A secret expert vote became the scientific foundation for decades of carbon-capture policy

In 2005, scientists finalising the Intergovernmental Panel on Climate Change's first major report on carbon capture and storage faced a basic problem: there was too little empirical evidence to quantify long-term leakage. Only one large project had rigorously tested permanent storage, and its injected carbon had unexpectedly risen 700 feet before meeting a shale barrier. An early draft warned that a broad claim about leakage would be almost meaningless.

The authors nevertheless produced a number. At a meeting in Oviedo, they anonymously voted on what they expected future evidence to show and agreed not to disclose the exercise. Five months later, the IPCC said well-selected and managed sites were likely to retain 99% of injected carbon for at least 1,000 years, that there were no major technical or knowledge barriers, and that carbon capture could supply as much as 55% of the century's climate-mitigation effort. The report did not tell readers that its central confidence estimate came from the off-record elicitation rather than published literature.

ProPublica and Drilled reconstructed the process through records and interviews with more than 100 experts, including a dozen meeting participants. Saudi Arabia and other exporters had pressed for the report; Shell, Chevron and Saudi Aramco representatives joined the chapter team. That does not establish that the scientists fabricated a conclusion. Chapter leader Sally Benson says the process synthesised a large body of underlying data, was known to IPCC leadership and did not violate its rules. Other IPCC veterans told the reporters that using an undisclosed survey to assess literature was inappropriate.

The methodological choice became policy infrastructure. The report has been cited thousands of times, embedded in climate models and used to justify public subsidies and net-zero plans. Yet only 12 large geological-storage projects have operated since 1996; together they bury less carbon than one power station emits annually, and seven encountered technical, financial or performance problems. Carbon capture may still be necessary. The investigation shows why a technology's uncertainty must remain visible when governments build entire transition pathways on top of it.