A Public Instrument of Record

The professional forecasting class keeps no record of its own performance. Nostradamus Intellect was built to be the missing ledger — and to live inside it.

A track record that starts empty and fills in public is the only kind worth having, because it is the only kind that cannot be forged.

A Civilization That Never Checks Its Answers

Every day, the professional forecasting class produces thousands of confident claims about the future. Recessions are called, wars are predicted, currencies are pronounced dead, technologies are declared inevitable or impossible. The claims are broadcast, quoted, monetized — and then, almost without exception, never checked. There is no profession on earth that keeps fewer records of its own performance than public prediction.

This is not a new complaint, and it is not a rhetorical one. In the largest systematic study of expert judgment ever conducted, the political scientist Philip Tetlock tracked 82,361 forecasts from 284 professional experts over two decades. The average expert barely outperformed random chance. More troubling: the experts most often quoted in the press — the boldest, the most vivid, the most certain — tended to be the least accurate. Fame and calibration were inversely related. The market for foresight, left unaudited, rewards precisely the wrong things.

The structure of modern attention has made this worse, not better. Roughly six in ten questions typed into a search box now end without a single click through to an underlying source; the answer arrives pre-digested, stripped of provenance, consumed in seconds and never revisited. In an economy where information is inhaled rather than examined, a confident wrong answer and a careful right one earn exactly the same glance. Confidence travels farther than accuracy, because confidence is legible at a glance and accuracy takes months to reveal itself — by which time the audience, and usually the forecaster, has moved on.

Call the mechanism what it is: pundit amnesia. The forecast is issued in the future tense and consumed in the present tense. When reality finally votes, no one is in the room to count the ballot. The analyst who called eleven of the last two recessions suffers no penalty; the analyst who quietly assigned careful probabilities and was right earns no premium. There is no ledger.

Other high-stakes professions solved this long ago. Aviation has the black box and the incident report. Medicine has the morbidity and mortality conference. Baseball has the box score, which is why no one in baseball can lie about a batting average for long. Public prediction — the discipline that claims to guide nations, portfolios, and lives — has nothing. No black box, no autopsy, no box score. That absence is not an oversight. It is a business model. Unchecked prediction is lucrative precisely because it is unchecked. Nostradamus Intellect exists to build the missing ledger, and to live inside it.

What Changes When a Forecast Can Be Wrong

The proposition behind Nostradamus Intellect is not that we predict better than others. That claim would be exactly the kind of unaudited boast we were built to retire. The proposition is narrower and harder: every forecast we publish is constructed so that it can be checked, and every check is performed in public, on a schedule announced in advance.

Concretely, each projection on the site carries four mandatory elements. An explicit probability — not “likely,” not “risks are rising,” but a number that can later be scored. A confidence interval — an honest statement of how much the number itself should be trusted. A falsifiability condition — a plain-language sentence, printed on the face of every projection card, stating exactly what observable event would prove the forecast wrong. And a public resolution date — the day on which the world’s answer is read off and entered into the ledger, whatever it says.

These four elements convert a rhetorical gesture into an instrument reading. A forecast without them can never be wrong, which is another way of saying it can never be right. A forecast with them becomes a wager against reality, timestamped and non-retractable.

How the Scoring Works

The ledger is scored with the Brier score, the same measure used for decades to grade weather forecasters — the one profession that is routinely and publicly checked, and, not coincidentally, one of the best-calibrated professions in existence. Decades of verification research show that when an experienced forecaster says seventy percent chance of rain, it rains close to seventy percent of the time. Accountability produced calibration. It was not the other way around.

The arithmetic is simple enough to do at a kitchen table. Take the probability the forecaster assigned. Subtract what actually happened — one if the event occurred, zero if it did not. Square the difference. Lower is better: zero is perfection, and 0.25 is the score of a coin flip, since a forecaster who answers “fifty percent” to everything scores 0.25 on every question. Squaring is what gives the measure its teeth — it punishes confident error far more severely than honest uncertainty.

Notice what the table implies. The pundit who thunders “this will certainly happen” and is wrong scores 0.81 — catastrophically worse than the modest analyst who said sixty percent and missed. Under Brier scoring, hedging is no refuge and bluster is no strategy. The only way to a good score, over enough predictions, is to actually know how much you know. That property — that the measure cannot be gamed by style — is why we chose it, and why the entire ledger, methodology included, is public before a single prediction has resolved.

Our first predictions were sealed on the third of July, 2026. Ten of them, every one carrying its probability, interval, falsifiability condition, and resolution terms. As of this writing, all ten stand at the only honest status a new instrument can claim: pending. The first public Brier scoring takes place in July 2027. Until then, our track record is — precisely, deliberately, and in writing — zero.

