In this paper, I clarify a process that produces power structure, under whi
ch some people choose a cooperative act by altruistic motive and other peop
le choose an uncooperative act by selfish motive. By using evolutionary gam
e models, I prove the following propositions: 1) When it is assumed that pl
ayers can choose their payoff functions under the condition that the probab
ility of continuation of the game is low, altruistic players who choose a c
ooperative act appear at a certain proportion, and the state is stable; 2)
In that case, selfish players exploit altruistic players. Therefore, it is
concluded that interactions with no constraints are able to produce power s
tructure, and that selfish players exploit altruistic players in such power
structure.