The Black Swan: The Impact of the Highly Improbable

by Nassim Nicholas Taleb

am skeptical about confirmation—though only when errors are costly—not about disconfirmation. Having plenty of data will not provide confirmation, but a single instance can disconfirm.