The meta-inductive approach to induction justifies induction by proving its optimality. The argument for the optimality of induction proceeds in two steps. The first “a priori” step intends to show that meta-induction is optimal and the second “a posteriori” step intends to show that meta-induction selects object-induction in our world. I critically evaluate the second-step and raise two problems: the identification problem and the indetermination problem. In light of these problems, I assess the prospects of any meta-inductive approach to induction.