Close this tab to return to your game.

Teacher Guide: Sixty to One — The Two Families

A guide for high school and introductory college instructors assigning this game in a U.S. history, economics, sociology, or civics classroom.

Quick facts: browser-based, no login or install, free to play, runs on a laptop or tablet. Two families — the Freemans (Black) and the Walkers (white) — walk the same six eras of American history from 1865 to 2025, generation by generation, meeting the same decisions and events. A measured reference playthrough (full prologue, every mini-game played, text read at a normal pace) ran about 10–11 minutes; depending on reading pace and how much text a player skips, a run can realistically take anywhere from about 5 to 20 minutes. That comfortably fits inside a single class period with time left for discussion. The game saves and reloads regardless, so a run does not need to fit one class period.

What this game is — and isn't
The audit assignment
The preserved inheritance disagreement
Discussion frameworks
Practical matters: length, content, accessibility
Limitations, written straight

1. What this game is — and isn't

What it is. A simulation built from published economics and history research. Every recurring racial disparity the game applies — a wage ratio, a loan-approval probability, a markup on a contract sale, the size of a typical inheritance, the 60-to-1 starting wealth ratio in 1865 — traces to a cited source in the project's research ledger (RESEARCH.md, 188 statistics drawn from primary sources, independently fact-checked). The two families' fortunes diverge over the course of a run because those sourced disparities push them apart, not because the game scripts an outcome or because one player makes better decisions than the other.

What it isn't. It is not a claim that any real family's outcome is predetermined by race, and it does not say a Black family in 1950 or 2020 was destined to end up poorer than a white neighbor. It models documented mechanisms — laws, lending practices, denied claims, wage scales — and shows what happens when two otherwise-identical households are run through them. The game's own closing text is explicit about this: nothing forces the gap open except the sourced disparities themselves, and the model deliberately follows a single typical family per side rather than surveying the full range of real households.

That distinction is the whole premise, and it's worth stating to students on day one: this is a model of mechanisms, not a verdict on any person.

2. The audit assignment — the centerpiece lesson

The game's citation system is not decoration — it is built so a skeptical player can check the game's homework. That is the assignment: have students audit a claim the game makes, the way a fact-checker would.

What's available to a student who only has the game (Tier 1 — every student)

Almost every screen in the game — an era's opening card, a decision's outcome, the ending's "What drove the gap" section, the About page — carries a small "Source" disclosure with a Learn more link. That link goes to the actual primary or secondary source: a Federal Reserve brief, a Census table, a peer-reviewed journal article, a named investigative report. Nothing in the game cites itself, and nothing cites an AI-generated summary.

What's available if you bring in the repository (Tier 2 — optional, for an advanced class or a whole-class demonstration)

The game's underlying research ledger (RESEARCH.md) and its structured data file (src/lib/data/sources.json) go further than any in-game popover does: every sourced figure there carries an explicit confidence rating — high, medium, or low — based on how solid the underlying research is, plus a conflict flag wherever sources disagree. This detail lives in the codebase, not the deployed game's UI, so treat it as the teacher-provided layer: print or share relevant excerpts, or project the file, rather than expecting students to find it unaided.

The assignment itself (45–60 minutes, pairs or individuals)

Pick three claims. While playing (or from a teacher-selected set of screenshots), each student picks three factual claims the game makes — one from an era's opening narration, one from a decision's outcome text, one from the Ending's "What drove the gap" section.
Follow the link. For each claim, click "Learn more." Confirm where it actually goes: is it a named government agency (Federal Reserve, Census Bureau, Bureau of Justice Statistics), a peer-reviewed journal, or a secondary summary of one? Read enough of the source to find the specific number or finding the game cites. Also note whether the link lands on the exact finding or just an agency's homepage — a handful of citations in the underlying data point to a general agency URL rather than a deep link to the specific table or report. That's a fair, recordable critique in its own right, not a reason to skip the claim.
Grade it. Using a simple four-point rubric — Accurate, Accurate but simplified, Overreaching, Can't verify — students judge whether the game's one-sentence claim represents what the source actually found. (Tier 2 classes: also check the source's confidence rating in sources.json, and confirm that any "medium" or conflict-flagged figure is presented with appropriately hedged language — "widely cited," "estimated," "an open dispute" — rather than false certainty.)
Report and compare. As a class, tally the ratings. Which sources were government or peer-reviewed? Which were journalism or advocacy research? Did anyone find a claim that overreached?

