A thousand mornings of flawless evidence, and the one morning that outranks them all.
The lineage is three centuries deep. Hume (1748) asked what entitles us to step from bread that nourished us yesterday to bread that will nourish us tomorrow, and found habit where we wanted proof (An Enquiry Concerning Human Understanding, Section IV). Russell (1912) moved the problem into the farmyard: a chicken, fed every day of its life, expects the feeding hand right up to the morning that same hand wrings its neck (The Problems of Philosophy, ch. VI, “On Induction”). Taleb (2007) swapped Russell's chicken for a turkey and pinned the arithmetic: fed for 1,000 days, and the surprise arrives on day 1,001 (The Black Swan, Random House, ch. 4).
And the salt: the turkey's statistics were never bad. Laplace's rule of succession (Essai philosophique sur les probabilités, 1814; first derived in his 1774 memoir on the probability of causes) is honest maths — after n good mornings with no bad ones, the chance of one more is (n+1)/(n+2), and after 1,000 it says 1001/1002, about 99.9002%. The calculation was right; the reference class was wrong. It counted mornings the turkey observed, not mornings the butcher scheduled, and the hazard lived only in the second list — so every new day of data made the turkey more certain and not one day safer. The fatal day is the day the record feels safest, and no pile of yesterdays proves tomorrow. the lying average →