Four regimes of uncertainty

The argument

Waves suggest gradual, cyclical, and predictable accumulation. Digital rain is dense and distributed, with countless small signals arriving simultaneously, requiring meaningful patterns to be separated from noise. A volcanic eruption or a meteorite represents sudden discontinuity, a moment when the existing model no longer explains reality.

Taken together, it is an argument about different regimes of uncertainty.

• Accumulation

  Waves to the shore

  Gradual, cyclical, predictable. Each wave carries one note, and the next one is much like the last.

• Distributed signal

  Digital rain

  Dense and simultaneous. Forty-eight notes at once, and the work is separating pattern from noise.

• Endogenous discontinuity

  Volcano

  Pressure builds inside the system. The warning signs are there to be read, and the collapse makes sense afterwards.

• Exogenous discontinuity

  Meteorite

  It arrives from outside the frame, without precursor. Watching more closely cannot help.

Why the two catastrophes are not the same claim

The distinction is built

Both catastrophe pieces carry the same instrument panel, and the difference between the two panels is the argument.

In the volcano, the hazard of eruption is a rising function of an observable state. Magma accumulates in a shallow reservoir, and the uplift of the ground above it and the seismicity near the crater row both climb with the accumulated volume. The panel plots that volume while it fills, so the eruption can be watched becoming likely. Someone who watches can learn to call it.

In the meteorite, arrival times are drawn from a memoryless process at a fixed rate. The hazard is constant, so the expected wait to the next impact is the same at every instant, however long the sky has been empty and whatever is on screen. The same panel plots a flat line. There is no precursor to find. That is a property of the process, so studying it harder cannot help.

One piece rewards attention. The other cannot, by construction.

The volcano is Svartsengi

The volcano was the weaker piece while it ran on invented units, so it now follows a real system: the magma reservoir beneath Svartsengi on the Reykjanes peninsula, which has fed repeated eruptions along the Sundhnúkur crater row since December 2023. Magma gathers at roughly four to five kilometres depth, the ground above rises measurably, a dyke travels to the crater row and erupts, the reservoir drains, and the filling starts again. The precursor is not hypothetical there. It was measured, published and acted on, many times over.

The detail that earns the piece its place is the threshold. It is not a line. Geodetic models put the volume needed for the early events at roughly 8 to 13 million cubic metres, and later in the series the estimate was revised upward, to roughly 12 to 15 million. The magma inflow rate itself declined gradually. Forecasts were published as rising probability over a window, never as a date.

So the piece draws the threshold as a shaded band, erupts at a volume drawn from inside it, moves the band between cycles, and afterwards reports where in its band that eruption actually fell. Watch it for a few minutes and both things are true at once: the precursor works, and the gate stays uncertain. A warning you can read is a different thing from a date you can name, and a piece that erupted on a crisp line would be making a promise the real monitoring never made.

The meteorite is Söderfjärden

The crater in the fourth piece is modelled on Söderfjärden, about ten kilometres south of Vaasa in Ostrobothnia. The piece runs deep time after the strike, so the visitor sees the event and the record it leaves.

What Söderfjärden adds to the argument is the second half of the exogenous case. The event could not have been predicted, and it also could not be read for a very long time. From the ground the structure is a flat plain. The circle was noticed in Landsat imagery in 1975, and the impact origin was only confirmed in 1985, on shock evidence: planar deformation features in quartz, kink bands in mica, polymict breccia in the drill cores.

Its age is still contested. Argon dating of clast-rich impact melt rock gives a range of 640 to 1880 million years, against an older reading that placed the impact in the Cambrian. So the exogenous event was unknowable in advance, and then unreadable for a very long time. It became legible from orbit, on evidence someone had to go looking for, and its date is still open. The piece reports the disagreement.

The colours

Every piece in the series maps musical pitch into a perceptually audited colour space with the Palettecore engine: hue is the pitch class, lightness the absolute pitch. The colours are generated and audited on the exact values each artwork displays, so no piece can drift from its numbers.

The two catastrophes use the four-class ladder, the one that stays separable under simulated colour-vision deficiency, because these are the pieces whose subject is whether a signal could have been read. The digital rain uses the full chromatic code, where the collapse is the subject: it clears the exploratory floor in normal vision and fails under every simulated dichromacy. None of these is an accessible encoding, and each piece reports its own audit on screen.

The series suits hospitals, libraries, university spaces, digital art festivals, science museums, and music museums with sound-art programmes. Contact hello@vahtian.com.

The two catastrophes

• Volcano

  Magma accumulates, the ground above it rises, and the instrument shows the eruption becoming likely. The threshold is drawn as a band, and it moves between cycles. Sound optional.

• Meteorite

  Nothing to read, then the record. A quiet sky, impacts drawn from a memoryless process, and a crater that fills and buries while the same instrument shows a flat line. Sound optional.

The pieces open fullscreen and the interface fades when the pointer rests. The other two are at Waves to the shore and Digital rain.