What students should find, if they do it right

The large majority of claims trace to a real, checkable, named primary source — not to the game developers' own assertion.
A nontrivial share of the underlying research is only "qualified," not "confirmed." RESEARCH.md's own accounting: of 188 statistics gathered, 113 were independently confirmed and 75 were qualified (a number that varies by source or couldn't be independently re-verified, though the direction holds) — zero were refuted outright, but zero is also not the same as "everything checked out perfectly." The game does not quietly upgrade a qualified figure to a confirmed one.
Calibration is labeled as calibration. The game distinguishes sourced racial ratios (the wage gap, the loan-denial odds, the inheritance gap) from disclosed calibration — round dollar levels, era income baselines, cost-of-living figures picked so the simulation's overall trajectory matches the real 60:1 → roughly 6:1 historical arc. The About page states this distinction outright; a sharp student should be able to find at least one example of each kind of number in a single playthrough.
Some real disputes are presented as disputes, not resolved. See Section 3.

This is the pedagogy the game's sourcing was built to support: a skeptical reader — student, parent, or administrator — is not asked to take the game's word for anything. They're handed the citation and invited to check it themselves. That's the point of Section 3 below.

3. The preserved inheritance disagreement

This is the section of the guide the owner's review singled out, and it's worth stating plainly to any skeptical parent or administrator who asks whether this game is pushing a political thesis: on one of the most consequential open questions in the field — how much of the Black-white wealth gap owes to inheritance versus housing — the game takes no side, and says so on its own About page.

The disagreement, as it actually stands in the research

One Federal Reserve study (Boston Fed, 2023) estimates that intergenerational transfers explain only about 13–16% of the wealth gap, calling their role "limited."
Other economists (notably Hamilton and Darity, whose work underlies the "baby bonds" reparations proposal) treat inherited wealth transfers as among the largest drivers of the gap.
Home equity is estimated to explain "about half" of the gap by many decompositions — but other estimates put housing's share lower, at 20–30%.

These are not fringe positions talking past each other; they are competing decompositions from credentialed economists using the same underlying wealth data, and the field has not converged.

What the game does about it — nothing, on purpose

The game's data does not encode a "housing explains X% of the gap, inheritance explains Y%" formula anywhere. Instead, it wires in only the pieces of this picture that are uncontested: the raw homeownership rates by race (75.8% white vs. 46.4% Black) and the raw inheritance-receipt rates and amounts (about 30% of white families receive an inheritance or gift, versus about 10% of Black families; conditional medians of roughly $305,000 versus $68,000). Those rates are not in dispute. What's in dispute — how much of the total gap each factor should be credited with causing — is exactly the piece the game leaves alone.

The About page states this directly, in its own words, as the worked example of the game's sourcing philosophy: "Where researchers genuinely disagree, this game says so instead of picking a side or averaging the disagreement away... Both readings are on record here, not smoothed into a single number."

Why this is the section that survives a skeptical reader

A parent or administrator's most natural objection to a game like this is: is this pushing a conclusion, dressed up as data? The honest answer this game can give is: no — and here is the proof. On a question where real economists disagree, the game doesn't quietly pick the more dramatic reading and bury the other. It states the disagreement, cites both sides by name, and builds its simulation only out of the parts neither side contests. A model willing to show you exactly where the experts disagree, and refuse to resolve it for you, is not a model built to win an argument. It's built to be checked.

Use this section as the answer to "is this a fair thing to show my class?" It is the single clearest piece of evidence that the project treats its own limits honestly.

4. Discussion frameworks

Match the game's own posture in every discussion: the mechanisms, not moral verdicts. The game's internal design rule is "the interface never celebrates or scolds; it shows" — no event is narrated as a family's fault, no windfall is narrated as a family's virtue. Frame classroom discussion the same way: ask what the mechanism was, not who "deserved" what.

