[climate · public services · service design]
The warning arrived but nobody could leave
The warning worked.
The message arrived.
The person stayed.
What went wrong?
Researchers led by the University of Bristol have examined why early-warning systems across southern Africa do not always translate into protective action. Their findings, reported on 2 November in the Bulletin of the American Meteorological Society, expose a problem familiar to service designers.
Delivering information is not the same as enabling somebody to use it.
Household surveys in Madagascar, Malawi and Mozambique found that 90, 80 and 77 per cent of respondents respectively had received warnings about approaching tropical cyclones.
Yet fewer than half evacuated.
That difference is the story.
A conventional evaluation might count warnings issued, messages delivered, forecast accuracy and the time between alert and impact.
All are important.
But none tells us whether somebody can actually leave.
The researchers identify several obstacles: warnings arriving too late, scepticism about their content, limited resources and what they call involuntary immobility.
That last phrase deserves attention.
A person may understand the danger perfectly and still have no transport, money, safe destination or practical means of moving dependants and possessions.
The problem is not necessarily communication.
It may be capability.
This changes the unit of analysis.
Instead of treating an early-warning system as a meteorological instrument connected to a messaging service, consider the entire sequence from recognising danger to reaching safety.
Forecast.
Interpretation.
Trust.
Decision.
Transport.
Shelter.
Recovery.
A warning is only one component.
The research team proposes six priorities, including stress-testing warning systems, strengthening national forecasting services, integrating scientific and local knowledge, community participation and better coordination between agencies.
They also argue that systems must prepare for unprecedented events, rather than relying exclusively on historical experience.
That introduces another difficulty.
Communities often possess sophisticated knowledge of familiar hazards. Climate change can produce events beyond the conditions in which that knowledge developed.
Neither scientific forecasts nor local experience is sufficient alone.
The useful system must connect them.
There is an important methodological question here too.
When a warning is followed by no evacuation, researchers must resist treating the outcome as evidence of indifference or irrational behaviour.
The apparent failure might be an entirely rational response to circumstances outside the individual's control.
Perhaps the nearest shelter is unsafe.
Perhaps leaving means losing income.
Perhaps an earlier warning proved inaccurate.
Perhaps there is nowhere to go.
Those are research questions, not excuses.
The United Nations' Early Warnings for All initiative aims to extend protection through warning systems worldwide by the end of 2027. The Bristol study suggests that achieving coverage will not, by itself, demonstrate that people are protected.
The distinction is fundamental.
A service can be delivered successfully and still fail its purpose.
Research Today has repeatedly encountered this boundary: between an interface and the activity it is supposed to support, between an intervention and its consequences.
Today's example is particularly stark.
The warning arrived.
The person could not leave.
If an intervention succeeds technically but people cannot act upon it, where should responsibility for the outcome sit?