The Refusals: What We Will Not Do

An instrument is defined as much by what it refuses to measure as by what it measures. Three refusals are structural to this project, and they are worth stating in the plainest available language.

We will not fabricate a track record. Nearly every entrant in the prediction business arrives carrying a portfolio of retrofitted triumphs — the crash they “called,” the election they “saw coming,” assembled after the fact from ambiguous statements no one could have scored at the time. We arrive carrying nothing. The ledger begins at zero on a known date, and nothing said before that date counts, including anything we might be tempted to claim informally. A track record that starts empty and fills in public is the only kind worth having, because it is the only kind that cannot be forged.

We will not traffic in unfalsifiable mysticism. Here the name over the door requires explanation, because the name is a deliberate irony. Michel de Nostredame, publishing his quatrains in 1555, achieved the most durable forecasting franchise in Western history by a method worth studying in reverse: he attached almost no dates, named almost no verifiable quantities, and wrote in an ambiguity so perfect that five centuries of readers have been able to find any event they wished inside his verses. The quatrains cannot fail because they cannot be tested. They are, in the strict sense, unfalsifiable by design — and that design is exactly why the franchise survived. We have kept the ambition of the name — the audacity of looking far forward — and attached to it the one thing the original never faced: accountability. His quatrains were unfalsifiable by design. Ours are falsifiable by contract. The falsifiability condition printed on every projection card is that contract, and any claim that cannot carry one is refused admission. The site’s Library shelves the real quatrains beside Thucydides, Ibn Khaldun, and the modern quantitative historians, not as authority but as exhibit: this is the tradition we inherit, and this is precisely what we changed about it.

We will not give advice. Every forward-looking figure on the site is labeled probabilistic simulation, because that is what it is. Nothing we publish is financial advice, medical advice, or instruction for living. An observatory reports what its instruments show; it does not tell the sailor to sail. The house rule is carved at the top of our methodology and applies to every sentence we publish, this one included: if a claim cannot be tested, it does not belong here.

One more clarification belongs under this heading. The analytical core of the system is called the Core, after the clockwork instrument that modeled the motions of the heavens on a tabletop. The name is chosen with care. An orrery does not hear voices; it has no intuitions; it holds no séances with the future. It is gears and ratios — a mechanism that makes visible the structure of a system too large to see whole. Ours models not planets but the rhyme of history. The Core is not an oracle. It is a convergence engine: twelve continuous public data streams, five measured components, six macro-historical frameworks, converging on a structured resemblance between the present and the indexed past — and only then, downstream of all that measurement, generating forward branches with probabilities attached.

The Lens: Five Components, Three Centuries of Evidence

A single “global risk score” would be astrology with extra decimal places — one opaque number, unarguable and uninformative. The Core instead decomposes every historical resemblance into five measurable components, each fed by public data, each weighted by weights that are themselves published. The five were not chosen for elegance. They were chosen because, across three centuries of indexed stress episodes, they are the variables that recur — the load-bearing walls of every documented crisis in the corpus. Six frameworks discipline the reading of them: Kondratieff’s long waves, the Thucydides Trap, Ibn Khaldun’s asabiyyah, Turchin’s elite overproduction, Perez’s technology surges, and Dalio’s long debt cycle.

Debt load. When obligations grow faster than the income that services them, a society’s room for error narrows until an ordinary shock becomes an extraordinary one. The canonical illustration is 1929–1933: a decade of accumulating private leverage met a fall in prices, and because the debts did not fall with the prices, liquidation fed deflation and deflation fed liquidation — the spiral Irving Fisher described from inside it in 1933. The interwar reparations chain — Germany owing the Allies, the Allies owing America — shows the same variable operating between nations rather than within one: a debt structure so rigid that it transmitted every tremor to every party. Our debt component reads the modern equivalents from FRED, the IMF, and the BIS, against Dalio’s template of the long debt cycle.

Great-power rivalry. Thucydides located the cause of the Peloponnesian War not in any incident but in a structure: the fear that a rising power instills in a ruling one. Graham Allison’s survey of the last five hundred years found sixteen such structural confrontations; twelve ended in war. The clearest modern instance is the era our index calls “The Long Fuse,” 1910–1914, when Germany’s industrial output overtook Britain’s and the dreadnought race converted economic anxiety into naval steel on a published schedule — a rivalry so legible that contemporaries could count it, and so unmanaged that counting did not save them. The rivalry component tracks its present-day analogues through SIPRI’s armament data and the geopolitical risk index, among other streams.

