Indeed, and to be clear I was specifically referring to Frederick's idea of introducing "deduction failure guides" rather than the std::elite proposal as a whole. I'd like to see the proposal developed further, but personally am not entirely sold on introducing a core language rule tied to this specific type.