There is a huge difference between accidentally shipping a bug, and accidentally shipping a fully functional distributed end-to-end silent audio streaming system that works at Facebook's scale. They are in no way comparable. All the intricate crosscutting interoperating client, server, and middeware components of an end-to-end Facebook scale silent audio streaming pipeline do not just accidentally emerge without anybody noticing and magically dovetail together perfectly because nobody was paying enough attention to detail.

Large distributed systems don’t spontaneously assemble themselves without anyone understanding what they’re doing or intending for it to happen.

>I will repeat: "with Facebook, who knows."

Apparently we all know the obvious except for you. And you're not living up to your user name, which perhaps should be "rationalizer" not "rationalist". Instead of begin rational, you're bending over backwards to implausibly carry the water for Facebook beyond all credulity. It's a bad look, and shows you're vastly underestimating the complexity of developing distributed streaming software at that scale. So I will repeat what rsynnott said:

>And if you believe that, you'll believe anything.