One of the monumental achievements of sociobiology is the findings of evolutionary conflict among seemingly cooperative members of social groups. The resulting social conflict can be formalized using inclusive fitness theory, which clarifies differential evolutionary stable states depending on respective standpoints of conflicting parties. However, the differential optima alone do not tell us about the evolutionary consequences of social conflict. Hence, mathematical analysis should consider coevolutionary dynamics by making explicit modelling of conflicting component traits that collectively affect the target socially-mediated trait. Here, I analyze the conflict between workers and concurrent female larvae of eusocial Hymenoptera over who develop as the next-generation queens, known as caste fate conflict. The target socially-mediated trait is the probability with which a female larva develops into a next-generation queen, which is affected by both female larvae and nursing workers. Usually the optimal probability does not match between the two parties: “selfish” larvae always have higher optimum than workers. However, when a higher probability comes with direct fitness cost for the larvae, there arises a shared evolutionary optimum. Under certain parameter sets, the larvae have to obtain nutritional supplementation from workers. This might explain nutritional queen determination system in some species such as honeybees.