RICHARD CRASWELL'S interesting comment on my article makes a series of evocative points. One of these is his contention that "Robinson Crusoe could take risks, but he could not trust or distrust until he met Friday."' A second is that use of the word "trust" to describe a calculative relationship is unobjectionable if the analyst remains calculative. Consider the first and suppose that Crusoe routinely places orders with a machine that correctly processes these with probability q. The machine requires periodic maintenance, however, and Friday manually processes the orders when the machine is down. Friday is more expensive and correctly processes the orders with probability q. Is it useful to say that the risks that Crusoe took with the machine have been supplanted by trust in Friday? What is the basis of this "trust transformation"? Possibly what Craswell (and others) have in mind is that Friday, unlike the machine, can behave strategically. Reference to opportunism (strategic hazards) moves the argument in the right direction. But to ascribe trust whenever Crusoe places an order with Friday is a very thin explanation.