[klyrs]

You're looking at actuarial estimates. Overestimating unknown-unknowns is more pragmatic than underestimating them. No, the article doesn't show how that particular sausage is made; we're all out here guessing.