AI analysis requested by The Unjournal

AI-generated analysis requested by The Unjournal, not an Unjournal evaluation. No person has checked this code or its numbers. Claude is made by Anthropic, which publishes the paper (conflict note).

Who asked for this, and how it was made

David Reinstein (co-director, The Unjournal) asked on 29 October 2026, after reading the paper’s literature section (pp. 5–6), whether an equally reasonable set of modelling choices would have led to very different results. At his request Claude (Anthropic’s Claude Opus 5.5, working in Claude Code) drafted the menu below and ran one pilot change the same day. The menu lists Claude’s suggestions, drawn from the paper’s own citations where possible, for evaluators to amend.

The pilot builds on Econ-ARK’s independent reproduction of the model (MIT licence). It is separate from the isitcredible review and the earlier AI assessments, and cites their results only where it says so.

Structural sensitivity: would other reasonable model choices change the results?

The paper builds its model from a series of choices, each defensible and each drawn from a literature: a task-based production structure, a capital-supply schedule, search and matching, a wage rule. Would an equally reasonable set of choices have led to very different results? This page lists the main choice points with alternatives, for evaluators to amend, and reports one pilot change.

In brief

What earlier work tested. The isitcredible review varied the paper’s parameters jointly and swapped a few smaller pieces of structure: the routing of labor-force exits, within-task substitution, and how the extra capital is paid for. It did not swap the main building blocks: the capital-supply rule, the wage rule, the matching function, or tasks within occupations.
The menu. 11 choice points, each with the paper’s choice and page, alternatives, the claims it could change, and what has been tested. Comments from evaluators will guide which to run next; the form is private.
The pilot. Capital reaches the paper’s supply schedule gradually rather than instantly. In the extreme scenario, the long-run elasticity of about 6 with a two-year half-life gives nearly the paper’s 2030 GDP (32.9 against 32.4 percent) but a different split: the cognitive wage is −17.0 percent against the paper’s −11.5, and cognitive unemployment 15.7 against 17.9 percent.
Why it matters. The two choices interact. Because the paper’s wage rule makes the cognitive wage sticky relative to the economy-wide wage, slower capital moves the cost from unemployment into lower pay. A different wage rule would likely move it the other way. Choices should be varied together, not one at a time.

The menu: choice points and alternatives

A draft for evaluators to amend: which alternatives are as reasonable as the paper’s choice, which are not, what is missing, and which values are defensible. Download it as plain text to comment offline.

Choice point	Block	Alternatives	Claims it could change	Effort
Capital supply: how elastic, and how fast	Production and capital	4	K1, K2, K4, K5	Pilot done
Wage setting	Labor market	4	K3, K4, K6	Hours for each
Matching function	Labor market	1	K3, K6	Hours
Moving between occupation groups	Labor market	3	K3, K6, K8	Hours for directed search (partly coded in the review); a day for the others
Who enters and leaves the labor force	Labor market	2	K3, K4, K6	Hours, to combine with the other choices
Worker heterogeneity and number of groups	Labor market	3	K3, K5	A day for different pay; days for more groups
Tasks within occupations	Production and capital	2	K2, K3, K4	A research prototype: days or more, and the result would be stylized
Substitution within a task	Production and capital	1	K2, K4, K5	A day, to run it through the full monthly model
New tasks for people	Production and capital	3	K1, K2, K3	Hours for a lag; a day for the others
Innovation	Production and capital	2	K1	Hours
Shape of the capability path	Scenario inputs	1	K1, K3	Hours

1. Capital supply: how elastic, and how fast

The paper’s choice. Capital is supplied instantly along a constant-elasticity schedule, Δln K = ε Δln r (Equation 6, p. 9), the reduced form of Moll, Rachel and Restrepo’s (2022) long-run wealth schedule. ε = 3 is “higher than five years of saving at historical rates would imply”, because the capital doing automated cognitive work is mostly compute, “financed in a world market and built in a year or two” (p. 26). Table 5 varies ε (p. 37).

Alternatives.

Partial adjustment: the stock closes part of its gap to the long-run schedule each month, with the long-run ε of about 6 that the paper derives from Moll et al. (p. 9, footnote). Piloted below.
Time to build: investment is committed months before the capital can be used (Kydland and Prescott 1982).
Accumulation from saving, at a fixed saving rate or by forward-looking households (an Euler equation), instead of a supply schedule.
Separate compute from other capital, each with its own supply.

