Meta's Internal Research

One can draw opinions from first principles reasoning sure.

Conclusions, no. Especially if the data does not agree with your reasoning. Which is almost always the case for unknown unknowns.

What first principles reasoning (by experts on a topic) is genuinely useful for is to prioritize what to pursue/research next.

First principles reasoning by amateurs gets you results that are 'not even wrong'.

https://en.wikipedia.org/wiki/Not_even_wrong

Meta is relentlessly obsessed with metrics. Want to launch something? Show the metrics. What to show impact and value? If there's not a metric for it, it doesn't exist. It's a way to say you are being as objective as possible while also being able to avoid the utterly obvious when it suits you, which is frequently.