Scan: When will a supervolcano next erupt?

← Back to scans

When will a supervolcano next erupt? — Before Jan 1, 2050

KXERUPTSUPER-0-50JAN01 · Climate and Weather · 2026-05-15

Agent

24%

Market Price

-17.9%

Edge

70%

Confidence

Volume: 29,107

Spread: 2.7c

Days to resolution: 8631

Markets in event: 1

Final Rationale

The geological base rate for VEI 8 supervolcano eruptions over 24 years is well under 1%, and even VEI 7+ interpretations yield only 1-5%; no supervolcano has erupted in recorded human history. I partially accept the Devil's Advocate critique on definitional ambiguity and the zero-spread ensemble suggesting false precision, adjusting from 5% to 6% to account for resolution risk if Kalshi uses a looser 'supervolcano' interpretation, but reject the suggestion to go higher because Campi Flegrei's centuries-strong unrest historically resolves without VEI 7+ eruptions and major quiet systems (Yellowstone, Toba, Taupo) show no precursors. I diverge sharply from the 21% market because it's an illiquid, wide-spread (5-88%) market dominated by entertainment traders rather than geoscience-informed participants — the price reflects vibes about 'supervolcanoes erupting sometime' rather than the actual geological base rate. The market signal here is weak enough that scientific priors should dominate.

Price vs. prediction

How the market moved after the agent's call. Solid blue line is the actual market price; dashed line is the agent's static forecast at $t{=}0$.

$n = 58$ follow-ups

Re-scan Context

This market has been scanned before. Previous predictions:

Date	Predicted	Market Price	Confidence
2026-05-08	6%	19%	70%
2026-04-30	4%	25%	70%
2026-04-04	4%	32%	50%

1. Planner Opus · Fermi decomposition + tool plan

Tools chosen: wikipedia_lookup article_search web_search kalshi_data code_execution

Sub-questions (Fermi decomposition)

Is the current base rate of supervolcano eruptions (VEI 8 or equivalent) consistent with at least one eruption occurring in any given 24-year window? (w=0.45)
Are any of the world's known supervolcanic systems (Yellowstone, Toba, Campi Flegrei, etc.) currently showing elevated unrest or precursor activity that meaningfully raises the probability of eruption before 2050? (w=0.40)
Does scientific consensus indicate that eruption forecasting models for supervolcanoes provide meaningful advance warning sufficient to shift short-term probability estimates above base rates? (w=0.15)

Planner reasoning

This question requires estimating the probability of a supervolcano eruption (typically defined as VEI 8, or sometimes VEI 7+) occurring within a ~24-year window (2026–2050). **Reference class and base rates**: The geological record suggests VEI 8 eruptions occur roughly once every 50,000–140,000 years (some estimates as low as 1 per million years for the very largest). Using a Poisson model, the base rate probability of at least one VEI 8 eruption in 24 years is extraordinarily small (~0.01–0.05%). If the question uses a broader definition (VEI 7+), recurrence intervals shorten to ~700–1,000 years, giving a somewhat higher but still very small probability (~2–3%). **Causal mechanisms and current monitoring**: The key question is whether any known supervolcanic system is showing anomalous precursors. Campi Flegrei (Italy) has experienced elevated bradyseismic unrest in recent years. Yellowstone is continuously monitored with no unusual escalation beyond normal background. The scientific literature emphasizes that even elevated unrest rarely precedes a major eruption and that current monitoring cannot reliably predict eruptions decades in advance. **Temporal analysis**: For YES to occur, a supervolcanic system would need to undergo a full eruption sequence within ~24 years. Even in the most optimistic unrest scenarios (Campi Flegrei), scientists estimate the probability of a major eruption as very low. The 24-year window is extremely short relative to geological timescales. No credible scientific body has issued warnings suggesting elevated short-term risk. **What makes this different**: Unlike many geopolitical or economic questions, this is dominated by an extremely stable geological base rate. No news event or human decision meaningfully changes the probability. The key uncertainty is definitional (VEI 7 vs VEI 8) and whether any system is in a late-stage pre-eruptive state that monitoring hasn't yet detected. Market prices on Kalshi will be informative about crowd wisdom on this ultra-low-probability event.

