good point, but that was left out of the earlier statement about inference. I suppose I should have inferred it however.

Yea, my bad I guess. I tend to think people mostly get that biometrics are, well, mostly immutable and that not being able to switch them up in response to a suspected breach is a huge inherent weakness. So the only defense I really get of them from anyone is that the effort for the user is minimized while the effort for the attacker is still fairly high. The problem with that is why I mention inferrability: The existence of a computer system that can authenticate via a biometric implies the existence of one that can spoof it, and we don't have any reason to believe this involves, say, more of a cost disparity than cracking a password, let alone anything approaching a strong one-way function.