It is slightly better than you make it sound, because without many years of number theory we would have basically no confidence about the algorithmic difficulty of number-theoretic problems.
I’m not sure about this. RSA is published in 1978 and Diffie-Hellman was published in 1976. There was some amount of work on trying to efficiently factor integers before that and most of that work was focused on factoring numbers of special forms, but not nearly as much as their was in the next decade. And before DH, there was very little work on the discrete log question. In the case of factoring the difference in the work level can be seen in the drastic improvements in factoring in the few years after (especially the number field sieve and elliptic curve sieve.) Similarly, determining if a number was prime dropped from being almost as difficult as factoring in the mid 1970s to being provably in P 30 years later, and I don’t think almost anyone in the late 1970s saw that coming (although Miller-Rabin did sort of point in that direction).
I’m not sure about this. RSA is published in 1978 and Diffie-Hellman was published in 1976. There was some amount of work on trying to efficiently factor integers before that and most of that work was focused on factoring numbers of special forms, but not nearly as much as their was in the next decade. And before DH, there was very little work on the discrete log question. In the case of factoring the difference in the work level can be seen in the drastic improvements in factoring in the few years after (especially the number field sieve and elliptic curve sieve.) Similarly, determining if a number was prime dropped from being almost as difficult as factoring in the mid 1970s to being provably in P 30 years later, and I don’t think almost anyone in the late 1970s saw that coming (although Miller-Rabin did sort of point in that direction).