2. Research Parallel tool calls · legacy v2

Tool	Status	Time	Items
wikipedia_lookup	OK	1.8s	-
article_search	OK	0.7s	-
web_search	OK	46.4s	-
kalshi_data	OK	1.4s	-
code_execution	OK	0.0s	-

3. Evidence Synthesis Sonnet · 13 items · legacy v2

#	Strength	Credibility	Direction	Source	Claim	Priced In
1	STRONG	85	↓ DOWN	code_execution	Poisson model calculations show VEI 8 eruption probability in a 24-year window is 0.017%–0.048% depending on recurrence interval assumptions (50k–140k years). VEI 7+ events give a higher but still small 1.2%–4.7% probability using 500–2,000 year recurrence intervals.	Yes
2	STRONG	90	↓ DOWN	web_search	The most recent confirmed VEI 8 eruption was Toba approximately 74,000 years ago; no VEI 8 eruption has occurred in human history, consistent with recurrence intervals of tens of thousands to hundreds of thousands of years.	Yes
3	MODERATE	80	↓ DOWN	wikipedia_lookup	The largest volcanic eruption of the 21st century is the 2022 Hunga Tonga–Hunga Ha'apai eruption, which was not a supervolcano event (VEI 5–6), confirming no supervolcanic eruption has occurred in recorded history.	Yes
4	STRONG	78	↑ UP	article_search	Campi Flegrei (Italy) has been experiencing its strongest recorded earthquakes in 2025, including events in March and June that forced evacuations, with the INGV director describing unrest not seen for centuries.	No
5	STRONG	92	↓ DOWN	web_search	Yellowstone caldera activity remains at background levels as of April–May 2026, with only 97 located earthquakes in April (largest M2.5) and no significant uplift or subsidence detected since January 2026.	Yes
6	MODERATE	82	NEUTRAL	article_search	Echinus Geyser at Yellowstone reawakened after years of dormancy in early 2026 (~40 eruptions in February, declining to 4 in April), but this is a geyser phenomenon unrelated to caldera magmatic unrest and carries no supervolcano eruption implication.	Yes
7	MODERATE	72	NEUTRAL	web_search	Axial Seamount (submarine volcano, Pacific) is forecast to erupt by mid-to-late 2026, but scientists confirm it poses no danger to human life and is not a supervolcanic system capable of a VEI 7+ eruption.	Yes
8	MODERATE	65	↑ UP	web_search	Campi Flegrei unrest data as of April 2026 (sourced from arXiv preprint) indicates significant ongoing bradyseismic activity, but no scientific body has issued an imminent eruption warning or elevated the probability of a supereruption in the near term.	No
9	STRONG	88	↓ DOWN	web_search	No credible scientific consensus or official warning of an imminent supervolcano eruption exists for 2026 or any near-term period through 2050.	Yes
10	STRONG	78	↓ DOWN	web_search	Scientists note that predicting volcanic eruptions weeks or even months in advance is 'nigh-impossible' for most volcanoes; the only reason Axial Seamount could be forecast is exceptional monitoring density, unavailable for most supervolcanic systems.	Yes
11	MODERATE	70	↓ DOWN	article_search	Scientific literature (implied by Campi Flegrei reporting) indicates that even elevated bradyseismic unrest rarely precedes a major eruption, and current monitoring cannot reliably predict supervolcanic eruptions decades in advance.	Yes
12	STRONG	60	↑ UP	kalshi_data	Kalshi prediction market currently prices this event at 21.3%, far above the scientifically-derived base rate of <5% even under generous VEI 7+ assumptions, suggesting significant market mispricing, entertainment-driven demand, or a much broader definitional scope than strict VEI 8.	Yes
13	MODERATE	55	NEUTRAL	kalshi_data	Kalshi market price has ranged from 5% to 88% over 194 days with an average daily volume of 150 contracts, suggesting thin liquidity and high volatility that undermines the market's reliability as a probability signal.	Yes

