CG-139Humanity & Planetary InfluenceOpen access

When Millions of Phones Become a Seismic Network

One phone trembling on a table proves nothing. Thousands of stationary phones registering the same first pulse can locate an earthquake before the most damaging wave reaches distant cities.

Human report: measured effectsPhone: accelerometerNetwork: consensusLimit: warning ≠ prediction
Interactive explanatory model

At what point does noise become evidence?

UNVERIFIED MOTION1 deviceinsufficient evidence

A conceptual reconstruction, not live earthquake data. Real systems combine many devices, timing, location, signal shape and independent seismic information.

In 1811 there were no seismographs in the Mississippi Valley. The New Madrid sequence was reconstructed from letters, damage, changes in the river and layers of sand forced through the ground. People were not merely witnesses: their reports became measurements after later scientists compared, classified and mapped them.

A person is not a calibrated seismometer. But a structured population of reports can measure effects that a seismometer does not: fear, falling objects, broken walls and the boundary between “felt” and “not felt.”

When testimony became a data product

The U.S. Geological Survey's Did You Feel It? system asks people what they felt and what happened around them. It converts answers about rattling objects, movement of furniture, difficulty standing and building damage into an aggregated intensity for an area.

The crucial operation is aggregation. Individual responses remain noisy and unchecked. Many reports from the same area can converge on a community intensity. In regions with few seismic stations, these human-derived values can help constrain ShakeMap; where instruments are dense, human reports help relate measured ground motion to experienced intensity.

01Experience“The cupboard opened.”
02QuestionnaireEffects become categories.
03AggregationResponses become an area value.
04ComparisonPeople and instruments calibrate the map.
Field record 01 · fictional Code Gaia research layer

At 09:17 the instrument map still had a pale gap west of the ridge. At 09:19, 412 questionnaires had arrived from that gap.

They did not agree on the sound. They agreed on what fell from the shelves.

The blank area was not quiet. It was simply under-instrumented.

Dr Lena Orlov · rapid impact desk · extract 19-BFictional record. The method it describes is based on documented DYFI aggregation and ShakeMap use.

The seismometer already in your pocket

A smartphone contains an accelerometer—the component that notices rotation and movement. When a stationary phone detects motion resembling the first, faster P-wave, it can send a coarse location and signal to a server. One phone may have been bumped. Many phones reporting a compatible pulse across space and time create a different proposition.

The Android Earthquake Alerts system uses this distributed principle. According to the system's 2025 scientific report, it had detected more than 18,000 earthquakes and issued alerts for more than 2,000 events by the end of the study period. Berkeley's MyShake project likewise turns participating phones into sensors while combining citizen reports with authoritative earthquake information.

RUPTUREPDATA / ALERTSCITY

Electronic communication can outrun seismic waves. It cannot outrun a wave that has already reached the phone, and people near the epicentre may receive little or no warning.

Field record 02 · fictional Code Gaia research layer

Device 8,104 moved. Then 8,110. Then forty-seven more.

The first decision was to distrust them. A freight train, a concert, a dropped charging rack—cities manufacture synchrony.

Then the front advanced outward at seismic speed. Noise had acquired geometry.

Arun Mehta · distributed detection log · 03:41:12 UTCFictional record. It illustrates the real need to reject non-earthquake motion and confirm a spatially coherent signal.

The network is the instrument

The individual component may be cheap, inconsistent or subjective. The scientific instrument appears one level higher: in the protocol that compares thousands of imperfect observations.

LayerWhat it sensesWhat can go wrong
WitnessExperienced shaking and damageMemory, wording and unequal participation.
PhoneLocal accelerationMovement, machinery and poor contact with the ground.
PlatformAgreement across many observationsCoverage follows devices, connectivity and population.
Scientific networkLocation, magnitude, intensity and uncertaintySpeed competes with accuracy; false and missed alerts remain possible.

What this does not mean

It detectsshaking that has begun

Early warning is possible because information can travel faster than damaging seismic waves.

It does notpredict the next earthquake

The network cannot state the date and place of a rupture before it starts.

It expandscoverage and impact evidence

Especially where conventional instruments are sparse.

It also inheritshuman inequality

No phone, power, data connection or safe reporting channel means no signal in the network.

The unsettling conclusion

Human civilisation has built a second skin of accelerometers, cameras, vehicles, power grids, reports and databases. Much of it was created for other purposes. Under the right protocol, fragments of that infrastructure can become instruments for observing the planet.

The sensor is no longer always an object. Sometimes it is a rule for deciding when thousands of unreliable objects agree.
The next system is open

Can a living system alter the conditions of its own survival?

CG-012 · PLANETARY SYSTEMSThe Amazon Makes Rain for Itself

Phones become an observing network when separate sensors agree. In the Amazon, billions of trees form a network that returns water to the air and passes it deeper into the continent.

Open the next dossier →
Question before proceeding

If each tree changes the water flux only slightly, when does their sum become climate?