One Planet,
No Controlled Rerun

The storm stalls over Aster Bay

A persistent storm brings intense rainfall. Rivers rise, the saturated ground sheds water quickly, and Aster Bay floods.

What made this event so severe?

The event is the evidence, not yet the explanation.

Several things changed together

Warm water, unusual circulation, abundant moisture, wet soil, and local land conditions were all present. Their co-occurrence describes the event, but does not isolate what difference any one factor made.

Imagine the experiment we wish we could run

Keep the storm setup, circulation, season, and land state comparable. Change the ocean condition, then compare rainfall.

But Earth follows only one history

We observe rainfall under the ocean condition that occurred. We do not simultaneously observe rainfall under the alternative condition.

Where can causal information come from?

We cannot observe the missing Earth directly. But variation, timing, causal structure, and physical knowledge can constrain what the alternative would have implied.

Model experiments are one possible source of evidence. They are not causal inference by themselves.

Relationships can play different causal roles

Ocean state, circulation, moisture, rainfall, soil, and flooding form pathways, not a flat list of correlated variables.

Does ocean state influence rainfall through moisture, or do both mainly reflect circulation?

Not every causal question can be answered from the available data

Different causal structures can produce similar observed correlations. More measurements help only when they distinguish the competing explanations.

Can the observations distinguish them?

Sometimes yes, sometimes only with additional assumptions, and sometimes not. Then the causal effect is not identifiable from the available information.