But this is another area where if P=NP you might expect hints to leak through. As Aaronson points out elsewhere, there's a number of problems where people had

But this is another area where if P=NP you might expect hints to leak through. As Aaronson points out elsewhere, there's a number of problems where people had found practically efficient algorithms, or polynomial-time randomized algorithms, or something, long before they found an actual polynomial-time worst case algorithm. (In addition to primality, linear programming is another example.) For NP-complete problems there's none of that, no such hints leaking through.