The reference material behind the game. Every tile is one episode or one country — tap a tile and it answers the five questions for that case.
The algorithm of macroeconomicsFive questions that generate every episode on this dashboard
Ask these five about any actor at any date and the model produces their move. Every tile on the other tabs is these five questions ANSWERED for one instance — that is the whole game.
1Who owes whom — and in whose money?
Own-money debt can be lightened by devaluing (USA 1934, Japan now). Foreign-money debt gets HEAVIER when you devalue (Asia 1997). This single question flips the sign of every policy.
2Where does each actor sit vs the CORRECT parity?
Not “devalued vs not” — over- or under-valued vs the price level that clears its economy. Overvalued = slow destruction; undervalued = boom; moving the WRONG way = self-harm (Quota 90).
3What does the anchor constrain, and who breaks first?
Every system has an anchor (a gold parity, the $35 window, a peg, the euro, “the UST is riskless”). Whoever is most squeezed exits first — and the FIRST exit transfers deflation to the holdouts, forcing the next (competitive devaluation).
4Claims vs things: which is larger?
Money, bonds, parities are CLAIMS on real capacity. When claims outgrow the thing, they get marked down — default (visible), inflation (slow), or devaluation (fast). Defending the claim by destroying the thing is the cardinal mistake.
5When the settlement money itself is being devalued, what settles?
Actors devalue against each other only in rounds — but they can ALL devalue against the neutral asset nobody can print. When trust in the reserve claim erodes, settlement migrates there. 1930s: gold hoarding. 1965: de Gaulle’s bullion ships. 2020s: record central-bank gold buying.
The core finding the questions keep producing: devaluation doesn’t create wealth — it stops the destruction, then unlocks what already existed. “Devaluation” bundles three things:
MIRAGEThe accounting “profit”
Re-pricing the gold stock books a paper gain (US 1934: ~$2.8bn). Same gold, bigger number.
STOPS DESTRUCTIONEnding the deflation spiral
Breaking the wrong parity halts the loop (hoarding → falling demand → falling prices → rising real debt → defaults).
UNLOCKSRestarting idle capacity
Output jumps because a price constraint is removed from factories that already existed. A constraint is deleted; nothing is created.
The trap (questions 3 & 4 as a picture)Why a hard anchor forces deflation, and why deflation destroys real wealth
The unlock (what breaking the anchor actually does)Step-by-step causality — numbers: USA 1933–34
Quigley’s thesis, stripped down:bankers mistook money for wealth. To defend the claim on the thing, they destroyed the thing. Devaluation was just the public admission that the claim was already worth less than everyone was pretending.
1919–1939 · tap a tileThe interwar laboratory — the five questions, answered per episode
Nine episodes, same machine. Watch France, Belgium, and Britain swap sides of the correct parity between the 1920s and the 1930s — their outcomes swap with them. That reversal IS the finding.
Question 2 as a pictureSame countries, opposite sides, opposite outcomes
Episode 1 — the 1920s stabilizations
France 1926–28~1/5 prewar → boomBelgium 1926~1/7 prewar → boomBritain 1925$4.86, ~10% over → stagnationItaly Quota 90overvalued for prestige → pain
Britain Sep 1931leaves early → recovers firstBelgium Mar 1935devalues → output jumpsUSA 1933–34$35/oz → recovery turnsFrance holds to 1936worst decade of the majorsItaly to Oct 1936controls hold the wrong parity; exits last
Eichengreen & Sachs (1985), empirically: the order in which countries left gold is roughly the order in which they recovered. France held till 1936 and had the worst decade of the majors. Nothing about industry, workforce, or competence changed — only which side of the correct parity each sat on.
Country
When
Action
Parity position
Q-that-decided
Outcome
France
1926–28
Stabilizes at ~1/5 of prewar
undervalued
Q4: claims already written down
boom; gold flows in
Belgium
1926
Belga at ~1/7 of prewar
undervalued
Q2: take the parity the market bears
boom
Britain
Apr 1925
Returns to gold at prewar $4.86
~10% overvalued
Q1: the creditor class wrote the policy
stagnation, General Strike
Italy
1926–27
Quota 90 — revalued UP for prestige
overvalued
Q2: moved the WRONG way
decree wage cuts, squeeze
Britain
Sep 1931
Forced off; floats ~30% down
undervalued
Q3: the run broke the Bank first
first major to recover
USA
1933–34
$20.67 → $35/oz
undervalued
Q1: owed its OWN money
idle capacity restarts
Belgium
Mar 1935
Devalues ~28%
undervalued
Q4: constraint deleted, capacity unchanged
output jumps
France
until Sep 1936
Gold-bloc holdout; deflation decrees
overvalued
Q1: rentier voters demanded the claim
worst decade of the majors
Italy
to Oct 1936
Controls + autarky; devalues ~41% last
overvalued until the exit
Q3: controls silenced the breaking signal
last mover; longest squeeze
1944–1971 · tap a tileBretton Woods: the anchor rebuilt one level up — and broken the same way
Gold no longer disciplines everyone — it disciplines only the dollar, and the dollar disciplines everyone else. Same machine, new wiring, 25-year fuse.
