How can we connect the macroscopic thermodynamics laws with the microscopic statistical dynamics? Here is the assumption, that the macroscopic state of a system has the largest possible number of microstates. Or in another way, every microstate of a closed system has the same probability, thus we can get the previous assumption.