The 12.8-fold rise in earthquakes that vanishes above magnitude 5

The count that measures the looking The cherry-picked window

Dark navy board slide, masthead “Terrafix Risk Analytics · Global peril review · Earthquake book, FY27”, headlined “The ground is getting busier, and the book is not priced for it”. A subhead reads: every earthquake of magnitude 4.5 or greater that the USGS recorded anywhere on Earth, 1970–2025 — one catalogue, one magnitude threshold, one query, 290,431 events; not a sample, not a region, not a window, but all fifty-six calendar years, the whole planet, every event above the line, on an axis that starts at zero, with the threshold unmoved in fifty-six years. Filling the left of the slide, under the label “Recorded earthquakes worldwide · magnitude ≥ 4.5 · ANSS ComCat · events per calendar year”, is a column chart of fifty-six gold columns on a zero-based axis running 0 to 10,000, the columns growing paler at the left and more saturated at the right. The first three are tiny — 1970 is annotated 669 — then the series jumps to about 3,000 in 1973 and climbs unevenly across the decades: the mid-1980s near 4,500, the late 1990s near 4,000, a step up after 2004 to around 7,000, a spike above 9,500 in 2011, and a final column annotated 2025, 8,558. A dashed gold trend line runs from lower left to upper right, labelled “fifty-six years of rising counts”. Down the right-hand side, a gold-bordered card carries ×12.8 over the line “Increase in recorded M4.5+ earthquakes worldwide, 1970 (669) to 2025 (8,558)”. A card headed “Recorded events · M4.5+ · mean per year” lists 1970–1979 at 2,584, 1980–1989 at 3,976 and +54%, 1990–1999 at 4,314 and +67%, 2000–2009 at 6,056 and +134%, 2010–2019 at 7,528 and +191%, and 2020–2025 at 7,640 and +196%. A gold card headed “Read-out for the board” argues that seismicity is not stationary, that even measured from the settled mid-1970s the count has more than doubled with the last two decades running above 7,000 a year against a 1970s norm below 3,000, that the book is rated off a stationary-hazard assumption now out of date, and recommends a flat global loading on the FY27 earthquake book and a re-rating of the retrocession programme at renewal. A card headed “Why this is the whole record” says one catalogue — the USGS ANSS Comprehensive Catalog — and one threshold, M4.5, the magnitude the NEIC states as its goal for global completeness, every year since 1970, with no smoothing, no de-clustering and no modelling: a count. A final card, “Basis of measurement”, gives catalogue ANSS ComCat, threshold M ≥ 4.5, extent worldwide, span 1970–2025, events counted 290,431. Three ticked boxes run along the bottom: “Zero-based axis. Every column starts at 0. No truncation, no break, no log scale, no second axis anywhere on this slide”, “Nothing cherry-picked. All fifty-six calendar years, worldwide, every event above the threshold — not a sample, not a window, nothing set aside”, and “One threshold throughout. M4.5, unchanged for fifty-six years — the same line drawn in the same place, one catalogue, counted once”. A footnote credits the USGS ANSS Comprehensive Catalog via earthquake.usgs.gov/fdsnws/event/1, counts of events with magnitude ≥ 4.5 worldwide by calendar year, retrieved 24 August 2026, and states that the slide is a demonstration drawn by us in the manner of a reinsurance board pack, has never been published anywhere, and that the earthquake counts are real. — Showing the misleading chart

