Ord, T. (2020). The Precipice: Existential Risk and the Future of Humanity. New York: Hachette; London: Bloomsbury.
Not read. The summary is drawn from the abstract or from what the work is standardly cited for. A reading should check it first.
Why this is here
The fullest estimate of existential risk, and an argument that obligations to the future include obligations inherited from the past. For QUESTION-299.
Read in its place
The work cannot be read here, so the record holds it at second hand.
Ord 2024, “The Precipice Revisited” (NOTE-1317). The author’s talk about five years on. It restates the book’s per-century estimates for its four largest risks (climate and nuclear 1 in 1,000, pandemics 1 in 30, unaligned AI 1 in 10), quotes the book on great-power war, and reports that the book’s AI chapter followed the Yudkowsky–Bostrom agent scenario, that its nuclear estimate stayed low because no known pathway leads from current arsenals to extinction, and that it proposed “existential security”. It does not restate the one-in-six total (Thorstad, NOTE-1308, reports it) and judges that no estimate has moved at its rounding. The LIT summary’s numbers fit. Missing still: the book’s moral case, its reasons “from the future, from what we owe the past, and from the present”, which this talk does not touch; the natural-risk estimates; the derivation of any number; and the book’s policy proposals.
Ord, “Shaping humanity’s longterm trajectory” (NOTE-1269). Ord’s chapter of 2023 (published 2025) is not about the book’s estimates; it formalises the book’s priority. It gives the book’s definition of existential risk (“irrevocably destroying humanity’s longterm potential, such as through extinction or an unrecoverable collapse of civilisation”, citing Bostrom 2013 and Ord 2020), the reason existential risk does not “wash out” (near-term, irreversible, preventable), the case from the future’s duration and quality, the rejection of pure time preference, and the place of value lock-in, which it locates in the book at pp. 153–8; it cites pp. 43–49 for the book’s statement of longtermism. Missing still: the book’s risk estimates and their derivation (the companion talk gives the headline numbers); its arguments from the past, the present and humanity’s virtues, of which the chapter has none; and anything on motivation or policy.