Anthropic Shadow: Observation Selection Effects and Human Extinction Risks

Abstract

We describe a significant practical consequence of taking anthropic biases into account in deriving predictions for rare stochastic catastrophic events. The risks associated with catastrophes such as asteroidal/cometary impacts, supervolcanic episodes, and explosions of supernovae/gamma-ray bursts are based on their observed frequencies. As a result, the frequencies of catastrophes that destroy or are otherwise incompatible with the existence of observers are systematically underestimated. We describe the consequences of this anthropic bias for estimation of catastrophic risks, and suggest some directions for future work.

Evaluating Methods for Estimating Existential Risks

“…The assessments are also vulnerable to another type of bias, known as the anthropic shadow. Ćirković, Sandberg, and Bostrom argue that risks of extinction events are underestimated because observers of actual extinction events cease to exist and cannot report their observations . Fourth, decisions need to be made regarding which events to place in set B.…”

“…Cirković, Sandberg, and Bostrom argue that risks of extinction events are underestimated because observers of actual extinction events cease to exist and cannot report their observations. (41) Fourth, decisions need to be made regarding which events to place in set B. Who decides whether set B is acceptable, and, similar to the holistic assessment approach, who ought to be asked to provide the probability assessments?…”

Small Theories and Large Risks—Is Risk Analysis Relevant for Epistemology?

“…OR: why have not we noticed traces of past such events?—fail to take into account observation‐selection effects. ( 38 ) However, the situation is even worse, epistemically speaking, since the publicized numerical bounds on risk probabilities often include theories intended to be tested , either explicitly or implicitly. The notion that the specific “risky” prediction constitutes sufficient grounds for falsification is hard to justify, both epistemologically and morally.…”

Anthropic arguments outside of cosmology and string theory

“…Therefore, it is subject to a specific bias called the anthropic shadow in Ćirković, Sandberg, and Bostrom (2010): a part of the parameter space of large catastrophic events is not adequately sampled, due to its incompatibility (or weaker compatibility, so to speak) with our existence as observers at present. In other words, the usual procedure, which does not distinguish between the directions of time (the distribution function is assumed to be the same, or similar, or obtained by the same procedure in the future as in the past) fails to account for the selection effects.…”