Who Catches You?

Instructor and research guide

The same shock. Different lives. Created by Yoav Duman / יואב דומן

A solo introduction to comparative political economy. Allow 50–70 minutes, including reflection and a short debrief, as an unpiloted planning estimate. The activity connects household protection, wealth and political decision-making. No economics course or prior Piketty reading is required. A two-session version can pause after the household room.

Central puzzle: People can face similar risks and support greater protection, yet receive different protection. How does need become a right, payment, service or unmet demand?

Five learning outcomes

After the activity, students should be able to:

1. Distinguish income, net wealth, liquid reserves and services.
2. Trace how eligibility, employment and unpaid care affect protection against a shock.
3. Explain a Piketty-inspired mechanism while distinguishing an absolute wealth gap from an ownership share.
4. Compare decisive-voter and organization-and-coalition explanations through conditional predictions and controlled comparisons.
5. Use dated country evidence with its population, measurement and causal limits.

Seven stops, one explanatory record

Running the activity

Choose For class and enter a name or class alias for the report. Class always uses Full. The eight reasoning prompts require 15, 20, 25 or 40 words, as shown beside each prompt. These are completion minima, not automated judgments of argument quality. Just for fun offers Core and Full, both with optional writing. Both routes visit all seven stops and retain the matched wealth, voting and assembly comparisons. Core requires a matched household pair and one dated country rule check; further household and country tests are optional. Full also requires tests for two households including a care test, at least three household replays, and all three country rule checks. No timer, student account, required dragging or reflex control is used. Labeled navigation and native controls provide routes through the illustrated setting.

Changing a control explores a result; Record this replay, Keep this experiment, Record the benchmark, and Record the agreement / this round retain evidence. The task strip shows what has been kept and what remains. A Compare kept runs panel identifies a qualifying pair, the changed condition, fixed conditions and outcomes; it remains available after further exploration. Other recorded runs remain in the report. The final My evidence tray brings kept comparisons, country values and the student's country note beside the revision prompt. It does not generate an argument.

In voting, changing both top income and participation together does not satisfy the required pair. In the assembly, keep the policy modules fixed and change only funding, one delegation's pledge, or outsider coordination. Removing coordination also removes the outsider pledge as its programmed consequence; it does not remove voting rights. Changing a module clears commitments and is useful exploration, but does not supply that required replay. A controlled comparison can leave both agreements viable, and an unchanged outcome can be informative.

Support and retained reasoning

Every writing field now presents a direct Your question, a short In your answer checklist and Where to look. A kept-comparison button opens that task's own evidence when available. The first forecast is a starting hunch, not a scored choice; the country comparison explicitly uses observations from two different countries. The existing eight writing tasks and minima remain.

Get a hint opens an orientation question. Ask a guiding question helps diagnose the reasoning problem; Try a different example offers an unsolved analogous problem. No student hint supplies the current voter, selects an experiment setting, gives a winning coalition or provides a complete recording recipe. Automatic help after repeated blocked advances uses the same orientation layer. Worked solutions from earlier editions have been retired; reopening an older help view now displays guidance. Historical help records remain truthful rather than being relabeled. Corrective rule explanations remain after a response, while incorrect voter selections keep the actual median and result hidden until correct identification.

In For class, answer textareas accept typed responses and block clipboard paste and dropped text. A note beside each field explains the rule. Just for fun permits pasted notes. Names, course details and progress-file restoration are unaffected. This input rule is not evidence of independent authorship.

The report records support opened, including requested versus automatic display, type and available settings. Opening records software display, not proof that a student read or relied on it. Older or incomplete help records are described neutrally. Neither missing support records nor a completed route establishes independent authorship; the game does not detect AI use or grade independence.

Drafts autosave. The header’s Aa · Display offers three local fonts and three text sizes; preferences are saved with the attempt and included in the progress file. Tools + help keeps key terms, sources, the guide, instructor studio and saving controls together. The Instructor studio, opened with the established code 6717, provides model notes, report criteria and the printable guide. It does not bypass student tasks.

Drafts autosave. Keep this version optionally records a snapshot; continuing records the response too. At the final stop, assembling the report records both final responses. No additional save click is required to advance. The first saved version is distinct from the first submitted opening baseline, captured when the student first advances from the opening. The final screen returns to that submitted baseline. Later edits do not overwrite these retained records. The readable report distinguishes current writing, retained first versions and the submitted baseline; a blank optional opening response is a valid retained baseline. Detailed older interaction history can be incomplete, particularly in older backups.

Saving, recovery and submission

The opening screen offers Just for fun and For class. Core and Full controls appear only in Fun. The header shows the current mode and route; Mode, route, name & course edits these settings. Switching modes or routes retains choices, comparisons, first answers, revisions and help history. Returning to For class selects Full, checks the work at every earlier stop and lists anything still needed. Existing backups without a route field retain the previous Full scope. The report records mode and route changes. A completed Fun record does not certify class completion.

