlation structure — see below and the Sources section. NoteFrom workshop evidence to model revision The May 8, 2026 workshop is complete. Its outputs now form a linked evidence-to-model workflow: the summary records the technical arguments, the beliefs analysis shows the forecast spread, the demand bridge connects factory-gate cost to an adoption model, and the proposed modeling hack turns the remaining disagreements into testable model changes. Workshop summary → · Public beliefs form → · Cost → demand bridge → · Modeling hack proposal → The named beliefs-analysis dashboard remains an internal review artifact and is intentionally not linked from this public site; a redacted public synthesis should be produced separately. NoteWe Want Your Feedback For substantive or longer-form discussion — please post on 💬 GitHub Discussions. That’s where the conversation can get involved, where others can reply and build on each other, and where everything stays threaded and organized. See the 📖 Discussion Map for where to post what, or jump to a hub: 🧠 Substantive hub — bio / econ / stats / engineering / welfare (the main event) 🎯 PQ framing · 💬 Workshop logistics · 🖥️ Platform & UX For quick inline notes on specific text or a parameter, use Hypothesis (click the < tab on the right edge). For anything beyond a brief highlight, prefer GitHub Discussions so the conversation stays organized and discoverable. 🎧 Listen: Technical Review (22 min MP3) — Audio walkthrough of model architecture and areas for review Other ways to reach us: Open a GitHub issue · Email contact@unjournal.org function toggleHypothesis() { const btn = document.getElementById('toggle-hypothesis') || document.getElementById('vc-hyp'); // Track state via body class (avoids checking Hypothesis's internal display/transform state) const hidden = document.body.classList.contains('hyp-force-hidden'); if (hidden) { document.body.classList.remove('hyp-force-hidden'); if (btn) btn.textContent = '◉ Annotations'; } else { document.body.classList.add('hyp-force-hidden'); if (btn) btn.textContent = '▷ Annotations'; } } function toggleFullWidth() { const body = document.body; const btn = document.getElementById('toggle-fullwidth') || document.getElementById('vc-wide'); // Quarto full-page layout uses column classes and sidebar const contentSelectors = [ 'main.content', 'main', '.page-columns', '#quarto-content', '.column-page', '.column-body', '.column-body-outset', '.panel-fill', '.panel-sidebar', '#quarto-sidebar', '.page-layout-full .content' ]; if (body.classList.contains('fullwidth-mode')) { body.classList.remove('fullwidth-mode'); // Remove injected style const injected = document.getElementById('fullwidth-style'); if (injected) injected.remove(); btn.textContent = 'Expand Content'; } else { body.classList.add('fullwidth-mode'); // Inject a style tag to override Quarto's layout constraints if (!document.getElementById('fullwidth-style')) { const style = document.createElement('style'); style.id = 'fullwidth-style'; style.textContent = ` body.fullwidth-mode #quarto-content, body.fullwidth-mode main.content, body.fullwidth-mode main, body.fullwidth-mode .page-columns, body.fullwidth-mode .column-body, body.fullwidth-mode .column-page { max-width: 100% !important; width: 100% !important; padding-left: 1rem !important; padding-right: 1rem !important; } body.fullwidth-mode #quarto-sidebar, body.fullwidth-mode #quarto-margin-sidebar, body.fullwidth-mode .sidebar { display: none !important; } body.fullwidth-mode .page-columns { grid-template-columns: 1fr !important; } `; document.head.appendChild(style); } btn.textContent = 'Normal Width'; // Also hide hypothesis when going fullwidth hideHypothesis(); } } function toggleParams() { const btn = document.getElementById('toggle-params'); const vcBtn = document.getElementById('vc-params'); const sidebar = document.querySelector('.panel-sidebar'); const fill = document.querySelector('.panel-fill'); if (!sidebar) return; const isHidden = window.getComputedStyle(sidebar).display === 'none'; if (isHidden) { sidebar.classList.remove('params-hidden'); sidebar.style.removeProperty('display'); if (fill) { fill.style.removeProperty('grid-column'); fill.style.removeProperty('width'); } if (btn) btn.textContent = '◀ Hide Parameters (expand charts)'; if (vcBtn) { vcBtn.textContent = '◀ Parameters'; vcBtn.style.background = '#e8f4e8'; vcBtn.style.borderColor = '#5a7a5a'; } } else { sidebar.classList.add('params-hidden'); if (fill) fill.style.gridColumn = '1 / -1'; if (btn) btn.textContent = '▶ Show Parameters'; if (vcBtn) { vcBtn.textContent = '▶ Parameters'; vcBtn.style.background = '#f8f9fa'; vcBtn.style.borderColor = '#ccc'; } } } // Wide sidebar: turn the parameters panel into a fixed-position drawer overlay. // Using position:fixed avoids all CSS grid conflicts — the sidebar lifts out of // the document flow and floats on top. A dimmed backdrop lets users click outside // to close. The OJS cells inside still work because the DOM nodes are the same. function toggleWideSidebar() { const body = document.body; const sidebar = document.querySelector('.panel-sidebar'); const backdrop = document.getElementById('params-backdrop'); const btnV = document.getElementById('vc-wide-params'); const btnI = document.getElementById('toggle-wide-params'); if (body.classList.contains('sidebar-wide')) { // Close drawer body.classList.remove('sidebar-wide'); const s = document.getElementById('sidebar-wide-style'); if (s) s.remove(); if (backdrop) backdrop.style.display = 'none'; if (btnV) { btnV.textContent = '⟺ Wide params'; btnV.style.background = '#f8f9fa'; btnV.style.borderColor = '#ccc'; } if (btnI) btnI.textContent = '⟺ Wide view'; } else { // Open drawer: make sidebar a fixed overlay panel body.classList.add('sidebar-wide'); if (!document.getElementById('sidebar-wide-style')) { const s = document.createElement('style'); s.id = 'sidebar-wide-style'; s.textContent = ` body.sidebar-wide .panel-sidebar { position: fixed !important; top: var(--quarto-navbar-height, 62px) !important; left: 0 !important; width: min(62vw, 820px) !important; max-width: 92vw !important; height: calc(100vh - var(--quarto-navbar-height, 62px)) !important; max-height: calc(100vh - var(--quarto-navbar-height, 62px)) !important; overflow-y: auto !important; overflow-x: hidden !important; z-index: 6000 !important; background: white !important; box-shadow: 8px 0 32px rgba(0,0,0,0.22) !important; padding: 0.5rem 2rem 5rem 1.5rem !important; border-right: 3px solid #3498db !important; scrollbar-width: thin !important; } `; document.head.appendChild(s); } if (backdrop) backdrop.style.display = 'block'; if (btnV) { btnV.textContent = '↘ Narrow params'; btnV.style.background = '#e8f0fa'; btnV.style.borderColor = '#3498db'; } if (btnI) btnI.textContent = '↘ Narrow view'; // Scroll drawer to top when opening if (sidebar) sidebar.scrollTop = 0; } } // Re-initialise Tippy tooltips for abbr[title] and span[title] elements added by OJS. // OJS renders after Quarto's DOMContentLoaded Tippy pass, so elements inside OJS // html`` templates are missed on first pass. Span elements (used for ⓘ inline notes // in the main content area) are also included — native browser title tooltips clip // long text and are unreliable across browsers. function initOjsTooltips() { if (typeof tippy === 'undefined') return; document.querySelectorAll( '.panel-sidebar abbr[title], .panel-fill abbr[title], .panel-sidebar span[title], .panel-fill span[title]' ).forEach(function(el) { const t = el.getAttribute('title'); if (!el._tippy && t) { el.removeAttribute('title'); // prevent duplicate native browser tooltip tippy(el, { content: t, placement: 'bottom', maxWidth: 500, allowHTML: false }); } }); } setTimeout(initOjsTooltips, 2500); // first pass after OJS settles // Re-run whenever sidebar OR main content changes (OJS re-renders on slider change) document.addEventListener('DOMContentLoaded', function() { var obs = new MutationObserver(function() { initOjsTooltips(); }); var sidebar = document.querySelector('.panel-sidebar'); var fill = document.querySelector('.panel-fill'); if (sidebar) obs.observe(sidebar, { childList: true, subtree: true }); if (fill) obs.observe(fill, { childList: true, subtree: true }); }); // Reset all parameters: clear URL state and reload (most reliable approach, // since OJS viewof inputs can't be reset reliably from plain JS) function resetAllDefaults() { window.location.href = window.location.pathname + window.location.hash; } var _expandAllActive = false; var _expandAllObserver = null; function collapseAllDetails() { _expandAllActive = false; if (_expandAllObserver) { _expandAllObserver.disconnect(); _expandAllObserver = null; } document.querySelectorAll('details[open]').forEach(function(d) { d.removeAttribute('open'); }); } function expandAllDetails() { _expandAllActive = true; document.querySelectorAll('details').forEach(function(d) { d.setAttribute('open', ''); }); // OJS re-renders reactive cells after every slider change, replacing <details> elements // with fresh closed ones. The MutationObserver re-applies 'open' to any new ones. if (!_expandAllObserver) { _expandAllObserver = new MutationObserver(function(mutations) { if (!_expandAllActive) return; mutations.forEach(function(m) { m.addedNodes.forEach(function(node) { if (node.nodeType !== 1) return; if (node.tagName === 'DETAILS') node.setAttribute('open', ''); if (node.querySelectorAll) node.querySelectorAll('details').forEach(function(d) { d.setAttribute('open', ''); }); }); }); }); _expandAllObserver.observe(document.body, { childList: true, subtree: true }); } } function toggleViewControls() { const content = document.getElementById('view-controls-content'); const icon = document.getElementById('vc-minimize'); if (!content) return; const isHidden = content.style.display === 'none'; content.style.display = isHidden ? 'flex' : 'none'; if (icon) icon.textContent = isHidden ? '−' : '≡'; } function toggleToc() { const btn = document.getElementById('toggle-toc') || document.getElementById('vc-toc'); // Quarto TOC can be in several locations const selectors = [ '#quarto-margin-sidebar', '#TOC', '.sidebar.toc-left', 'nav.toc', '#quarto-sidebar' ]; let toc = null; for (const sel of selectors) { toc = document.querySelector(sel); if (toc) break; } if (!toc) { btn.textContent = 'TOC not found'; return; } if (toc.style.display === 'none') { toc.style.display = ''; btn.textContent = 'Hide Table of Contents'; } else { toc.style.display = 'none'; btn.textContent = 'Show Table of Contents'; } } function hideHypothesis() { document.body.classList.add('hyp-force-hidden'); const btn = document.getElementById('toggle-hypothesis') || document.getElementById('vc-hyp'); if (btn) btn.textContent = '▷ Annotations'; } View ≡ ◀ Parameters ⟺ Wide params ▷ Contents ▷ Annotations ↔ Expand ▲ Collapse all ▼ Expand all /* ── Fullwidth mode: hide parameter sidebar, let charts fill the page ── */ .fullwidth-mode .page-columns { grid-template-columns: 1fr !important; max-width: 100% !important; width: 100% !important; } .fullwidth-mode #quarto-content, .fullwidth-mode main.content, .fullwidth-mode main, .fullwidth-mode .column-page, .fullwidth-mode .column-body, .fullwidth-mode .panel-fill { max-width: 100% !important; width: 100% !important; padding-left: 1rem !important; padding-right: 1rem !important; } /* Hide parameter sidebar and TOC in fullwidth mode */ .fullwidth-mode .panel-sidebar, .fullwidth-mode #quarto-sidebar, .fullwidth-mode #quarto-margin-sidebar, .fullwidth-mode .sidebar { display: none !important; } /* Hide Hypothesis in fullwidth mode, and when manually toggled off */ .fullwidth-mode .annotator-frame, .fullwidth-mode .hypothesis-sidebar, .fullwidth-mode hypothesis-sidebar, .hyp-force-hidden .annotator-frame, .hyp-force-hidden .hypothesis-sidebar, .hyp-force-hidden hypothesis-sidebar, .hyp-force-hidden iframe[src*="hypothes.is"] { display: none !important; visibility: hidden !important; pointer-events: none !important; } /* Sidebar overflow fix — offset by Quarto navbar so top buttons aren't hidden */ .panel-sidebar { overflow-y: auto !important; overflow-x: hidden !important; position: sticky !important; top: var(--quarto-navbar-height, 62px) !important; max-height: calc(100vh - var(--quarto-navbar-height, 62px)) !important; } .panel-sidebar details[open] { max-width: 