I think Gates's point has merit with WhatsApp and like, India, where the group chat mob is very very real and very dangerous [1], and since it's all E2EE, there's no easy way to monitor the proliferation of this sort of misinformation. (Though my understanding was that they put some filters in to the front-ends of the app to reduce some of the more aggressive spreading, and further reduces the number of forwards you could make.) And either way, banning E2EE isn't the answer and is a horrifying proposition for all of the slippery slope reasons we're all familiar with.
You're also absolutely right that most of the spread is just happening in the clear on normal FB or Twitter posts, and not as encrypted WhatsApp messages.
I think that FB has a moral imperative to significantly reduce the easy of sharing, frankly. It's a single click to commit a share/forward that passes along misinformation instantly with no effort on the part of the sharer. It should require _some_ effort: having to create your own post, for instance. It would dramatically slow the viral-spread-like-wildfire aspect of the most egregious pieces of viral misinformation: where people simply see a salacious headline that confirms their biases or aligns with their views and mindlessly clicks share.
That is addressable by reducing the patterns meant to increase viral spread, which admittedly is a huge part of what lead to Facebook's success, but it's now causing far more harm than good.