AFAIK they would get the info "which user said what". The source would be what ever instance you are from tho, not your own ip. So the owner of a instance knows as much as everyone else.
The 1949 geneva conventions have been rattified by (afaik exclusively): all UN members, palestine, the holy see and the cook islands. The singning binds the signee to uphold these rules for themselves. How then could the UN, who as you said is "just an extension of its member states", not follow the morals you yourself held as the absolute standard?
"Elementary, my dear watson" is a sherlock holmes "quote", which is sometimes used to roughly say "it is actually quite simple".
Take away the "my dear" and the post roughly sounds like the "quote".
I think your estimations are reasonable, but some further information: it is not just a normal screen (with normal picel density) at a bigger size, but the pixels are rather far apart. So I would assume that this reduces the cost.
How would you derive that someone "dislikes jews because they are jews"? Do you listen to them talk and make a decision following your gut? Do you make an MRI image of their brain and measure their brain waves to read their mind?
You would have to work with their past actions. And then it becomes non trivial to define "anti-semite".
So it is not really about "what does the word mean" but "how do you decide who fits the definition and who doesn't". People are concerned about this because it is very hard to make a law that 100% only fits to the people you want to target (leaving asside wether the principle is correct or not).
But I don't think an X-RAY would even qualify as NSFL, would it? I mean you see that factually, that probably hurts, but you get zero impression of it.
Kinda like fotos of the atomic bomb explosions from WW2. You know some people are suffering in the very moment that foto was taken, but you can't see it, so it's not NSFL.
Or do you have a dofferent oppinion?
Hmm, afaik florisboard doesn't support word suggestions at all right now. You can read up on it a bit on github but to make it short: the main dev is struggling with creating his own performant word suggestion algorithm.
More specifically: "4 people died from covid" is also true