100%; overflow-wrap: break-word; } /* When params sidebar is hidden, let charts take full width */ .params-hidden { display: none !important; } NoteNew to Cultured Meat? Read our deep dive: How Cultured Chicken is Made — a detailed guide covering cell banking, bioreactors, media composition, growth factors, and why each step affects costs. Quick summary: Cultured chicken is produced by growing avian muscle cells in bioreactors. The main cost drivers are media (amino acids, nutrients), growth factors (signaling proteins), bioreactors (capital equipment), and operating costs. For a side-by-side comparison of how published TEAs differ in their assumptions and estimates, see our TEA Comparison page. This model stops at a factory-gate cost; for what a given cost implies for consumer adoption and market share, see our Cost → Demand bridge, which connects to Pablo AMC’s demand-side model. CELL BANK → SEED TRAIN → PRODUCTION → HARVEST → PRODUCT [O] [OOO] [OOOOOOO] [===] [≡≡≡]
So many folds at the top. Better than having them unfolded but I feel like there must be a better way of displaying this, perhaps with some horizontal boxes, perhaps some of these should go to external links, etc
The four initial $100 response slots are provisionally filled as of 31 July 2026. Earlier individual commitments still stand1If you were promised an incentive for responding by a particular date, we will hold to that promise. If you have any questions, email contact@unjournal.org. A further $50 is reserved for eligible respondents who return for the follow-up/update round. Please check with us before assuming that a new response is compensated.
Actually make this compensation but a folded box -- not central
. Please check with us before assuming that a new response is compensated.
make this last sentence a tooltip
Once we've gathered this round of independent forecasts, we'll circulate the combined picture and the main points of disagreement, and invite you to revisit and update your estimate. For now, please answer from your own knowledge and reasoning. Naturally, we encourage you to consult your own notes, do a background web search, run calculations, etc.
make this latter part of the paragraph a tooltip
Give your in
"Give (or update)"
How many tonnes of cultivated-chicken biomass will be sold worldwide during calendar year 2036?
Put this one first
A1
"The production-weighted average cost per kg of cultivated-chicken cell biomass referred to in CM_01 A1"
Reasoning or explanation (optional)
leave the pre-fill blank here
Conditional on B,
"B" is not clear here. Better: "Conditional on some firm(s) producing at the 2k metric tonnes scale"...
Include worlds without actual large-scale production. The scale assumption does not imply successful commercialization or price parity.
This isn't a direct quote -- no need for a quote offset. Instead "If you imagine no plants producing at this scale, consider the hypothetical cost at this scale". And then the tooltip on 'hypothetical cost at this scale' should say, "in worlds where there is no production at this scale, you should consider what the cost would be if a firm had chosen to produce at such a scale"
in 2036, at a scale of at least 2,000 tonnes per plant per year? Definition Use actual comparable plant data where available; otherwise estimate a hypothetical industry using the technology and input-supply conditions you expect in 2036. The average is total capital plus operating cost divided by total biomass output, not the cheapest or a representative plant. If hypothetical, specify the imagined production mix in your reasoning. Granting this scale does not assume technology breakthroughs, a mature supply chain, or a particular learning history.
in the tooltip -- how can they use 'actual plant data'? this is for a future world that doesn't exist yet
Only the second sentence is helpful here, and it can be a tooltip on 'production-weighted average'
If hypothetical, specify the imagined production mix in your reasoning. Granting this scale does not assume technology breakthroughs, a mature supply chain, or a particular learning history.
