Fun story: after the last report which called them out for not talking about write loss by default, MongoDB updated their Jepsen page to say that the analysis didn't observe lost writes. I guess they assumed that people wouldn't... read the abstract? Let alone the report?

For me, MongoDB has track record of bolstering a lot ("webscale") and hiding/denying mistakes.

PostgreSQL is quite the opposite on that front, confident yet open to critics and abble to admit mistakes. Hell, I've even them present their mistakes at conferences and ask for help.