Hoare examines why software had become highly reliable despite little use of formal proof, and how mathematical theory and engineering practice reinforce one another.