> The disagreement centers on a chain letter that spread on Facebook Messenger in Myanmar in September. The messages warned Buddhist communities of an imminent Muslim attack. Meanwhile, Muslim populations received a separate message cautioning them of violence from militant Buddhist groups.
If you take that description to a smart, compassionate person who doesn't happen to know the 2015-2020 politics of Myanmar, they'll have no idea what the right moderation strategy is. (They're actually pretty likely to come up with a worse answer than Facebook, because the Burmese government is led by Aung San Suu Kyi, globally known from 1990 to 2015 as a brave fighter for freedom and human rights.) So this isn't just a matter of moderation tools; in order to moderate appropriately, Facebook has to be aware of all current human rights controversies in all countries. That's a high bar!
And, to address your new point about moderation being hard when you have billions of users: The thing about having billions of users, is that it gets you billions in revenue. In fact, revenue probably scales super-linearly with user count. Either way, the required-moderators-per-user actually decreases as you get more users because you can spread the fixed development costs of moderation tools (including ML models, obviously) over a larger number of revenue-generating users.
From your comment, your key point is apparently:
> just to mention the inconsistency with the blame solely being on Facebook itself
I don't think anyone is making the claim that it's "solely" a problem with Facebook. People are just pointing out and discussing mistakes that Facebook has made.
All of that said, Facebook is certainly doing better than they were a few years ago. For example, they now have people who can actually speak Burmese on their moderation team.[1] So that's a good start...
[1] https://www.nytimes.com/2018/04/09/business/facebook-myanmar...