Era 1 — Reconstruction & Betrayal (1865–1896)

What specific federal decision (Special Field Order No. 15, then its reversal) created and then erased the "40 acres" promise? Who made that decision, and was it a change in law or a change in enforcement?
The Freedman's Savings Bank collapsed in 1874 and depositors were never made whole. Was this a policy failure, a fraud, or both? What does "up to 62% reimbursement, often much less" tell you about the difference between a legal remedy and an actual one?
Sharecropping kept a family from ever banking a surplus. Is that a story about effort, or about contract terms?

Era 2 — Jim Crow & the Nadir (1896–1932)

The Tulsa Race Massacre (1921) destroyed a thriving Black business district in a matter of hours. Insurance claims were filed and almost universally denied. What does "the claim was denied" reveal that "the neighborhood was destroyed" alone does not?
Black land ownership had been rising before this era, then collapsed by roughly 90% over the following decades. Why is it important that the loss came after a period of real progress, not instead of one?

Era 3 — New Deal & War (1932–1950)

The FHA and GI Bill are usually taught as universal, race-neutral programs that built the postwar middle class. The game's sourced numbers (roughly 98% of FHA-backed loans to white borrowers; 2 of 3,229 VA-guaranteed loans to Black veterans in one 1947 Mississippi sample) argue otherwise. What's the difference between a law's text and its administration — and why does that distinction matter for evaluating any policy, then or now?
Social Security's original exclusion of agricultural and domestic workers is a genuinely disputed case in the historical record: was the motive administrative convenience, or deliberate exclusion? The game presents both readings rather than picking one. What kind of evidence would it take to settle a question like that?

Era 4 — Suburban Boom (1950–1968)

Contract selling let a family "buy" a home with no legal protection and no equity, at a markup of roughly 84% over market value. How is a contract sale different from a mortgage, mechanically — and why would a family accept those terms?
Urban renewal and interstate construction displaced roughly a million people nationally, disproportionately people of color. Ask students to find a specific local example (their own city, if documented) rather than relying on the national percentage alone.

Era 5 — Civil Rights & Backlash (1968–2000)

The Fair Housing Act passed in 1968. The sourced research the game draws on found no measurable structural break in Black suburban residence rates around that date. What does that imply about the difference between passing a law and closing a gap?
The game's "education doesn't close the gap" data point (median wealth of a Black college-graduate household was roughly a tenth that of a white college-graduate household in 2013) is one of the sharper numbers in the whole game. Push students past the number to the mechanism: what other factors (the eras before this one) would already have been baked into a family's wealth by the time a degree is earned?

Era 6 — New Economy (2000–2025)

Subprime mortgage targeting (2004–2008) and the neighborhood-devaluation research (homes in Black neighborhoods valued roughly 23% lower than comparable homes elsewhere) both predate the 2008 crash and compounded its damage. Ask: was the crash itself colorblind, even if its setup was not?
The game's closing paradox, drawn directly from real 2022 Federal Reserve data: Black median household wealth grew faster in percentage terms than white wealth from 2019–2022, and the dollar gap still widened. Have students explain, in their own words, why both of those things can be true at once. (This is arithmetic, not opinion — a good check of whether the point landed.)

The NULL-TEST toggle — a built-in "what if" tool, and a lesson in model validity

After a student finishes their first playthrough, the Ending screen unlocks a real, clickable NULL-TEST toggle: "What if none of it had happened?" One click re-walks the exact same run — same seed, same choices — with every sourced racial disparity removed, so both families face identical wages, odds, and starting wealth. Clicking it back ("Show what actually happened") restores the real numbers instantly; nothing about the saved run is changed by trying it.

This is the ready-made classroom demonstration: have a student play their run once, read the ending as authored, then flip the toggle and ask them to explain — in terms of the mechanism, not the outcome — why the gap collapses to almost exactly 1.00-to-1. It's the same logic as a placebo arm in a clinical trial or a null hypothesis in a statistics class: why does a model that produces no gap when you remove every input that could create one serve as evidence the model isn't rigged? A good bridge lesson if the class also touches research methods, and a stronger, hands-on version of the same honesty test the About page describes in prose.

