When implementing reliable messages on top of unreliable messages, a sender can wait for an acknowledgment message, and retransmit if it doesn’t receive it within a bounded period of time. But how does the receiver know that its acknowledgment has been received? Why doesn’t the sender have to acknowledge the acknowledgment (and the receiver acknowledge the acknowledgment of the acknowledgment . . . )? (For more information on the design of fast, reliable protocols, you might want to consult a text on computer networks.)