Polarization, Technology, and the Plumbing of Money

Internal polarization. Ibn Khaldun, writing in the fourteenth century, identified asabiyyah — the shared social cohesion of a group — as the asset every dynasty accumulates in adversity and spends in comfort, and he treated its depletion as measurable in generations. Peter Turchin’s modern quantitative work adds a mechanism: when a society produces more claimants to elite position than positions to claim, the surplus contenders radicalize, and internal conflict follows with disquieting regularity. The American stretch from 1968 to 1974 — assassinations, burning cities, a campus shooting by national guardsmen, a presidency ending in resignation — is the corpus’s reference case for cohesion failing inside an otherwise wealthy society, and it is worth noting soberly that this is the era the present most resembles in our current index. The component reads polarization through Pew’s long-run survey series and allied measures.

Technology shock. Carlota Perez documented the recurring sequence: a transformative technology arrives, finance rushes in, paper valuations detach from installed reality, and the gap closes violently before the technology’s real deployment begins. The 1920s ran the full sequence with electrification and radio — RCA’s share price rising roughly tenfold in the late twenties before losing almost everything after 1929, while the underlying technologies went on to remake the century anyway. The dot-com episode of 1999–2000 ran it again in compressed form. The shock, in both cases, was not that the technology failed; it was that the society repriced it faster than it could absorb it. The component tracks adoption curves and investment intensity through ITU data and the technology series in our corpus — with obvious contemporary relevance.

Monetary strain. The quietest component and, historically, the fastest-moving. Money is the plumbing; strain in it turns local failures into general ones. On the fifteenth of August, 1971, the United States closed the gold window and ended the Bretton Woods system over a single weekend — a monetary regime that had anchored the postwar world dissolved by announcement, and a decade of inflation followed. Four decades earlier, the interwar gold standard had performed the inverse service, transmitting deflation from country to country with mechanical efficiency and forbidding the remedies that might have stopped it. The strain component reads reserve structures, currency stress, and the modern experiments in money — including the crypto-asset series — through BIS, IMF, and CoinGecko feeds.

The five components are scored separately and then read together, because history’s worst passages are precisely the ones where they converge. Our Historical Resonance Index currently scores July 2026 against nine indexed eras, with public weights: the strongest resemblance is to 1968–1974 at 83 of 100, with 1910–1914 and 1873–1896 tied at 82, then 1929–1933 and 1848–1849 at 79, 1937–1939 at 78, 1815–1819 at 77, and 1999–2000 and 1987 in descending order. We publish those numbers with a standing caution that is also the method’s honest boundary: resemblance is not destiny. The index can tell you the present rhymes with a dangerous stanza. It cannot tell you how the verse ends — that is what the forward branches, with their probabilities and falsifiability conditions, are for.

The Goal: Three Horizons

Stated plainly, the goal has three horizons. Now: an instrument of record. Ice cores let climatologists read the atmosphere of centuries they never breathed. We are attempting the equivalent for judgment: a public, timestamped record of what a disciplined analytical system believed at a given moment, at what probability, on what evidence, and with what result. A historian in 2076 examining the middle of our century should be able to consult this ledger and know — not reconstruct, not infer, but read — what a calibrated observer standing in July 2026 considered likely, and how the reckoning went. No such record exists for any previous era. Every generation’s forecasters vanish behind their own ambiguity. We intend to be the first who can be checked by people not yet born.

Next: a training ground for calibrated citizens. Reading a probability is a skill, and it is nowhere taught. A public that cannot distinguish “70 percent likely” from “certain” — or from “pundit talking” — is a public that can be stampeded by anyone with a confident voice. The Calibration Ledger is designed to be a gymnasium as much as an archive: any reader can log the sealed predictions, record their own numbers beside ours, and compute their own Brier score when resolution day arrives. From 2027, the Observer Protocol formalizes this. A contributed signal passes through provenance checks, then a consistency test against the twelve-feed corpus, then weighting by the author’s own public Brier score, before it is absorbed into Pattern Memory or routed to human review. The design principle is simple: the brain grows smarter with every honest, well-calibrated observer — and only with those.

Eventually: an engine sustained by its observers. The roadmap is published and deliberately unhurried. The beta now live is Phase 0, free and open. Phase 1, in 2027, brings observer accounts, reputation, and the first scoring cycle. Phase 2, across 2027 and 2028, is the one we consider genuinely new: observers pool AI inference credits from any provider they choose, funding continuous Monte Carlo scenario sweeps across multiple frontier models — with the divergence between the models treated as a signal in its own right, since the questions on which capable models disagree are exactly the questions reality has not yet priced. One analyst asks a question. A hundred thousand observers keep the engine asking every question, all the time. What comes after Phase 2 is deliberately unpublished until the scores exist. That ordering is not a scheduling detail; it is a fourth refusal. The instrument must earn its record before its record is worth governing.