Counterfactual scenarios — an engine capability, not (yet) an in-game control

Separately from the NULL-TEST toggle, the simulation engine can also replay a run with a single historical wrong undone (e.g., "no redlining," "GI Bill administered equally") — this machinery lives in the codebase (src/lib/engine/counterfactual.js, src/lib/data/counterfactuals.json) and is exercised by the project's own tests. It is real, sourced, and auditable by a Tier 2 class that opens the repository, but there is no button or toggle for it in the game a student plays in the browser today. If your class does the Tier 2 repo audit, this is worth pointing out as an example of a documented, tested capability that hasn't yet been surfaced as a player-facing feature — a fair thing for students to notice and even suggest as a future addition.

The ruin mechanic's honesty stance

The game includes a "bust" or game-over state a family can reach under a severe run of bad luck — and it is now a real, measured outcome, not a hypothetical. The project's own design rule for this mechanic, worth quoting directly to students as a model of intellectual honesty in software: failure rates must be "emergent and reported, never target-tuned." In plain terms — the developers did not decide in advance what percentage of families should go bust and then rig the odds to hit that number; the odds come from the sourced probabilities already in the model (chiefly a medical-catastrophe mechanism now present in five of the game's six eras, striking a family that is carrying too thin a cash cushion), and whatever bust rate results is measured, by running the simulation many thousands of times, and reported rather than picked.

As currently measured, under ordinary play roughly one run in sixty ends this way for the Freeman family — a small but real and non-zero share, consistent with the game's own 1–5% target envelope for the first time since this mechanic was built. Under the same ordinary play, the Walker family's measured bust rate is effectively zero — not because the code treats him differently, but because the same sourced shock lands on a much larger cushion. And the honesty gate holds here too: run the equalized-inputs test from the section above, and both families' bust rates converge to the same (very low) number.

The project originally set a much higher target for a deliberately worst-played run — bust as the majority outcome. That target has since been formally closed, and the reasoning is worth teaching alongside the number: the developers measured worst-play bust at roughly 6.5% (its best-case result), diagnosed exactly why it couldn't go higher without inventing an unsourced mechanism (one era in the model is structurally a "dead zone" for this event by design, and a worst-played family never sheds the owned assets that would make it vulnerable to it), and accepted the lower, measured number rather than forcing the math to hit the original target. The project chose recording why a target wasn't reachable over inventing a mechanism to hit it — that choice is itself the honesty stance worth pointing students to, not a loose end. Treat the canonical ~1-in-60 figure as a snapshot that could still move as the game develops further, but treat the worst-play conclusion as settled and on the record, not pending.

5. Practical matters: length, content, accessibility

Session length

A measured reference playthrough — full prologue, every mini-game played, text read rather than skipped — took about 10–11 minutes. Reading pace and how much text a player skips will move that; a realistic range is 5 to 20 minutes. Either way, a full run fits comfortably inside a single class period with room for discussion. The game also saves and reloads, so a run can be split across two class periods without losing progress if you'd rather spread it out. For a single 50-minute period, plan on: a short intro (5–10 min), the playthrough itself (5–20 min), and time for the ending screen, the audit assignment (Section 2), and initial discussion (Section 4) in the same sitting.

Content notes — what's actually depicted, for age-appropriateness planning

The game covers slavery's aftermath, family separation, a documented race massacre, forced displacement, and financial exploitation. Concretely:

The Prologue covers the history before 1865, including slavery and the sale of family members away from one another. It is written and illustrated with restraint — text and illustration, not graphic depiction.
The Tulsa Race Massacre (1921) appears as a full-screen event: a single sourced sentence and a restrained illustration, no interactivity, no "win/lose" framing. The game's tone rule — "the interface never celebrates or scolds; it shows" — applies here specifically: the massacre is presented as a historical fact the family survives, not as a spectacle or as a consequence of the player's choices.
Later eras include eviction, wage garnishment, bankruptcy, and a modern financial-ruin state, presented as economic mechanics rather than as personal failure.

This content is appropriate for high school and college audiences. For younger students, or a class with no prior exposure to Reconstruction-era or Jim Crow-era history, preview the Prologue and the first two eras yourself before assigning the game, and consider a short framing discussion beforehand.