Information Gaps

No detailed recent data on Campi Flegrei's current alert level or official Italian Civil Protection risk assessment — the web search result was truncated before completing the Campi Flegrei April 2026 summary.
No data on the current status of Toba (Indonesia), Taupo (New Zealand), or other major supervolcanic systems beyond Yellowstone and Campi Flegrei.
No peer-reviewed quantitative estimate of Campi Flegrei's probability of VEI 7+ eruption in the next 24 years from volcanologists.
No clarity on the question's precise definition of 'supervolcano eruption' — whether VEI 7 or strictly VEI 8 is required, which changes the base rate by 2–3 orders of magnitude.
No data on Long Valley Caldera (California), Aira Caldera (Japan), or other active supervolcanic systems' current monitoring status.
No explanation for why the Kalshi market is priced at 21.3% — whether this reflects definitional ambiguity, entertainment bias, or genuine crowd assessment of supervolcano risk.

Key Uncertainties

Definition ambiguity: If 'supervolcano eruption' includes VEI 7 events, the base rate probability is ~2-5% over 24 years rather than <0.05% for strict VEI 8 — a 100x difference that dominates all other considerations.
Whether Campi Flegrei is in a genuinely late-stage pre-eruptive state that monitoring has not yet fully characterized — this is the single most plausible path to a near-term supereruption.
Unknown supervolcanic systems: There may be poorly monitored or undiscovered supervolcanic systems that could erupt without warning.
Whether the Kalshi market at 21.3% reflects genuine crowd epistemic wisdom or is driven by entertainment/novelty buyers with no geological knowledge.
Temporal clustering: Geological record shows supervolcano eruptions may not be Poisson-distributed; if Earth is currently in an active phase, base rates could be higher than time-averaged estimates.

Market Context

Price: 0.213 · 7d: +0.019 · 30d: -0.004
Volume: LOW — 150 contracts/day is very thin for a meaningful prediction market · Depth: SHALLOW
Reliability: LOW — extremely wide price range (5%-88%) over 194 days and very low volume suggest this market is driven by casual or entertainment participants rather than informed geoscience forecasters. The 21.3% price is likely a severe overestimate relative to scientifically-grounded base rates.

4. Independent Forecaster Opus · No market price · legacy v2 Did NOT see market price

Independent estimate: 5% (70% confidence)

base rate

70% conf

evidence driven

70% conf

contrarian

70% conf

Spread

0pp

Low disagreement

sq1: Is the current base rate of supervolcano eruptions consistent with at least one eruption in any 24-year window?

Strict VEI 8 base rate over 24 years is 0.017-0.048% using Poisson with 50k-140k year recurrence. VEI 7+ interpretation gives 1.2-4.7%. Weighting toward strict interpretation (VEI 8 is the standard definition of 'supervolcano') but allowing some probability for looser resolution: ~2%.

sq2: Are any known supervolcanic systems showing elevated unrest meaningfully raising eruption probability before 2050?

Campi Flegrei is the standout: strongest recorded earthquakes in 2025, INGV describes unrest unseen for centuries. However, Campi Flegrei is typically classified as capable of VEI 7 eruptions (Campanian Ignimbrite was ~VEI 7), and even strong bradyseismic unrest historically resolves without major eruption (1538 Monte Nuovo was VEI 2). Yellowstone is at background. No official imminent eruption warnings. This raises probability modestly above pure base rate to ~6%.

sq3: Does scientific consensus indicate forecasting models can shift short-term probabilities meaningfully above base rates?

Strong consensus that supervolcano eruption prediction is essentially impossible decades in advance. Monitoring detects unrest but cannot reliably predict eruption timing. This question essentially asks whether models elevate probability above base rate - answer is largely no. ~5%.