Use Save / resume at any point. Save now updates the local copy. Save & pause returns to the entry screen, and Resume your investigation restores the same stop. Download progress file (.json) and Restore a progress file remain available throughout and move the screen, drafts, retained evidence, mode and route history, and display settings between browsers. In Class, readable reports, text downloads and printing/PDF stay locked until every Full requirement is met and the learner selects Complete & assemble report at the final stop. Changing required work after completion rechecks the lock. Fun exports remain available as clearly labeled exploration records. The readable report includes the name/alias and optional course or section, writing, model runs and matched comparisons, the prediction before first country reveal, dated case evidence, support records, definitions and sources. Once unlocked, it can be downloaded as text or printed; choose Save as PDF in the browser's print dialog. Use the course's normal submission system. There is no submission server, automatic Canvas hand-in, identity verification or authenticated completion certificate.

Autosave depends on storage in that browser. If a saved attempt cannot be read, the game protects its original stored text and opens recovery instead of replacing it. Download the original saved data preserves that text; restoring a compatible backup or starting fresh requires confirmation. The original download remains available if replacement storage fails. A storage-unavailable message means work may exist only in the current tab: download a backup before closing. Clearing browser data, private browsing or changing devices can remove local progress.

Read the fictional models carefully

The four households are structural examples, not a representative population. Income and reserves are different quantities; reserves already form part of assets. Each shock test resets: paid-work loss, a health bill and extra care hours are separate interruptions. Bundle costs cover this four-household shock cohort, with care priced at an invented two credits per hour; they are not annual national budgets. Full take-up is assumed. Tax incidence, administrative delays and control of shared household resources are not estimated.

Resident reactions are explicitly fictional model responses tied to cash or care results. They are not human testimony or approval of a preferred policy. Care services require other workers; their availability, pay and conditions remain outside the model. Sam's lack of the fictional qualifying contribution record does not mean self-employment always excludes people from real insurance.

The wealth workshop follows two debt-free financial accounts with equal earnings, certain equal returns and no inheritance event, housing, pension entitlement or tax-avoidance estimate. Gross return is 4%; growth and earnings growth are 1%. Students change starting ownership, reinvestment, a levy or saving. The levy leaves the accounts; its public use is not modeled.

The nine-citizen voting laboratory is separate from the households and 100-seat assembly. It fixes incomes and cash-only preferences for a binary 30% tax-and-equal-transfer proposal. Abstainers still pay tax and receive transfers. The original Meltzer–Richard model includes labor-supply responses omitted here; this is not an optimal-tax calculator.

The assembly's delegates, seats, policy conditions, coordination and funding are invented. Outsiders initially lack a shared channel for a binding pledge; other delegations already have one. Building it adds no seats or new policy preferences. Real organizing takes resources and may fail; those costs are not estimated. An unpledged seat is not a predicted No vote. Under the initial eight-credit budget, all three two-module combinations can form majorities through different coalitions. At four credits no affordable one-module platform can secure a majority under the stated positions. This deliberate impasse illustrates a constraint, not a hidden winning click. The assembly does not model electoral-system effects or program design changing future budgets.

Theory inside the play

Piketty: ownership and political rules. Income is a flow; wealth is a stock of assets minus debts. Initial ownership, saving, reinvestment, taxes and institutions shape wealth dynamics. Piketty rejects r > g as a sufficient law of ever-rising inequality [1–2]. The workshop displays both absolute gap in credits and A's share of the two accounts in percent: a rising gap can coexist with a falling ownership share. Students should name the measure they explain. A wage or transfer can protect current cash without equalizing ownership.

Meltzer–Richard: the decisive voter. The decisive voter's income relative to mean income helps determine redistribution under the original model's assumptions [3]. Participation can change the decisive voter without changing the population tax base. The fixed-income proposal illustrates one mechanism, not the full equilibrium or a national prediction.

Organized power and coalitions: political routes to protection. Power-resources, Korpi–Palme and Iversen–Soskice offer related but distinct accounts of organized capacity, program coalitions and institutions [4–5]. The assembly directly illustrates coordination and commitments; it cannot validate those full theories. A majority preference need not implement itself. Later Piketty also incorporates political conflict, so the contrast is not that Piketty ignores power.

Research can challenge an intuitive model. Machtei, Huber and Stephens report observational results in affluent democracies contrary to the simple mean/median prediction [6]. Brady and Bostic challenge parts of the targeting-versus-universalism account [7]. Neither makes every voter irrelevant or every design equivalent. Coverage, depth, financing, care and durability remain different questions. Esping-Andersen's state–market–family framework organizes the inquiry without permanent country boxes; Mares and Carnes help situate social policy beyond affluent democracies [17].

