I argue that a theory's assumptions always are and ought to be unrealistic.
Further we should attempt to make them more unrealistic in order to increa
se a theory's fruitfulness. Many sociologists believe that a theory's assum
ptions ought to be empirically realistic. I contend that this criticism pro
bably stems from the confusion of a theory's assumptions with its scope con
ditions. While Friedman's (1953) similar prescription is associated with th
e instrumentalist philosophy of science, I maintain that it is also consist
ent with the realist view if "unrealistic" is taken to mean "incomplete" ra
ther than "untrue." I discuss a recent theory of the value of children by F
riedman, Hechter and Kanazawa (1994) to point out how assumptions differ fr
om scope renditions and how empirically plausible and realistic hypotheses
can be logically deduced from highly unrealistic assumptions. I then discus
s Kollock's (1993a, 1993b) revision of Axelrod's (1984) Cooperation Theory
as an example of when assumptions need to be revised.