Show full reasoning trace

base rate: VEI 8 eruptions occur every 50,000-140,000 years, giving 0.017-0.048% probability over 24 years. VEI 7+ events give 1.2-4.7%. Last VEI 8 was Toba ~74,000 years ago; no supervolcano eruption in recorded human history.

evidence updates: Campi Flegrei unrest in 2025-2026 pushes UP by ~3-4% (strongest in centuries, but historical analogs rarely produce major eruptions). Yellowstone background activity is neutral/down. Scientific consensus on prediction impossibility pushes DOWN by keeping us anchored to base rates. Definitional ambiguity on Kalshi resolution adds ~1-2% upward.

combination method: Weighted average across sub-questions with sq1 base rate dominant at 45% weight, sq2 system-specific evidence at 40%, sq3 forecasting capability at 15%.

final: Final estimate: ~5%. This balances the extremely low geological base rate (which would suggest <1%) against (a) Campi Flegrei's notable unrest, (b) definitional ambiguity that could allow VEI 7 to resolve YES, and (c) the 24-year window being long enough that tail risks accumulate. I would not go below 3% due to resolution risk, nor above 8% without a credible scientific warning.

ensemble: {'ensemble_count': 3, 'perspectives': {'base_rate': {'p_yes': 0.05, 'confidence': 0.7}, 'evidence_driven': {'p_yes': 0.05, 'confidence': 0.7}, 'contrarian': {'p_yes': 0.05, 'confidence': 0.7}}, 'spread': 0.0, 'individual_p_yes': {'base_rate': 0.05, 'evidence_driven': 0.05, 'contrarian': 0.05}}

5. Devil's Advocate Sonnet · legacy v2 Market price revealed

Missing Info Reasoning Flaw Overconfidence

Challenges

The combination method is mathematically opaque. A 'weighted average' of three sub-questions that don't represent mutually exclusive or exhaustive pathways isn't a coherent probability decomposition. The actual question is simply P(supervolcano eruption by 2050), which should be computed directly from base rate + evidence updates, not by averaging conceptually heterogeneous sub-probabilities.
The resolution criterion ambiguity is identified as a critical information gap (changes base rate by 2-3 orders of magnitude) but is only given a ~1-2% bump. If there's meaningful probability (say 20-30%) the question resolves on VEI 7+ rather than strict VEI 8, and VEI 7+ base rate is 1.2-4.7%, this alone could justify 1-2% additional probability mass — but if the resolver uses an even looser definition (e.g., any caldera-forming event or media-labeled 'supervolcano'), the probability could be substantially higher. This definitional uncertainty deserves more weight given 8631 days to resolution.
With 24 years to resolution, tail risks from currently-quiet systems (Taupo, Toba, Aira, Long Valley) are essentially unmodeled. The forecaster acknowledges no data on these but doesn't widen uncertainty accordingly. The base rate already incorporates these globally, but the forecaster's confidence of 0.7 seems high given acknowledged information gaps.
The forecaster correctly identifies the market as low-volume and unreliable, but the market's 21% does encode something — possibly the definitional ambiguity risk or popular interpretation of 'supervolcano' that includes VEI 6-7 events. Dismissing it entirely as 'entertainment bias' without considering it may reflect a looser resolution interpretation is asymmetric reasoning.
Ensemble shows zero spread (all three perspectives gave exactly 0.05), which suggests the perspectives weren't genuinely independent — a red flag for false precision/confidence.

Suggested adjustment: +2pp

Modest upward adjustment (to ~7%) warranted because: (1) definitional ambiguity on 'supervolcano' is underweighted given the 24-year window and Kalshi's potentially loose resolution standards, (2) information gaps on multiple major caldera systems should widen uncertainty upward more than downward given fat-tailed geological risk, and (3) the zero-spread ensemble suggests false confidence. This is not adjusting toward the market — it's correcting for under-weighted resolution risk.

6. Aggregator Opus · final calibrated forecast

6% (70% confidence)