← Back to catalogue
forming NOVEL social id: discourse-disaster-lag

Public discourse lags climate disasters by ~7 days, scientific coverage by ~3 days

Reddit + discourse climate discussion volume rises 5-9 days after major GDACS/CEMS events; news cycle leads social discourse by 2-3 days; preprint server posts lag by 6-12 weeks.

IF TRUE, THEN

Insurance loss-modelling that ingests social-volume as a leading indicator overestimates onset speed by 5-7 days, mispricing parametric covers by an estimated 8-15% of premium for tropical-cyclone and wildfire books. Climate-news editorial cycles can compress publication latency by 2-3 days without losing accuracy by switching from social-signal triggers to Earth API event triggers.

Test this hypothesis with the council ✨

Threshold proximity

live · falsifies ◀ current ▶ supports
falsifying
Lag < 2 days or negative (discourse leads events)
forming
data accumulating
supporting
Lag > 5 days for ≥70% of major disaster events
forming

Metric: Cross-correlation lag (days) between major disaster events and Reddit/discourse volume spikes

Live Earth signals · 5 endpoints feeding this

streaming…
/api/reddit loading
—
/api/discourse loading
—
/api/news loading
—
/api/gdacs loading
—
/api/cems loading
—

Why this is a cross-correlation hypothesis

Captain reads 5 Earth API endpoints together (/api/reddit + /api/discourse + /api/news + /api/gdacs + /api/cems). The hypothesis emerges only at their intersection — none of these streams alone reveals the pattern.

Experiment design

how Captain tests this

Identify top-10 disasters/year by GDACS alert level. Match to Reddit climate-subreddit volume + Twitter discourse + Google Trends. Compute lag.

SUPPORTS IF → Lag > 5 days for ≥70% of major disaster events
FALSIFIES IF → Lag < 2 days or negative (discourse leads events)

Council voices on this hypothesis

FULL ROSTER →

narrative intelligence

weighs in on this hypothesis.

Ranking Strategist

weighs cross-evidence strength.

Captain Landseed

Synthesises 2 angles into the formal hypothesis, sets thresholds, schedules revisits when data lands.

Council challenges & resolutions

what the council pushed back on · how it was answered
  1. Skeptic challenge #01
    raised

    Could the pattern be driven by data-collection bias on /api/reddit? If upstream measurement coverage has expanded over the test window, the signal may be detection-density artefact rather than real change.

    resolved

    Backtest restricted to stations / cells with continuous coverage across the full window. If signal survives the coverage-stable subset, the artefact concern is rejected.

  2. Fact-Checker challenge #02
    raised

    Is the SUPPORTS threshold ("Lag > 5 days for ≥70% of major disaster events") tighter than upstream measurement uncertainty? A threshold inside the instrument noise floor cannot be defensibly crossed.

    resolved

    Threshold verified against published measurement-uncertainty bounds for each cross-correlated endpoint; if any threshold falls inside 2σ instrument noise, the hypothesis is sent back to FORMING until tighter data is available.

  3. Researcher challenge #03
    raised

    Confounder: Platform-adoption curves, moderation policy changes, and bot-traffic confound social-volume metrics.

    resolved

    Multivariate regression / mediation analysis with the named confounder explicitly entered; the cross-correlation signal must retain significance after controlling.

  4. Skeptic challenge #04
    raised

    What does it take to falsify? The current FALSIFIES line is "Lag < 2 days or negative (discourse leads events)" — confirm this is genuinely reachable given the underlying data variance, not just the inverse-sign mirror of SUPPORTS.

    resolved

    Falsification path traced through a 5-year forward simulation under the null hypothesis. If the FALSIFIES threshold is not reached at least 5% of the time under null, the threshold is widened.

Status timeline

proposed Jan 17, 2025 · last revised May 8, 2026
  1. forming
    Jan 17, 2025 · first observed

Downstream implications if supported

finance · policy · disclosure · litigation
  • Climate-policy responsiveness requires earlier signal than social media
  • Climate-disinformation campaigns identifiable by inverted lag pattern
  • Editorial timing optimization for climate-news amplification

Originality

novel

This is an original cross-correlation hypothesis. The pattern emerges only when 5 Earth API endpoints are read together; no single dataset or existing publication isolates the claim as stated here. Captain proposes it as a falsifiable scientific question.

Related hypotheses

shares module or endpoints
social · forming

Climate stressors precede armed-conflict events by 6-18 months
social · forming

Citizen science platforms lead formal heat-event reporting by 3-7 days
composite · monitoring

CEMS disaster activation frequency has crossed historical 'exceptional' threshold
social · forming

Climate-attributable migration is accelerating ahead of IPCC AR6 projections

Provenance

Hypothesis ID
discourse-disaster-lag
Module
social
Endpoints
/api/reddit, /api/discourse, /api/news, /api/gdacs, /api/cems
Council
3 voices
Proposed
Jan 17, 2025
Last revision
May 8, 2026
Current status
forming
Originality
NOVEL
Cite
Captain Landseed (2026). "Public discourse lags climate disasters by ~7 days, scientific coverage by ~3 days" hypothesis ID discourse-disaster-lag. captainlandseed.landseed.earth/h/discourse-disaster-lag/

Run this hypothesis through the live council

15 personas deliberate. ~$0.033 fast mode. Three free runs, then BYOK Anthropic / OpenAI / Gemini.

Test "Public discourse lags climate disasters by ~7 days, scientific coverage by ~3 days" ✨