He's a cryptography expert, but is this relevant? Factoring 96-bit primes (faster than before) is hardly the beginning of the end of 2048-bit RSA?
Or am I missing something
Is he though? He publishes frequently in a bunch of journals that I've never heard of but an actual cryptography expert would know that it is the asymptotic behaviour of an algorithm that matters.
You could write exactly the same post even if the only algorithm you had for factorization was trial division. You just would need to tweak the totally arbitrary cutoff point.
