This study applies Frank's evolutionary game model (1988), under the assumption that the payoff matrix is the type of Prisoner's Dilemma, to simulate the numbers and rewards of altruist and hypocrite. The results show: (1) In the long run the evolutionary game will reach a "break-even point". Altruists and hypocrites will both survive with a certain population ratio. (2) Reducing "the cost of identifying hypocrite" will increase the number of altruists in a society and will increase both altruists' and hypocrites' benefits. However, hypocrites' benefits will be higher than altruists'. It means the decrease of social justice. (3) Promoting altruism (increasing the number of altruists) will not only increase altruists' benefits, but also more hypocrites' benefits. Hence, altruism became a faith of no opponents. Everybody (altruist and hypocrite) "speaks" the good of altruism. (4) According to the results above, the present study leads to a new point of view to discuss the traditional concept of "Morality Reward Gain". "Morality Reward Gain" was supposed to be a faith of social justice (good is rewarded with good and evil meets evil). However, promoting the faith will decrease social justice (hypocrites' benefits will be higher than altruists'), but will increase the social benefit. This is called "the Paradox of Morality Reward Gain" in the study.