Real cases: three countries, one observation year

The worktable uses 2024 estimates published in the World Inequality Report 2026, with the population unit per capita, full population [8–9]. They are not model outputs.

Income is after pensions and unemployment insurance, before personal income tax: neither pure market income nor fully disposable household income. Net wealth deducts debts; a negative bottom-half share means aggregate debts exceed assets in that group. Income and wealth rank people separately. The top 1% is included within the top 10%; do not add their shares or subtract wealth shares from income shares to calculate redistribution. Pension entitlements, housing, debt and services complicate private-wealth comparisons.

The table describes patterns of concentration. It contains no controlled before/after redistribution comparison or measures that identify effects of unions, parties, voting or electoral rules. Students identify additional evidence needed to distinguish political explanations; the worktable cannot test them alone. These are selected cases, not a representative world sample, and the fictional bundles do not stand for countries.

United States: the tax-year-2024 federal Earned Income Tax Credit example concerns River, single, age 23, without a qualifying child. Low wages alone do not satisfy the ordinary childless age rule; this says nothing about access to other programs [11].

Sweden: 2024 parental benefit combines 480 days per child in total, with income-related days and basic/flat amounts subject to eligibility and ceilings. Broad reach can coexist with earnings-related provision [12].

Brazil: the original 2023 Bolsa Família law provides the rules for this 2024 snapshot. Household income per person, registration, supplements and protection rules matter; formal employment is not automatic exclusion. Later rules differ [13].

These are program examples, not portraits of whole welfare states. Their different currencies and program units cannot be ranked directly as generosity scores.

Assess the existing report

Look for an accurate measure, a mechanism, relevant evidence, a fair rival, revision and a limit. Read the final explanation and Port Azure recommendation with the retained comparisons. Correct rule checks and word counts indicate completion; they do not assess meaning, causal reasoning or writing quality.

An optional instructor rubric can score four criteria, 0–2 each: concepts, mechanism, bounded comparison and precise evidence, and revision with limits. Apply it to the work already requested; no additional exit response is needed. A thoughtful recommendation for targeted support can earn the same score as one for broad provision. No policy preference, tax rate, household prosperity or number of selected benefits earns an automatic grade. This rubric has no validated scoring reliability.

The activity already asks students to revise a claim, use a model observation and a real observation, consider a rival, and name a blocker, tradeoff and observation that would change their Port Azure recommendation. A short debrief can use those responses rather than add writing.

Development and review evidence

Original development used six AI working groups covering disciplinary content, comparative cases, game/narrative design, teaching and art/UX. This revision then received six fresh independent blind AI reviews of a frozen candidate: three simulated student perspectives focused on laptop use, phone use and engagement, and three faculty perspectives covered political economy, welfare sociology/care/Global South, and learning/instructional design. Their initial judgments were made without earlier verdicts or intended repairs. These were AI perspectives, not recruited students or academics.

Disciplinary source checks and read-only source/Node-VM acceptance checks examined measures, arithmetic, comparison gates, saved-work protection, retained reasoning, support records and completion feedback. Actual browser QA is a separate evidence category: the observed class walkthrough, 320/390-pixel viewport checks, saved-progress reopening and export-test limitations are recorded separately from the blind reviews and VM checks. The latest shared-controls checks observed mode switching with a retained draft, display preferences and paused state after reload, resume, the instructor code flow, and a compatible progress-file restore through the browser. Browser download capture still failed, while direct serialization/import checks passed. Standalone file-protocol operation and print/PDF output are not claimed as browser-verified. Neither form of technical QA establishes learning, timing, accessibility conformance or transfer. No human classroom pilot or measured learning gain is claimed; 50–70 minutes remains a planning estimate.

The exact shared ChatGPT URL could not be retrieved. Related current portfolio records WCY-01–04 informed the earlier revision; they were not treated as the contents of that unavailable conversation.

Question-clarity follow-up

Four fresh simulated ordinary-student perspectives read a frozen version in route order: first-year, phone, question-first skimmer and careful reader. Each logged the first confusing phrase, its initial interpretation, the nearby clarification sought and whether that resolved it. Initial logs were independent; subsequent hint audits and two acceptance checks were separate, nonblind phases. They were source/VM simulations, not recruited human students.

Repairs clarified recorded household tests, the contribution-specific Sam note, benefit versus job registration in Brazil, outsider delegates, pledge removal, later program funding/support and Port Azure's unspecified eligibility/ownership. The revised Brazil evidence was checked against the original Law 14,601, articles 5–6. Browser QA separately observed typed class input, blocked class paste, allowed fun-mode paste, restored progress, three hint tiers, concealed results after a wrong voter choice, correct-choice confirmation, and opening a kept comparison from its question. A 390-pixel viewport showed the new question layout without horizontal page overflow. These checks do not establish a classroom learning effect.