YOu can leave off this last paragraph ... it's confusing and cumbersome
If your cost estimate excludes technically infeasible outcomes, explain that below.
why would an estimate include technically infeasible outcomes? That wouldn't be a good estimate, I think.
Technical feasibility (optional)
this dropdown is unclear
A shorter route into the technical and empirical disagreements behind 2036 cost projections. Start with the five-card overview; open only the cruxes or questions you care about. This page uses public S1 and S3 material, consent-checked beliefs, public-cleared written contributions, and official regulatory sources. No S2 discussion is used.
This could be a tooltip or fold
Here, a crux is not just a general uncertainty. It is a question whose answer would materially change the cost forecast or the chance that the forecasted industry exists.
"not this but that" -- AI tic -- let's improve this
PUBLIC WORKING SYNTHESIS · Forms checked Tue/Fri; new responses require human privacy review · Manual review: 31 July 2026 · AI-assisted synthesisAI drafted and updated the synthesis prose and calculations. The latest human check-in was 31 July 2026, when privacy handling, email follow-ups, and the proposed synthesis update were reviewed and adjusted. Human review does not mean that every claim has been independently verified.
this is too emphasized -- remove the yellow
Public working summary
make this a folding box. Put further details in a tooltip
Better Together? The Impact of Cash Transfers and Couples' Financial Planning on Household Dynamics and Economic Outcomessource titleLiteral title supplied by NBER metadata; not model-generated. Cause: Global health & dev. (LMICs) AI shortlist |Sarika Gupta, Jessica Leight, Daniel W. Maggio |NBER gpt-5.5 (codex headless, xhigh) Human 90/100 · n=1Sorry: this weighting is ad hoc. Quick --, -, ~, +, and ++ ratings map to 10, 30, 50, 70, and 90 and have weight 1. A full rating has weight 2 without detailed discussion; with detailed discussion it has weight 3 for public raters and 4 for team members. “Detailed” currently means at least 120 characters or 20 words. The AI score has weight 3 in the synthesis, in quick-rating equivalents. Each person’s latest rating counts once. Effective human weight for this paper: 1 quick-rating equivalent. Source-provided abstract (not AI-generated): Households' most consequential economic decisions are usually made jointly, yet behavioral interventions are typically t...Households' most consequential economic decisions are usually made jointly, yet behavioral interventions are typically targeted to individuals and optimized to overcome intra-personal constraints. We use a randomized controlled trial to study whether behavioral interventions targeted to couples can overcome interpersonal constraints related to intra-household decision-making. In rural Liberia, we layer a facilitated joint financial planning exercise atop a large unconditional cash transfer, assessing effects relative to cash only and a pure control. Relative to cash only, adding planning increases the amount of the transfer spent on productive investment and consequently improves economic outcomes by 0.16 standard deviation units. However, for households with the highest ex-ante conflict risk, these gains come at the cost of additional intimate partner violence. ▾ Open details, rate or discuss Full abstract, scoring context, and feedback options Quick rate---
to reduce visual clutter and overwhelm, when you "open details" for one paper, can other content be grayed out in some way?