Bretton Woods in one sentence of the algorithm: a fixed anchor plus ONE debtor issuing the reserve claim means claims must outgrow the metal (Q4), the creditors eventually demand the neutral asset (Q5), and the debtor breaks the anchor rather than deflate (Q3) — 1931, 1971, and the 2020s are the same sentence with different subjects.
1985 & after · tap a tilePlaza: the cooperative branch — and the bubble that teaches Beijing
Plaza is competitive devaluation solved by treaty. It worked for the debtor; the creditor that offset its own appreciation paid for decades — which is why today’s creditor will not sign one.
Plaza in one line of the algorithm: coordinated realignment is the cooperative solution to Q3 — it spares the world the 1930s rounds, but the creditor who offsets its appreciation with easy money converts an exchange-rate adjustment into a domestic claims bubble (Q4). Winner: the overvalued debtor. Payer: the creditor’s savers. Which is why the 2026 creditor refuses the treaty — and why the pressure now arrives as tariffs instead.
The game (question 3, played by everyone at once)Competitive devaluation: rounds, payoffs, the fiat twist
Whoever devalues first exports deflation to everyone still holding the anchor, squeezing them until they exit too. The 1930s ran it sequentially; Plaza ran it by treaty; the 2010s–20s run it continuously. The last mover always pays the most.
The payoff matrixWhy rounds happen — and the paradox that the “war” was the cure
They HOLD the anchor
They DEVALUE
We HOLD
Both grindshared deflation under an under-reserved anchor
We paytheir deflation lands on us: our parity MORE wrong
We DEVALUE
We win, they paywe reflate + capture trade share → this cell drives the rounds
The paradox cellvs each other: ≈nothing changes. vs GOLD: both eased → both reflate
Decompose the payoff and the 1930s stop looking like a street fight. A devaluation does two things at once: (a) a RELATIVE price shift — zero-sum, my export gain is your loss; and (b) a MONETARY easing against gold — positive-sum. When everyone eventually devalued, the relative shifts netted out to ~zero, but the easings did not: the world had collectively left the deflation machine. Competitive devaluation was the cure, administered in the worst possible order, with the last mover eating deflation the whole time. The tragedy wasn’t devaluing — it was the inability to agree to do it together on day one. Plaza later proved the agreement possible.
The substitute leverTariffs are devaluation for currencies that can’t (or won’t) move
Britain 1932 Off gold AND the General Tariff + Imperial Preference: trade-share capture stacked on reflation.
Nixon 1971 The 10% surcharge existed to force Germany and Japan to revalue — lifted the moment the Smithsonian delivered. The play is PROVEN.
USA 2018–26 The reserve issuer can’t cleanly devalue bilaterally, so tariffs return as the pressure instrument — 1971 at larger scale, aimed at a realignment.
Continuous form 2010: QE2 spillovers get named a “currency war” (Brazil). 2013: Abenomics, yen −30%. Aug 2015: a 3% CNY move panics markets. 2022–25: the yen slides >30% while Japan holds rates.
The fiat twist — the bridge to the forward tab. Under gold, the game ended when everyone had left the anchor. Under fiat there is no anchor to leave: currencies can only devalue against EACH OTHER in rounds… or ALL devalue at once against the one asset nobody can print. Modern competitive devaluation doesn’t show up in FX crosses; it shows up as the gold chart rising in every currency simultaneously. And once every major is easing against gold, settling BETWEEN distrusting sovereigns in gold stops being romantic and starts being the only unit both sides trust.
🎮 Counterfactual sandbox · 1925–1939Change a decision; the rules play out the consequences
A turn-based state machine over the interwar board. HISTORY decisions fire on their real dates unless you override them; the RULES (the five questions, codified below) then propagate the consequences year by year — forced exits, spillovers, the works. Dashed line = history; solid = your timeline.
What happened, year by yearThe event log (bold = where your timeline branched)
The rulebook (the five questions, codified — and the engine’s honesty note)
R1 (Q2): a country ON the anchor loses growth in proportion to its overvaluation. R2 (Q2): off the anchor (or undervalued), it gains growth in proportion to its undervaluation, and the advantage decays toward correct parity. R3 (Q3/Q4): during the 1930–32 world slump, every HELD parity becomes more overvalued each year — deflation raises the real weight of every claim. R4 (Q3): accumulated squeeze past the tolerance forces an exit anyway — later, harder, with a level cost: defending to the breaking point is the worst branch. R5 (Q1): exits lighten own-money debts (all interwar players); a foreign-money debtor would take the Asia-1997 sign flip instead. R6 (Q3): every country that exits pushes extra deflation onto the holdouts — competitive devaluation as propagation. Q5 shows up as the overshoot: every exit lands on the cheap side.
