In linguistic pragmatics and social anthropology, several influential researchers believe that politeness is essential for maintaining social order by way of disarming potential aggressiveness [Goffman 1967; Brown & Levinson 1987; Gumpers 1987]. In one of the most detailed of these theories, Brown and Levinson's, speakers pursue a single goal (e.g. getting the hearer to stop doing something) by using a mental model of the hearer to select a position on a one-dimensional spectrum of strategies that identifies the best balance between achieving the speaker's practical goal while avoiding offense to the hearer (as might occur from a purely brusque request). But are speakers actually limited to this one-dimensional spectrum of strategies? And given that people often pursue more than one goal at once, how might they do so in such a simplistic model of polite communication? I describe and evaluate a computational model that generates strategies for multiple simultaneous goals.