Claims it could change. K1: Extreme scenario: by 2030 GDP is 32 percent above its no-AI path, growing 15 percent a year; K2: Extreme scenario: the labor share falls from 60 to 45 percent; K4: Extreme scenario: the cognitive wage is 11.5 percent below its no-AI path; other wages are 34 percent above; K5: The aggregate gain is almost three times the cognitive loss; a transfer of about 9 percent of GDP would compensate.

Already tested. Table 5 varies ε. The isitcredible review charged for the extra capital under several financing assumptions. No dynamic adjustment before this pilot.

Effort on the Econ-ARK code. Pilot done. Time to build: hours. Saving-based accumulation: days, and a different solution method. Two capital types: a day or more.

2. Wage setting

The paper’s choice. The cognitive wage closes part of its gap to the market-clearing wage each month (Equation 30, p. 19, after Blanchard and Galí 2007). “What is rigid is the cognitive discount”: the rigidity is relative to the economy-wide wage. Other occupations’ wages clear. Layoffs are job rationing (Michaillat 2012; p. 19). Table 6 varies the rigidity ξ.

Alternatives.

Rigidity in the level of the cognitive wage (against its own no-AI path) rather than in its discount.
Rigidity downward only.
A wage norm or bargained wage with a slow-moving outside option (Hall 2005).
Rigid wages in both groups.

Claims it could change. K3: Extreme scenario: nearly one in five cognitive workers is unemployed; K4: Extreme scenario: the cognitive wage is 11.5 percent below its no-AI path; other wages are 34 percent above; K6: Substantial scenario: cognitive unemployment rises from 2.9 to 4.5 percent; GDP is 8.3 percent higher.

Already tested. ξ one at a time (Table 6) and jointly with the paper’s other values: the isitcredible review found wage rigidity accounts for 0.56 of the variance of extreme cognitive unemployment over the paper’s own values. The rule’s form was not varied. The pilot below shows it matters for how other choices play out.

Effort on the Econ-ARK code. Hours for each.

3. Matching function

The paper’s choice. Hiring follows the den Haan, Ramey and Watson (2000) form with curvature ι = 1.27 (p. 20; Table 1, p. 24); matching efficiency is set to hit the normal-times job-filling rate of 0.65.

Alternatives.

Cobb–Douglas matching with an elasticity near 0.5 (Petrongolo and Pissarides 2001), recalibrated to the same normal-times targets.

Claims it could change. K3: Extreme scenario: nearly one in five cognitive workers is unemployed; K6: Substantial scenario: cognitive unemployment rises from 2.9 to 4.5 percent; GDP is 8.3 percent higher.

Already tested. Not varied; the curvature is held at 1.27.

Effort on the Econ-ARK code. Hours. Probably small at these targets, but unchecked.

4. Moving between occupation groups

The paper’s choice. A single discount μ on job search across the two groups (p. 19, citing Kambourov and Manovskii 2009): 0.17 in normal times, fitted to switching data, and 0.08 and 0.04 in the substantial and extreme scenarios, which have no outside anchor (p. 28).

Alternatives.

Search directed toward the group whose wages are rising.
Mobility that differs across cognitive occupations.
A one-off retraining cost or delay instead of a permanent discount.

Already tested. μ jointly with the paper’s other values; illustrative, uncalibrated worker responses to wage gaps (isitcredible review).

Effort on the Econ-ARK code. Hours for directed search (partly coded in the review); a day for the others.

5. Who enters and leaves the labor force

The paper’s choice. All separations, including exits from the labor force, pass through the unemployment pool; entrants join the groups and the pool in proportion to their sizes (pp. 20–21).

Alternatives.

Exits leave the labor force directly.
Entrants who replace cognitive workers join the growing group instead.

Already tested. The isitcredible review built a family of routings: 2030 cognitive unemployment of 2.0–4.5 percent in the substantial scenario and 13.9–17.6 in the extreme, one change at a time.

Effort on the Econ-ARK code. Hours, to combine with the other choices.

6. Worker heterogeneity and number of groups

The paper’s choice. Two groups of occupations; “workers are homogeneous in production” (p. 10). The cognitive group’s share of the wage bill is set equal to its employment share, 0.624.

Alternatives.

Different pay levels: measured, the cognitive share of the wage bill is about 0.69.
Three or more groups that differ in exposure and mobility.
Displacement costs that differ by age or tenure.

Claims it could change. K3: Extreme scenario: nearly one in five cognitive workers is unemployed; K5: The aggregate gain is almost three times the cognitive loss; a transfer of about 9 percent of GDP would compensate.