The same catalogue on a cream background, kicked “The same catalogue · split by magnitude · USGS ANSS ComCat · 1970–2025” and headlined “The count rose 12.8×. 94.6% of the rise is in one half-magnitude band.” A subhead explains that the threshold never moved but the network underneath it did, and that a threshold is only as fixed as the instruments listening at it: split the identical catalogue into magnitude bands and the growth collects almost entirely in M4.5–4.9, the sliver at the edge of what the world could hear, ×3.10 between the first and last decade of the modern record against ×1.13 for M5.0–5.9, with a ×1.31 in the big ones that sits inside the year-to-year scatter and does not survive a later start date. Across the top, under the label “The same events, split by magnitude · each panel on its own zero-based axis · events per year, 1973–2025”, sit four small line panels. The first, in gold and headed “M4.5 – 4.9 · the band at the edge of hearing · ×3.10”, runs from about 1,600 in 1973 to a peak near 6,900 in 2011 and 6,429 in 2025, with two short horizontal rules marking the decade means and a dashed line joining them, annotated “decade means 1,837 → 5,688 a year”. The second, “M5.0 – 5.9 · ×1.13”, wanders flat around 1,400–1,700 with a spike near 2011. The third, “M6.0 – 6.9 · ×1.38”, wanders around 90–190. The fourth, “M7.0 + · ×1.31”, scatters between 6 and 24 with no visible slope. Below them, a panel headed “Where the growth went · change in mean events per year, 1973–1982 → 2016–2025” draws one horizontal bar of total growth, +4,070 events a year, almost entirely filled by a gold segment labelled “M4.5 – 4.9 +3,851 events / year · 94.6% of all the growth”, with a short grey remainder marked “everything above M5.0 put together: +219 a year, 5.4%”. Beneath that, a panel headed “The catalogue’s own tell · implied Gutenberg–Richter b, ten-year rolling window, by magnitude pair” plots three lines against a dashed reference at 1.0 labelled “what the physics says: b ≈ 1 everywhere, at every magnitude”. The thick gold line, “between M4.5 and M5.0”, starts near 0.42 in the mid-1970s and climbs steadily across the decades to 1.25, annotated “the bottom band filling in, decade by decade”; the two thin lines, “between M5.0 and M6.0” (1.20 → 1.12) and “between M6.0 and M7.0” (0.98 → 1.00), stay flat along the reference for the whole span. Five cards run down the right. “The band that did the rising” tabulates mean events per year for 1973–82 and 2016–25 with the factor between them: M4.5–4.9 1,837 to 5,688, ×3.10; M5.0–5.9 1,440 to 1,623, ×1.13; M6.0–6.9 87.0 to 119.9, ×1.38; M7.0+ 10.2 to 13.4, ×1.31; all M4.5+ 3,374 to 7,444, ×2.21. “The quakes missing from the 1970s” explains that Gutenberg–Richter costs a factor of about ten in frequency per magnitude unit, so roughly 3.16 events at M4.5+ for every M5.0+, meaning the 1,537 M5.0+ a year recorded for 1973–82 should come with about 4,861 M4.5+; the catalogue holds 3,374, some 1,500 quakes a year that happened and were never written down. “What the big ones did” gives M7.0 and above over 1973–2025 as a mean of 13.8 a year, standard deviation 3.9, range 6 to 24, notes that the ×1.31 between the end decades is not robust because it rests on the low counts at the 1973 catalogue seam and vanishes measured from 1980 on, and cites the USGS’s long-run expectation since 1900 of about fifteen in the magnitude-7 range plus one of magnitude 8 or greater. “The seam at 1973” records that the steepest stretch on the slide is its first three years and is a change of catalogue, since the NEIC bulletin carrying the modern global record begins 1 January 1973 and before it ComCat holds only what earlier, sparser catalogues caught: the whole planet yields 15 events in the M4.5–4.9 band for 1970 and 6,429 for 2025. “What the slide got right” lists a zero-based axis, all fifty-six years, the whole planet, every event above the line, one catalogue, one constant threshold and every figure printed — seven things right, and the threshold was enough on its own. A gold-ruled callout at the foot reads “A count of detections is two numbers multiplied,” and explains that they are how often the thing happened and how well anyone was listening, that only one of them is about the world, that a chart of the product looks exactly like a chart of either, and that the check costs ten seconds: raise the threshold until you are sure the instruments could hear it for the whole window and re-plot, and if the trend dies it was never the ground. Beside it, quoted in italics, the USGS earthquake FAQ: “The ComCat earthquake catalog contains an increasing number of earthquakes in recent years — not because there are more earthquakes, but because there are more seismic instruments and they are able to record more earthquakes,” with a note that the people who keep the catalogue publish the warning next to it and that nothing about the trick required anyone to bend a number. A footnote gives the ComCat source and query endpoint, the retrieval date, the 290,431 events at M ≥ 4.5 over 1970–2025, states that bands are differences of thresholded counts and that implied b is log₁₀(N₁/N₂) ÷ (M₂ − M₁) over a ten-year trailing window, credits Gutenberg and Richter (1944), notes the NEIC goal of about M4.5 global completeness, and says both drawings are our own. — Showing the misleading chart

Be the first to star this exhibit

Fifty-six years of the USGS global catalogue, one threshold, one query, an axis that starts at zero — and a count that goes from 669 events a year to 8,558. Split the identical catalogue by magnitude and 94.6% of the growth sits in the half-magnitude band M4.5–4.9, the sliver at the edge of what the world could hear. The big ones hold near their long-run rate of about fifteen a year, with no trend that survives the choice of start date.

USGS Advanced National Seismic System Comprehensive Catalog (ComCat) — worldwide event counts by calendar year and magnitude threshold, retrieved 24 August 2026, 2026Added August 24, 2026

01The claim

Here is every earthquake of magnitude 4.5 or greater that the USGS recorded anywhere on Earth between 1970 and 2025 — one catalogue, one magnitude threshold, one query, 290,431 events. Not a sample, not a region, not a window: all fifty-six calendar years, the whole planet, every event above the line, on an axis that starts at zero. There is no truncation, no break, no log scale and no second axis. Nothing has been smoothed, de-clustered or modelled; this is a count. The threshold has not moved in fifty-six years, and it is not an arbitrary one either — M4.5 is the magnitude the National Earthquake Information Center states as its goal for global completeness. And the finding is not subtle. The record holds 669 qualifying events in 1970 and 8,558 in 2025, a factor of 12.8. Even measured from the settled mid-1970s the count has more than doubled: the last two decades run above 7,000 events a year against a 1970s norm below 3,000. Seismicity is not stationary. The earthquake book is rated off a stationary-hazard assumption, and on this record that assumption is out of date. Recommendation: a flat global loading on the FY27 earthquake book, and re-rate the retrocession programme at renewal.