The Appointment: July 2027

In July 2027 the first sealed predictions to reach their resolve-by dates are graded, and their Brier scores are computed and published — every score, in public, exactly as the arithmetic falls. The set is small and it was fixed in advance: the pre-registered first scorecard names exactly which seals it covers and on what dates, so that no easier entry can be added later to soften a miss. If the numbers are good, they will be the beginning of a track record rather than the proof of one; a first scorecard is a data point, not a verdict. If the numbers are poor, they will be published with equal prominence, because a ledger that only records victories is not a ledger. Either way, something will exist in July 2027 that does not exist for a single one of the confident voices filling the world’s feeds today: a checked forecast, checked in the open, on a date fixed before the outcome was known.

That is the entire proposal. We do not ask to be believed. Belief is the currency of the old regime — the unfalsifiable quatrain, the amnesiac pundit, the score no one kept. We ask instead for the one thing the old regime could never survive. Judge us in twelve months.

Frequently asked

Why is the project named after Nostradamus if his prophecies were unfalsifiable?

The name is a deliberate irony. Michel de Nostredame built the most durable forecasting franchise in Western history on ambiguity that could never be tested. We kept the ambition of looking far forward and attached the one thing the original never faced: accountability. His quatrains were unfalsifiable by design; ours are falsifiable by contract — a plain-language failure condition is printed on every projection card.

What is a Brier score and why does Nostradamus Intellect use it?

The Brier score grades probabilistic forecasts: take the stated probability, subtract the outcome (1 if the event occurred, 0 if not), and square the difference. Zero is perfect, 0.25 is a coin flip, 1.00 is perfectly confident and perfectly wrong. It cannot be gamed by style — confident error is punished far more severely than honest uncertainty — which is why weather forecasting, the one publicly checked profession, became one of the best calibrated.

Does Nostradamus Intellect have a track record?

No — deliberately. Ten predictions were sealed in the public Calibration Ledger on 3 July 2026, and all ten currently stand at PENDING. The first public Brier scoring takes place in July 2027. Until then, the track record is precisely, and in writing, zero. A ledger that starts empty and fills in public is the only kind that cannot be forged.

Are the projections financial or investment advice?

No. Every forward-looking figure on the site is labeled a probabilistic simulation, because that is what it is. Nothing published by Nostradamus Intellect is financial advice, medical advice, or instruction for living. An observatory reports what its instruments show; it does not tell the sailor to sail.

What happens in July 2027?

The first sealed predictions to reach their resolve-by dates are graded, and their Brier scores are computed and published — every score, in public, exactly as the arithmetic falls. Which seals those are is not decided in 2027; it is pre-registered and timestamped, and published at /first-scorecard. Good numbers will be the beginning of a track record, not the proof of one; poor numbers will be published with equal prominence, because a ledger that only records victories is not a ledger.

Sources

Tetlock, P. E., Expert Political Judgment: How Good Is It? How Can We Know?, Princeton University Press, 2005 — 82,361 forecasts from 284 experts.
Brier, G. W., “Verification of Forecasts Expressed in Terms of Probability,” Monthly Weather Review 78(1), 1950.
Murphy, A. H. & Winkler, R. L., “Reliability of Subjective Probability Forecasts of Precipitation and Temperature,” Journal of the Royal Statistical Society C 26(1), 1977.
Fisher, I., “The Debt-Deflation Theory of Great Depressions,” Econometrica 1(4), 1933.
Allison, G., Destined for War: Can America and China Escape Thucydides’s Trap?, Houghton Mifflin Harcourt, 2017 — 16 cases since 1500, 12 ending in war.
Ibn Khaldun, The Muqaddimah, 1377.
Turchin, P., Ages of Discord: A Structural-Demographic Analysis of American History, Beresta Books, 2016.
Perez, C., Technological Revolutions and Financial Capital, Edward Elgar, 2002.
Dalio, R., Principles for Navigating Big Debt Crises, Bridgewater, 2018.
SparkToro / Datos, “2024 Zero-Click Search Study” — 58.5% of US Google searches ended without a click to the open web.
Nostredame, M. de, Les Prophéties, Lyon: Macé Bonhomme, 1555.
Public data streams: FRED, World Bank, SIPRI, GPR Index (Caldara & Iacoviello), Our World in Data, ITU, Pew Research Center, NOAA/NASA, IMF, BIS, Maddison Project, CoinGecko.