Already tested. The isitcredible review varied both shares together; a version with different pay was not built.

Effort on the Econ-ARK code. A day for different pay; days for more groups.

7. Tasks within occupations

The paper’s choice. Workers within an occupation are treated alike across its tasks. The authors note a literature in which they differ (Autor and Thompson 2025; Freund and Mann 2026; Althoff and Reichardt 2026) and write: “We do not currently incorporate this richness” (p. 6).

Alternatives.

Expertise tiers: as the paper summarizes Autor and Thompson (2025), automating an occupation’s inexpert tasks raises the expertise and wage of the job that remains, and automating its expert tasks lowers both.
Standardization that lowers wages without lowering employment, as the paper summarizes Fukui, Nakamura and Steinsson (2026).

Claims it could change. K2: Extreme scenario: the labor share falls from 60 to 45 percent; K3: Extreme scenario: nearly one in five cognitive workers is unemployed; K4: Extreme scenario: the cognitive wage is 11.5 percent below its no-AI path; other wages are 34 percent above.

Already tested. Not tested.

Effort on the Econ-ARK code. A research prototype: days or more, and the result would be stylized. It could change the sign of the cognitive wage result.

8. Substitution within a task

The paper’s choice. An AI-performed and a human-performed instance of the same task substitute at the same elasticity, σ = 0.5, as tasks do across the economy (p. 8).

Alternatives.

A nested production function with a separate within-task elasticity.

Claims it could change. K2: Extreme scenario: the labor share falls from 60 to 45 percent; K4: Extreme scenario: the cognitive wage is 11.5 percent below its no-AI path; other wages are 34 percent above; K5: The aggregate gain is almost three times the cognitive loss; a transfer of about 9 percent of GDP would compensate.

Already tested. The isitcredible review built a nested version of the full-employment economy: the extreme 2030 labor share is 44.6 percent at a within-task elasticity of 0.5, 42.4 at 1 and 37.4 at 3. Unemployment effects were not simulated.

Effort on the Econ-ARK code. A day, to run it through the full monthly model.

9. New tasks for people

The paper’s choice. A constant reinstatement ratio ρ (0.5, 0.25 and 0 across the scenarios): new tasks belong to cognitive occupations and enter at the same unit cost (p. 11). The modest value follows Acemoglu and Restrepo (2019); the other two have no outside anchor (p. 28).

Alternatives.

New tasks that arrive with a lag.
New tasks in other occupations.
A reinstatement rate that rises or falls with AI capability.

Already tested. ρ over the paper’s values (the second-largest driver of extreme cognitive unemployment, isitcredible review).

Effort on the Econ-ARK code. Hours for a lag; a day for the others.

10. Innovation

The paper’s choice. A semi-endogenous ideas stock (Romer 1990; Jones 1995), with parameters from Bloom, Jones, Van Reenen and Webb (2020) (p. 16; Appendix C).

Alternatives.

AI raises research productivity directly.
Ideas raise the productivity of capital as well as labor.

Claims it could change. K1: Extreme scenario: by 2030 GDP is 32 percent above its no-AI path, growing 15 percent a year.

Already tested. Across the block’s parameters, extreme 2030 GDP moves only between 32.1 and 32.4 percent; letting ideas raise capital productivity has small effects by 2030 (isitcredible review).

Effort on the Econ-ARK code. Hours. Low priority for the 2030 claims; more relevant beyond 2030.

11. Shape of the capability path

The paper’s choice. Capability and adoption follow logistic paths from mid-2026 anchors to each scenario’s 2030 values; the automation share and reinstatement are held constant (p. 11).

Alternatives.

The same 2030 endpoints, reached gradually or in a late jump.

Claims it could change. K1: Extreme scenario: by 2030 GDP is 32 percent above its no-AI path, growing 15 percent a year; K3: Extreme scenario: nearly one in five cognitive workers is unemployed.

Already tested. The isitcredible review added one steeper path for its compensation figures.

Effort on the Econ-ARK code. Hours. This changes the scenario inputs rather than the model; included because an earlier AI report suggested it.

The claims each alternative is checked against

Fixed before any run, so that results cannot pick the claims after the fact. Each is a sentence from the paper.