Preferred for detailed ratings
needs more explanation
If possible, give a 0–100 percentile relative to other papers we might evaluate. Otherwise use the five-point scale. Numeric ratings are included in anonymous public aggregates. Team ratings require your name and team email and are checked against the internal roster before they receive the one-assessor workflow shortcut. For a first pass without prior scores or analyses, use the independent-rating view.
public sharing disclaimer here too
for each paper, how strongly should The Unjournal commission an evaluation?
be more clear -- rating this for -- is it potentially impactful and is there evaluation value ... not '"it a good paper"
Targeted intake and prioritization requests - 74 records across 4 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. Use the Targeted intake filter above to include all records, exclude these deliberately oversampled searches, or inspect one family. A targeted label records how a record was found; it is not an endorsement, quality judgment, prevalence estimate, or completed Unjournal prioritization decision.AI impacts on global health and developmentResearch papers (29 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 (4)AI impacts on global health and development: LMIC labor and preparedness: Targeted public-paper search guided by the Coefficient Giving application discussion, manual source and thematic-fit verification, deduplication, and Codex subscription scoring. This pass follows the application discussion's risk-to-response framing: exposure estimates are inputs, not outcomes, and should be assessed alongside actual task content, adoption, infrastructure, institutions, service-trade exposure, and feasible policy responses. It deliberately includes competing estimates and early evidence on BPO and export-linked work, youth and expertise pathways, firm adoption, and frontline health care. LMICs are not treated as one labor market, and inclusion is not endorsement or a completed Unjournal team decision.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.AI impacts on global health and development: 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.AI, global health, and development curation: Manual curation of 25 candidate leads with public-link verification. This curation was preserved on a development branch but was not merged as a live intake batch. The labeled dashboard records were already present through other sources and are marked as surfaced rather than added. Candidate leads without a verified public paper page were not published. Coefficient Giving request for proposals. View 29 records →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 →Empirical conflict replication-game intakeResearch papers (14 papers; 2026-07-24) Recent empirical economics, political science, and related research with conflict or political violence as an outcome. Search-pass details (2)Empirical conflict replication-game intake: 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.Empirical conflict replication-game follow-up: 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 14 records →Soil-invertebrate crux sweepCruxes (6 cruxes; 2026-07-11) Decision-relevant cruxes involving soil invertebrates, soil animals, land use, and indirect animal-welfare effects. Targeted search of public forum posts followed by manual curation. Added after a specific request to look beyond the broad forum scan for soil-invertebrate cruxes. The topic was deliberately oversampled, so these entries should not be interpreted as representative of the overall corpus or as completed Unjournal prioritization decisions. View 6 records →Animal-welfare focused intakeResearch papers (19 papers; 2026-06-22) Public research on animal-welfare economics, food systems, consumer behavior, and production systems. Search-pass details (2)Animal-welfare economics public-paper follow-up: Manual follow-up restricted to independently public papers and DOI records. These papers were added through a requested, topic-focused follow-up rather than the broad recurring scan. Private or unpublished program leads were excluded. The topic was deliberately oversampled, so inclusion is not evidence of prevalence in the wider literature, author endorsement, or a completed Unjournal team decision.Animal-welfare focused intake: Focused OpenAlex search followed by local prioritization. This focused scan added three independently public OpenAlex papers. Unpublished conference-program topics are not named or exposed here. The topic was deliberately oversampled, so inclusion is not evidence of prevalence in the wider literature, author endorsement, or a completed Unjournal team decision. View 19 records → What is this score? (weights for the active lens) Active lens: Evaluation priority. The priority score and default sort are a weighted average of the six sub-scores below (each renormalized if a sub-score is missing; out-of-scope papers are floored).CriterionWeightWhat it capturesPotential decision relevance30%Relevance to important real-world / global-welfare decisions.Methodological potential10%Methodological rigor and how evaluable the analysis is.Real-world influence10%Whether decision-makers are already using or citing the work.Prominence0%Attention the work already has (venue, authors, citations).Timing value20%Early-stage work benefits most from timely feedback.Neglectedness / value of evaluation30%How much independent evaluation would add beyond existing review.Research relevance weights importance, rigor, and relevance for readers looking for useful work. Evaluation priority (default) adds timing and value-of-evaluation (neglectedness) for deciding what The Unjournal should evaluate next. Custom weighting & sorting