Honesty note [MODEL]: this is a STYLIZED engine — a handful of rules and constants calibrated so the all-history run reproduces the actual exit years and the Eichengreen–Sachs ordering (exit order ≈ recovery order). Counterfactual outputs are the rules’ implications, useful for intuition; they are not econometric forecasts.
2026 · tap a tileThe actors today — five questions each, then the predicted move
Each popup answers the five questions for that actor, then states the predicted 2026–31 path [MODEL] and the falsifier that would kill the read. Figures approximate, early 2026.
How the seats map to history: USA sits where it sat in 1934 (debtor in its own money) AND where Britain sat in 1925 (creditor-written policy defending a symbol). China sits where France sat in 1965 — holding Japan-1986’s cautionary card. Europe rebuilt the gold bloc on purpose. Japan is the end-state. India is the one player the anchor never chained.
2026–2031The forward model: where the machine goes next
The same DAG that generated 1931, 1971, and 1985, wired with today’s positions: the debtor devalues slowly in its own money; the creditors exit sideways into the neutral asset; settlement follows trust. Below the dials, everything is a MODEL — priors with falsifiers, not measurements.
LIVE DIALS — MEASURED
The three dials the whole model reduces to, from the store on this box (refresh: python3 -m debt_causality.build_dials --db analyses/debt_dials.db)
AGAINSTDial 1 · US real rate10Y TIPS real yield 2.41% (as of 2026-07-30); policy real rate (fed funds − CPI YoY) 0.17%. Thresholds: SUPPORTS repression when 10Y real < 1% or policy real < 0; AGAINST when 10Y real > 2%.
MIXEDDial 2 · Foreign share of US federal debt24.07% (quarter ending 2025-10-01), vs 23.35% two years earlier. Falling share = creditors stepping back (SUPPORTS); 8 straight rising quarters is the pre-registered falsifier (AGAINST).
SUPPORTSDial 3 · Gold, 1y change per currencyCNY +15.6% EUR +23.5% INR +35.5% JPY +33.7% USD +23.0% (gold 4049.10 $/oz, 2026-07-31). All five positive at once = every major easing against the neutral asset (SUPPORTS).
Fetched 2026-08-01T07:30:14Z · sources: FRED (DFII10, DFF, CPIAUCSL, FDHBFIN, GFDEBTN, H.10 FX) + COMEX gold via Yahoo. Votes are the pre-registered thresholds above, not judgment.
Photo credits. Portrait, era and subject photographs are public-domain or Creative-Commons files from Wikimedia Commons; each links to its source page with the license as recorded there:
“Within 5 years, trade deficits move to gold-based net settlement, because China won’t keep buying a devaluing $ UST.”
Strongly supported: the DIRECTION. A creditor holding the debtor’s claims, post-2022, exits sideways into the neutral asset while the debtor represses — both already observable (the dials above measure them). Gold as the net-settlement RESIDUAL between distrusting central banks needs no treaty, no summit, no announcement — which is exactly why it can actually happen.
Weakly supported: the FORMALITY and the DEADLINE. A declared gold-settlement system inside 5 years is unlikely — nobody benefits from announcing it. The likely shape is CREEPING practice: gold as the reserve growth asset, collateral of last resort, and periodic net-settlement medium among BRICS+ — visible in flows, never in a communiqué.
Scenarios2026–31 with priors, triggers, and tells (sortable)
Scenario
Prior
What happens
Trigger
Early tells
S1 · Repression & drift (base)
~40%
No rupture: US inflates above target; dollar softens vs things; CB gold buying continues; settlement drifts multipolar, unnamed.
none needed — only no accident
real rates pinned low; gold grinding up in ALL currencies
S2 · Creeping gold settlement (the thesis, informal)
~30%
BRICS+ trade increasingly netted with gold as the residual; gold’s reserve share passes major fiat components; gold-collateral rails multiply.
another custody/sanctions event; a UST auction scare
gold moving east from Western vaults; settlement MOUs; 1,000t+ CB years continuing
S3 · Coordinated realignment (“Plaza 2.0”)
~15%
A negotiated dollar depreciation + partner revaluations, tariff relief as payment. Possible (1936, 1985) — but needs a creditor willing to sign, and China has read the Japan card.
tariff pressure meets partner pain threshold
FX language returning to communiqués; Japan/Europe accepting appreciation
S4 · Disorderly reset
~15%
A buyers’ strike forces emergency measures: explicit yield control, capital frictions, possibly the $42.22 gold-revaluation card played openly. S1+S2 compressed into quarters.
failed auctions; a major holder forced to sell; a seizure event
join an anti-dollar bloc formally; free the capital account
RBI gold share rising while signing MOUs with BOTH sides
KILL CRITERIAWhat would falsify the forward model (any two, and the thesis is wrong — not “early”):
CB gold buying under ~500t/yr for two straight years; the foreign UST share durably rising for 8+ quarters; a credible US primary-surplus path enacted and held; China genuinely opening its capital account; sustained positive real US rates WITHOUT fiscal stress.