This seems so obvious that I’m almost sure I’m missing the counterargument but my central problem with existential risk people is that they cloak their analysis in the language of probabilities but how can these probabilities be properly assessed? You can’t run experiments.