K1. Extreme scenario: by 2030 GDP is 32 percent above its no-AI path, growing 15 percent a year (abstract; p. 38)
K2. Extreme scenario: the labor share falls from 60 to 45 percent (abstract)
K3. Extreme scenario: nearly one in five cognitive workers is unemployed (abstract; p. 38)
K4. Extreme scenario: the cognitive wage is 11.5 percent below its no-AI path; other wages are 34 percent above (p. 4)
K5. The aggregate gain is almost three times the cognitive loss; a transfer of about 9 percent of GDP would compensate (pp. 35, 39)
K6. Substantial scenario: cognitive unemployment rises from 2.9 to 4.5 percent; GDP is 8.3 percent higher (p. 34; Table 3)
K7. Modest scenario: AI adds less than half a point to growth and a tenth of a point to unemployment (abstract)
K8. The median survey respondent's answers are consistent with the substantial scenario (abstract; p. 36)
K9. The mechanics are the same in every scenario: output rises, cognitive employment shrinks, other employment grows (p. 38)

How a full version would work

Each alternative either contains the paper’s choice as a special case or is recalibrated to the same targets: the normal-times labor market, a labor share of 0.60, a capital-output ratio of 3.5, and the mid-2026 anchors. Otherwise a comparison mixes a different structure with a different economy.
With every alternative switched off, the code must reproduce all 226 published numbers that Econ-ARK’s validator checks.
Alternatives are varied jointly with each other and with the paper’s own parameter values, in a full grid or by sampling.
Results are reported claim by claim: the share of model versions in which a sentence still holds, a specification curve for each headline number, and how much of the variation comes from structural choices compared with parameters.
All of it is conditional sensitivity. A range across model versions is not a probability distribution over outcomes.

This follows the logic of specification-curve and multiverse analysis (Simonsohn, Simmons and Nelson 2020; Steegen, Tuerlinckx, Gelman and Vanpaemel 2016), global sensitivity analysis (Saltelli et al. 2008), and model uncertainty in policy evaluation (Brock, Durlauf and West 2003). These references and the ones in the menu that the paper does not cite are Claude’s suggestions and have not been checked against the sources.

Pilot: how fast capital reaches the paper’s schedule

The change. In the paper, the capital stock is always on its supply schedule. In the pilot, each month the stock closes a fixed fraction of its gap to the schedule; the half-life is the time to close half the gap. A half-life of zero is the paper. The paper itself says compute is “built in a year or two” (p. 26), so half-lives of 6–12 months are close to its own description; 24 and 48 months stand for slower physical limits.

Two long-run elasticities. The paper sets ε = 3, half the value of about 6 implied by Moll, Rachel and Restrepo (p. 9, footnote), with the short horizon in mind. Adding a lag to ε = 3 may count that delay twice; the cleaner comparison is ε = 6 reached with a lag. Both are shown. Everything else stays at the paper’s values.

Checks. With the half-life at zero, every monthly series of the 14 printed runs (Tables 3, 5 and 6) equals the unmodified Econ-ARK code bit for bit, and Econ-ARK’s validator finds all 226 published numbers it checks. With an infinite half-life, the stock stays on its no-AI path, and the results equal the paper’s rule at ε = 0 to within 10⁻¹⁵. At zero half-life, the ε = 6 column reproduces the paper’s Table 5.

Extreme scenario, January 2030

GDP, % above no-AI path

Cognitive wage, % vs no-AI path

Cognitive unemployment, %

Average wage, % vs no-AI path

Labor share, %

GDP gain ÷ cognitive wage-bill loss

Each point is one model run; hover over a point for its value. Half-lives are evenly spaced on the axis, not to scale, and each panel has its own vertical scale. The first blue point is the paper’s Table 3; the first orange point is its Table 5 at ε = 6.

Table view: extreme scenario

January 2030	ε 3 (paper)					ε 6 (long-run, Moll et al.)
January 2030	Instant (paper)	6 months	12 months	24 months	48 months	Instant (paper)	6 months	12 months	24 months	48 months
GDP, % above no-AI path	32.4	30.8	28.7	25.3	20.6	37.2	36.4	35.1	32.9	29.4
GDP growth, % a year	15.4	14.8	14.2	13.1	11.4	17.5	17.1	16.7	16.0	15.1
Labor share, %	45.2	44.3	43.6	42.4	40.8	46.9	46.3	45.8	45.0	43.8
Cognitive unemployment, %	17.9	15.8	15.4	15.1	14.8	18.6	16.7	16.1	15.7	15.4
Unemployment, all workers, %	11.9	10.7	10.5	10.3	10.2	12.2	11.2	10.9	10.7	10.5
Cognitive wage, % vs no-AI	−11.5	−19.2	−22.8	−27.4	−33.2	−3.3	−9.9	−13.0	−17.0	−21.9
Other wages, % vs no-AI	33.6	33.3	29.7	23.4	14.7	42.4	43.7	41.9	37.8	31.3
Average wage, % vs no-AI	9.7	4.8	1.2	−4.3	−11.5	18.3	14.9	12.2	8.1	2.3
Net return to capital, %	8.3	8.8	9.2	9.7	10.6	7.5	7.8	8.0	8.4	9.0
Capital stock, % above no-AI path	56.3	51.9	47.5	40.7	31.8	67.1	64.4	61.3	56.2	49.0
GDP gain ÷ cognitive wage-bill loss	2.79	2.36	2.06	1.66	1.22	3.92	3.44	3.11	2.67	2.15