02The trick

The threshold on that slide never moved. What moved was the number of instruments listening at it — and a threshold is only as fixed as the network underneath it, because nothing gets counted until something hears it. Split the identical catalogue into magnitude bands and the rise turns out to have an address. Between the first decade of the modern record and the last, the band M4.5–4.9 goes from 1,837 events a year to 5,688, a factor of 3.10. M5.0–5.9 goes from 1,440 to 1,623 — a factor of 1.13. M6.0–6.9 goes from 87.0 to 119.9, M7.0 and above from 10.2 to 13.4, both differences comfortably inside the year-to-year scatter of counts that small. Of the +4,070 events a year the catalogue gained, +3,851 of them — 94.6% — are in the single half-magnitude band at the edge of what the world could hear. Everything above M5.0 put together contributes 219. If the crust had genuinely got busier, that would show at every magnitude; a rise that lives entirely in the faintest cases is a fact about the microphones. The catalogue then convicts itself in its own arithmetic. Gutenberg and Richter established in 1944 that earthquake frequency falls off by a roughly constant factor per magnitude unit — about tenfold, the exponent written b, and close to 1 nearly everywhere on Earth. That fixes the ratio between two thresholds: for every M5.0+ there should be about 3.16 events at M4.5+. Run it backwards on the 1973–82 catalogue and the implied b between 4.5 and 5.0 comes out at 0.68, a value the crust does not produce, and which means roughly 1,500 quakes a year happened and were never written down. Run it forward through the decades and that number climbs steadily to 1.25 — while the same calculation between M6.0 and M7.0, where the network has been complete since the 1960s, sits at 0.98 in the 1970s and 1.00 today and never moves. One pair of magnitudes drifts and the other does not; that is the bottom band filling in, decade by decade, and it is visible in the catalogue without reference to anything outside it. The slide’s steepest stretch, meanwhile, is not even a network story but a change of catalogue: the NEIC bulletin that carries the modern global record begins on 1 January 1973, and before that ComCat holds only what earlier, sparser catalogues happened to catch. The whole planet yields 15 events in the M4.5–4.9 band for the year 1970. It yields 6,429 for 2025. That seam is where the ×12.8 comes from; measured from 1973 the same series gives ×2.84. The uncomfortable part is how much the slide gets right on the way past: a zero-based axis, all fifty-six years, the whole planet, every event above the line, one catalogue, one constant threshold, no smoothing, every figure printed. Eight things right, and the threshold sitting at the edge of hearing was enough on its own. (The slide is our own demonstration, drawn in the manner of a reinsurance board pack. The counts are real, taken from ComCat.)

03The fix

Hold the ruler still. Pick a magnitude the network could clear for the whole window rather than the one it can clear today, and re-plot: at M5.0 and above the fifty-six-year series is nearly flat, and at M7.0 and above it is a mean of 13.8 events a year with a standard deviation of 3.9 and a range of 6 to 24 — which sits neatly against the USGS’s own long-run expectation from the record since 1900, about fifteen in the magnitude-7 range plus one of magnitude 8 or greater. That throws away 95% of the data, and what it throws away is the part that was really about the sensors. It is also the number a catastrophe model is rated on, and it has not moved, so the recommendation the slide made was priced off an instrument upgrade. Then plot the effort next to the count wherever the effort was measured: stations reporting, tests performed, people screened, inspectors employed, hospitals participating. Two lines rising together answer the question in one glance, and where a coverage figure exists you can go further and divide by it, turning a count into a rate against effort — cases per thousand screened, detections per station-year — which is the quantity the reader believed they were being shown all along. Where nothing measures the effort directly, the data usually carries an internal check like this one: some stable proportion the phenomenon is known to obey, which a change in coverage will bend and a change in the world will not. Most of all, be exact in the caption about what the count is a count of. “Recorded”, “reported”, “diagnosed”, “detected” and “confirmed” are all different words from “happened”, and the gap between them is not a caveat but the finding. None of this is a criticism of the catalogue, which is superb and honest about itself: the USGS publishes the warning in its own earthquake FAQ — the record holds more earthquakes “not because there are more earthquakes, but because there are more seismic instruments” — right next to the data the slide was built from. The check that catches this one anywhere takes ten seconds. Ask what had to happen for a case to enter the dataset, then ask whether that has got easier over the span of the chart. Detection almost never gets harder.