Accessibility

Playable muted. All essential information is conveyed in text and visuals; sound is atmospheric. A mute control and independent music/ambient/sound-effect volume sliders are built in.
Keyboard-operable. Decisions, dialogs, and navigation are designed for keyboard-only play.
Reduced motion respected. The game checks the operating system's "reduce motion" preference and shortens or removes its own animations accordingly, rather than relying on browser defaults alone.
No independent, published accessibility audit (e.g., a third-party screen-reader or axe/Lighthouse report) is available at the time of writing. If a student in your class relies on assistive technology, do a quick personal check before assigning it broadly.

6. Limitations, written straight

The game's own About page has a "What this model leaves out" section; this expands on it plainly, so you can answer a sharp student's "but doesn't this leave something out?" honestly.

A median-family lens, not a census. The game follows one typical family per side through each era. It does not represent the full range of real Black or white households' experiences, and it says so.
Dollar levels are calibrated; racial ratios are sourced. Every wage ratio, approval-odds gap, markup, and inheritance-size disparity is a real, cited research figure and is frozen — the game cannot quietly change it. The absolute dollar amounts and era-income baselines around those ratios are tuned so the whole simulation reproduces the real historical trajectory (roughly 60:1 in 1865, narrowing to roughly 6:1 today); they are disclosed as calibration, not presented as an independently sourced figure.
Low-confidence claims are excluded from the math, not just downgraded. The research ledger documents figures the researchers explicitly could not verify to a usable standard (an unconfirmed dollar total for the 1898 Wilmington coup, an unverifiable "Black-owned businesses 1900–1914" count, and others). These are flagged narrative-only in the ledger and were not wired into the simulation's structured source list at all — they never drive a dollar or probability calculation.
Some real costs of deprivation are left out because no defensible sourced figure exists yet — for example, the added health cost of a utility shutoff, or how much a long-deferred home repair ultimately costs to fix. The game states this as a disclosed gap, with research queued to fill it, rather than inventing a plausible-sounding number.
Contested questions are preserved as contested, not resolved by the model. Beyond the inheritance-vs-housing decomposition (Section 3), the game explicitly preserves several other live disputes rather than picking a side: the motive behind Social Security's 1935 exclusion of farm and domestic workers (administrative feasibility vs. deliberate exclusion); how much of the racial gap in bankruptcy-chapter outcomes is attributable to bias specifically (the best causal estimate found attributes only 15–53% of the measured gap to bias, not the whole gap); and whether the standard "payday debt trap" framing overstates a phenomenon that is real but concentrated in a rollover-prone minority of borrowers.
Race functions as a data input, never as a branch in the simulation's logic. The underlying engine reads a family's cash, income, and buffer state — it does not contain an "if Black, then X" rule anywhere. The gap that emerges is a consequence of the sourced disparities in wages, approval odds, and inheritance being applied to otherwise-identical decision logic, not of the code treating the two families differently by design.
Education's role is deliberately left unquantified. The game's ending states this outright: no number in the simulation accounts for who reached a classroom, a pre-K seat, or a college lecture hall — even though the developers consider it likely one of the most important drivers running through the whole record. Leaving it out is disclosed as a limitation, not smuggled in as a hidden assumption.
The ruin/bust mechanic is real and measured, and its target-setting is now settled. A small share of ordinary runs (currently measured at roughly one in sixty) genuinely end early at the game-over screen, driven by a sourced medical-catastrophe mechanism against a thin cash cushion — a rate that is emergent and reported, never tuned to hit a target, and that sits inside the project's own 1–5% target envelope. A separate, higher target for a deliberately worst-played run (bust as the majority outcome) was measured, found structurally out of reach without inventing an unsourced mechanism, and formally closed rather than forced — the project recorded why the higher target wasn't reachable instead of rigging the math to hit it, which is itself the teachable honesty stance (Section 4). Treat the canonical ~1-in-60 figure as a snapshot that could still move as the game develops further; treat the worst-play target's closure as settled.

If you or a student finds an error anywhere in the game's sourcing, the project explicitly invites correction — treat that as an extension of the audit assignment, not a dead end.