Substantial scenario: charts and table

GDP, % above no-AI path

Cognitive wage, % vs no-AI path

Cognitive unemployment, %

Average wage, % vs no-AI path

Labor share, %

GDP gain ÷ cognitive wage-bill loss

January 2030	ε 3 (paper)					ε 6 (long-run, Moll et al.)
January 2030	Instant (paper)	6 months	12 months	24 months	48 months	Instant (paper)	6 months	12 months	24 months	48 months
GDP, % above no-AI path	8.3	8.0	7.6	7.0	6.0	9.1	8.9	8.7	8.3	7.5
GDP growth, % a year	5.4	5.3	5.2	4.9	4.6	5.7	5.6	5.6	5.4	5.2
Labor share, %	56.1	55.9	55.7	55.3	54.8	56.5	56.4	56.2	56.0	55.6
Cognitive unemployment, %	4.5	4.3	4.2	4.2	4.2	4.6	4.4	4.3	4.2	4.2
Unemployment, all workers, %	4.6	4.4	4.4	4.3	4.3	4.6	4.4	4.4	4.4	4.3
Cognitive wage, % vs no-AI	−0.3	−1.6	−2.4	−3.7	−5.4	1.3	0.3	−0.3	−1.3	−2.7
Other wages, % vs no-AI	5.9	5.7	5.1	3.9	2.1	7.3	7.4	7.0	6.3	4.9
Average wage, % vs no-AI	2.1	1.3	0.6	−0.7	−2.4	3.7	3.1	2.6	1.7	0.3
Net return to capital, %	7.0	7.1	7.2	7.4	7.7	6.8	6.9	6.9	7.1	7.3
Capital stock, % above no-AI path	13.8	13.0	12.0	10.5	8.4	15.7	15.2	14.6	13.5	11.8
GDP gain ÷ cognitive wage-bill loss	4.78	3.93	3.31	2.55	1.81	7.55	6.38	5.50	4.35	3.15

Modest scenario: table

January 2030	ε 3 (paper)					ε 6 (long-run, Moll et al.)
January 2030	Instant (paper)	6 months	12 months	24 months	48 months	Instant (paper)	6 months	12 months	24 months	48 months
GDP, % above no-AI path	1.6	1.6	1.5	1.5	1.4	1.7	1.7	1.7	1.6	1.6
GDP growth, % a year	2.4	2.4	2.4	2.4	2.3	2.4	2.4	2.4	2.4	2.4
Labor share, %	59.4	59.4	59.4	59.3	59.3	59.5	59.5	59.5	59.4	59.4
Cognitive unemployment, %	2.9	2.9	2.9	2.9	2.9	3.0	2.9	2.9	2.9	2.9
Unemployment, all workers, %	3.9	3.9	3.9	3.9	3.9	3.9	3.9	3.9	3.9	3.9
Cognitive wage, % vs no-AI	0.4	0.3	0.2	0.1	−0.2	0.7	0.6	0.5	0.4	0.2
Other wages, % vs no-AI	1.1	1.1	1.1	1.0	0.7	1.4	1.4	1.4	1.3	1.1
Average wage, % vs no-AI	0.7	0.6	0.5	0.4	0.2	0.9	0.9	0.8	0.7	0.6
Net return to capital, %	6.6	6.6	6.6	6.6	6.7	6.5	6.6	6.6	6.6	6.6
Capital stock, % above no-AI path	2.3	2.2	2.1	2.0	1.7	2.6	2.6	2.5	2.4	2.2
GDP gain ÷ cognitive wage-bill loss	14.50	11.17	9.09	6.71	4.52	73.56	39.31	27.02	16.85	9.84

The gain-to-loss ratio is not meaningful here: the cognitive wage-bill loss is close to zero.