A short reading ladder

Students can begin with the in-game summaries, dated cases and definitions; no full book is required. Instructors can follow Piketty's clarification [1], Meltzer–Richard [3] and the contemporary reassessment [6], then the program/coalition debate [4–5, 7]. Gethin, Martínez-Toledano and Piketty extend the inquiry beyond a single voter-income axis [10], and Cavaillé distinguishes fairness judgments [16]. Plenter's and Nonnemacher's undergraduate syllabi informed the sequence; this was a purposive check, not a census or ranking of scholars [14–15].

References

1. Piketty, T. (2015). Putting Distribution Back at the Center of Economics. Journal of Economic Perspectives 29(1):67–88. https://www.aeaweb.org/articles?id=10.1257/jep.29.1.67
2. Piketty, T. (2021). Capital and ideology: A global perspective on inequality regimes. British Journal of Sociology 72(1):139–150. https://onlinelibrary.wiley.com/doi/10.1111/1468-4446.12836
3. Meltzer, A. H., and Richard, S. F. (1981). A Rational Theory of the Size of Government. Journal of Political Economy 89(5):914–927. Institutional full-text copy: https://www.haverford.edu/sites/default/files/MeltzerRichard-RationalTheory.pdf
4. Korpi, W., and Palme, J. (1998). The Paradox of Redistribution and Strategies of Equality. American Sociological Review 63(5):661–687. Author working version: https://www.lisdatacenter.org/wps/liswps/174.pdf
5. Iversen, T., and Soskice, D. (2006). Electoral Institutions and the Politics of Coalitions. American Political Science Review 100(2):165–181. https://web.stanford.edu/~jrodden/oslo/iversen_soskice.pdf
6. Machtei, I., Huber, E., and Stephens, J. D. (2025). Goodbye to Meltzer-Richard: Testing major theories of redistribution. Journal of European Social Policy 35(3):233–247. https://journals.sagepub.com/doi/10.1177/09589287241311117
7. Brady, D., and Bostic, A. (2015). Paradoxes of Social Policy: Welfare Transfers, Relative Poverty, and Redistribution Preferences. American Sociological Review 80(2):268–298. https://journals.sagepub.com/doi/10.1177/0003122415573049
8. World Inequality Lab. World Inequality Report 2026, country-appendix data, 2024 observations, country_sheets_table1_in_excel.xlsx. Official download portal: https://wir2026.wid.world/methodology/ ; archive: https://wir2026.wid.world/www-site/uploads/2025/11/WIR_2026_country_sheets.zip
9. Andreescu, M., et al. (2025). World Inequality Report 2026: Technical Notes for Figures and Tables. https://wir2026.wid.world/www-site/uploads/2025/12/WIR26_TechnicalReport.pdf
10. Gethin, A., Martínez-Toledano, C., and Piketty, T. (2022). Brahmin Left Versus Merchant Right. Quarterly Journal of Economics 137(1):1–48. Author advance version: https://amory-gethin.fr/files/pdf/GethinMartinezPiketty2022QJE.pdf
11. IRS. Publication 596 for tax year 2024, Rule 11. https://www.irs.gov/pub/irs-prior/p596--2024.pdf ; credit ceilings in Revenue Procedure 2023-34, section 3.06: https://www.irs.gov/irb/2023-48_IRB
12. Swedish Social Insurance Agency. Social Insurance in Figures 2025, regulations in 2024, pp. 20 and 25. https://www.forsakringskassan.se/download/18.242206fc198acf662d68d/1757339678327/social-insurance-in-figures-2025.pdf
13. Brazil. Law 14,601 (2023), original publication, articles 5–8 and 10–11. https://www2.camara.leg.br/legin/fed/lei/2023/lei-14601-19-junho-2023-794341-publicacaooriginal-168138-pl.html
14. Plenter, J. Introduction to Comparative Welfare State Analysis, Summer 2025. https://www.johannaplenter.de/documents/Syllabus_Plenter_ComparativeWelfareStateAnalysis_SoSe25.pdf
15. Nonnemacher, J. Comparative Political Economy, Fall 2024. https://www.jeffreynonnemacher.com/uploads/1/3/3/1/133154167/fall24_psci3092_001_syllabus.pdf
16. Cavaillé, C. Fair Enough? (2023), Chapter 2. https://www.cambridge.org/core/books/abs/fair-enough/what-is-fair/4FCC75112C24E461D339FCABA36D6A84
17. Mares, I., and Carnes, M. (2009). Social Policy in Developing Countries. https://www.annualreviews.org/content/journals/10.1146/annurev.polisci.12.071207.093504