The paradox is that when we model future designs on past successes, we are inviting failure down the line; when we take into account past failures and anticipate potential new ways in which failure can occur, we are more likely to produce successful designs.