these folds are taking up a bit too much vertical space
Quick-rate: ON
probably we should have this be on by default
Search
enable boolean search ... and/or etc
Search: SortOrder papers by overall AI priority score, recency, decision relevance, or other criteria. "Deeper Model First" shows papers scored by the most capable AI models at the top. Priority (Human-AI synthesis) Most Recent First Deeper Model First Potential Decision Relevance Prominence Timing Value Neglectedness Crux/PQ relevance Custom weights Rating sourceChoose the score shown in the circle and used by the default priority sort. Sorry: the current weighting is ad hoc. Quick --, -, ~, +, and ++ ratings map to 10, 30, 50, 70, and 90 and have weight 1. A full rating has weight 2 without detailed discussion; with detailed discussion it has weight 3 for public raters and 4 for team members. “Detailed” currently means at least 120 characters or 20 words. The AI score has weight 3 in the synthesis, in quick-rating equivalents. Each person's latest rating counts once. Papers without human feedback retain their AI score in synthesis mode. Human-AI synthesis AI ratings only Human ratings only Evaluation-relevance weightingON (default): rank by Evaluation priority for The Unjournal — adds timing and value-of-evaluation (neglectedness) on top of relevance and rigor. OFF: rank by Research relevance for readers — importance, rigor, and relevance. Unfold "What is this score?" below to see the exact weights. Cause AreaThe primary global issue this research addresses, aligned with Unjournal's focus areas: global health, development, animal welfare, climate, AI governance, catastrophic risks, and more.
how do I clear all filters? make it easier to do that
· Discussion (Google Doc)
The Google Doc is probably not where we're working anymore? Maybe get rid of this or don't mention it as prominently.
We're continuing the discussion asynchronously and sharing key materials here. This site is evolving into a resource page. Read the workshop summary → · Survey results & synthesis → ·
Don't flag the "Survey Results and Synthesis" link you're giving here - that's a survey about whether people found the workshop useful. That's less interesting for outside participants - we can share that publicly, but it shouldn't be at the top. It shouldn't be prioritized. Instead, you can now link the belief elicitation results https://uj-wellbeing-workshop.netlify.app/beliefs-overview-w9k3m.html and the synthesis of this (is the latter up yet?)
WORKING DRAFT · Internal review only · Not linked from main site
Adjust this - we can now link it from the main site and have a link back as well.
WELLBY measurement value, conversion, and forecasting · March 16 2026 workshop · The Unjournal · Internal draft, last corrected August 3 2026 — estimates visible, for internal review only
Encourage people to look at the subquestions too. !!! I post that more prominently.
WELLBY measurement value, conversion, and forecasting · March 16 2026 workshop · The Unjournal · Internal draft, last corrected August 3 2026 — estimates visible, for internal review only
Link the summary/synthesis here - it's highly connected to this page and these results.
Data-quality warning: The form opened with values already set on some sliders — PQ1A at 70 with its 80% interval pre-set to 40 and 90, PQ3A at 25%, PQ3C at 35% — and it recorded only the final value, not whether a respondent moved anything. These were not neutral midpoints: on a 0–100 scale the form presented what looked like a complete, plausible answer, which is a stronger anchor than a midpoint would have been. Exact-default values may therefore be deliberate answers, anchoring effects, or untouched sliders, and the data cannot distinguish them. Among the 7 responses, one matches all three PQ1A defaults, two match the PQ3A default, and two match the PQ3C default.
Update - I think we've asked people to follow up on this... and they probably should be excluded by default.
June 18 update: Added Miles Kimball's June 17 response. The most visible effects are a higher forecast for GiveWell-linked charity uptake (new max 50%), a higher forecast for agreement among surveyed economists and practitioners (60%), and a much wider DALY/WELLBY conversion range because Kimball gives 1 WELLBY/DALY with a 0.1–10 interval while stressing that his team has not studied that conversion directly.
Make these "updates" folded sections.
at uj-wellbeing-workshop.netlify.app/beliefs
Make these hyperlinks instead of just pasted URL text.
Asked again August 3 which of his values were intended; no answer yet — see above
update -- he has now responded and made it pretty clear that where he gave the default answer, it was probably once he skipped .
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
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"
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