Technical supplement 02
Methods and assumptions.
Prof. Jean-Luc Annexe, Dr. Annette Preapproval
Institute for Precautionary Progress
We aim to be specific enough to be useful, while remaining sufficiently distant in time to avoid an immediate correction. The following procedures were applied consistently, except where consistency would have affected the results.
Stakeholder consultation
We consulted every organisation on the stakeholder register. Entry to the register requires evidence of previous consultation. The process therefore produced a consistent sample of stakeholders.
Precautionary modelling
All projected benefits were discounted for uncertainty. All projected risks were retained because uncertainty about a risk does not establish its absence. The model therefore recommends further assessment.
Proportionality review
Each proposed obligation was assessed against the cost of not having an obligation. Because that cost cannot be established in advance, the obligation was considered proportionate.
Democratic validation
The conclusions were approved by representatives accountable to institutions accountable to member states accountable to electorates. No single participant was accountable for the conclusions.
Pre-registration
The main prediction was registered in a document before the document was published. This rules out the possibility that the main prediction was added after publication. Other possibilities remain.
Peer review
The manuscript was circulated to the authors, who provided valuable feedback on their own sections. One reviewer suggested a different conclusion. We thank them for their time.
Conflicts of interest
The Institute's continued existence depends on the future remaining sufficiently uncertain to require further work. The authors do not consider this to have influenced their finding that more work is required.
Data availability
The scenario indicators are published directly alongside the narrative. The interactive compute forecast exposes its formula and assumptions. Narrative assumptions are distinguished from model assumptions in the accompanying notes.
All research supplements ↗