What the pilot shows

Same GDP, different split (K1, K3, K4). In the extreme scenario, ε = 6 with a 24-month half-life gives GDP 32.9 percent above the no-AI path against the paper’s 32.4. But the cognitive wage falls 17.0 percent instead of 11.5, the average wage gain is 8.1 percent instead of 9.7, and cognitive unemployment is 15.7 percent instead of 17.9: “nearly one in five” becomes about one in six.
With the paper’s ε and its own build time (K1, K5). At ε = 3 with a 6–12-month half-life, extreme GDP is 28.7–30.8 percent, the average wage gain 1.2–4.8 percent, and the gain is 2.06–2.36 times the cognitive wage-bill loss rather than “almost three times” (2.79). This pairing may double-count the delay, as above.
The labor share moves less (K2). Across the extreme runs it is 40.8–46.9 percent, and 42.4–46.3 at half-lives of 6–24 months, against the paper’s 45.2.
Substantial and modest (K6, K7). Substantial GDP ranges from 6.0 to 9.1 percent and cognitive unemployment from 4.2 to 4.6; the average wage falls below its no-AI path once the half-life reaches 24 months at ε = 3. The modest scenario barely moves. The paper already notes that a less elastic capital supply “can push the average wage below its no-AI path” (p. 38).
Why unemployment falls when capital is slower. The paper’s wage rule makes the cognitive wage sticky relative to the economy-wide wage (Equation 30, p. 19). Slower capital lowers the economy-wide wage, and the cognitive wage follows it down. In the extreme scenario at ε = 3, a 12-month half-life barely changes the cognitive discount (−19.0 to −19.5 log points), lowers the economy-wide wage from +7.0 to −6.1 percent, and cuts cumulative layoffs to 2030 from 6.9 to 5.1 percent of the labor force. Under a wage rule rigid in the level of the cognitive wage (menu item 2), the same lag would be expected to raise unemployment instead. That is untested.

Limits of the pilot

One change among the eleven choice points, at otherwise fixed paper values; not the joint grid a full version needs.
Partial adjustment is a simple stand-in for build lags. The half-lives are illustrative, not estimated; the only anchor used is the paper’s “a year or two”.
The lag applies to all capital, not only compute, and starts with the stock on its no-AI path in January 2024. Within a month, every capital market uses the same short-run schedule, built from last month’s stock.
The code was written by Claude and has not been reviewed by a person. Claude is made by Anthropic, which publishes the paper.

Comment on the menu

Responses go to The Unjournal team only; they are stored privately and not shown on this page. Leave name and email blank to stay anonymous. Annotations made with Hypothes.is (below) are public. If you are a commissioned evaluator and wish to stay anonymous, please use this form, an anonymous Hypothes.is account, or a private Hypothes.is group, rather than a public annotation under your own name.

Leave this empty

Which alternatives should be explored?

1. Capital supply: how elastic, and how fast

2. Wage setting

3. Matching function

4. Moving between occupation groups

5. Who enters and leaves the labor force

6. Worker heterogeneity and number of groups

7. Tasks within occupations

8. Substitution within a task

9. New tasks for people

10. Innovation

11. Shape of the capability path

What is missing or wrong? Alternatives we missed, with sources Defensible values or ranges for any alternative, with sources Which claims matter most for the decisions this paper could inform? Comments on the capital-adjustment pilot

About you (optional)

Role

May we quote your comments?

Name

Code, results and how to reproduce

The pilot changes two functions in Econ-ARK’s independent rebuild of the model (econ-ark/econ-scenarios, commit 87170fa, MIT licence; Zenodo DOI 10.5281/zenodo.22839070). reproduce.sh copies their package, applies the patch, runs the three checks above and the grid, and writes the logs listed here. Files end in .txt so that they open in a browser; drop the suffix before running. Credit for the model rebuild belongs to Econ-ARK; errors in the change are ours.

Prepared 29 October 2026. The working record, with every run, is kept in The Unjournal’s review folder for this paper.

Annotate this page with Hypothes.is

Select a passage to annotate it, or open the Hypothes.is sidebar tab on the right edge. Public annotations can correct a number, challenge an inference or point to evidence.

How annotations may be used: Hypothes.is annotations on this page may be incorporated into The Unjournal’s knowledge base and may be quoted, summarized, or reposted on other Unjournal pages or public channels. Do not include private or confidential information.

Hypothes.is help