I always shake my head when performance is touted for new functions/constructions. It’s more important to minimize the costs of legitimate use while maximizing

I always shake my head when performance is touted for new functions/constructions. It’s more important to minimize the costs of legitimate use while maximizing the costs of software and hardware attacks. Minimizing the former inevitably leads to minimization of the latter in brute force (cloud cluster, ASIC, FPGA, etc.) attacks. The fundamental impetuses of scrypt (a PBKDF rather than an unauthenticated hash function) was a righteous and necessary advance that must never be forgotten with each fashionable iteration of newness that seems to lose sight of how faster or newer isn’t always automatically better.