If the end result is politically unbiased, it's due to their conscious decisions, not some magic algorithm.
As opposed to centralized moderation (twitter, FB, IG, etc.)
See here for more: https://news.ycombinator.com/item?id=39471973
Calling it 'The Problem' like climate change or the national debt gives it too much power. Just use something else. People use group chats for real relationships now anyway.
So in practice you have to make choices, or you'll end up running the new 4chan and being sad about your life. As happened to the guy who ran the old 4chan: https://www.rollingstone.com/culture/culture-features/4chans...
The ideal, platonic version of this would be that moderators only see an "identity scrambled" version of each tweet/post when they make their moderation decision. Like a screen that blinds orchestra musicians when they audition, the human would see a statement like "I hate New Yorkers" and not know if the original message said "I hate New Yorkers" or "I hate Floridians." So they would have to make a decision based on the general principle of whether a statement of this form is allowable.
Anywhere you want to draw the line is fine with me, as long as you draw it consistently.
Take the use of reclaimed slurs, for example. When used against the discriminated group by a dominant group, their intention is often to cause harm. When used within the group, the intention is to reappropriate the term: https://en.wikipedia.org/wiki/Reappropriation
Similarly, harassers will use terms in ways that are plausibly read different ways depending on who they're talking to. So something that might sound innocuous or just odd when directed at me will be correctly read as a racist attack when directed at somebody else.
And that's not even counting when they'll just come up with new terms so they can be awful in ways that are novel enough that automated filters or out-of-date moderators won't catch. E.g.: https://www.vice.com/en/article/bv88a5/white-supremacists-ha...
In short, because there's a great deal of identity-based hate in the world, identity-blind moderation ends up being an aid to the identity haters out there.
That this moderation strategy would prevent the use of all slurs (even reappropriated ones) sounds like a feature to me, not a bug.
You're proposing erring on the side of censorship to avoid some gray areas. While this is a reasonable position, it doesn't satisfy some ideal of neutrality and won't really avoid the gray areas, and so still would require subjective judgement.
In practice, almost any nominally "neutral" position ends up allowing an enormous amount of abuse. Which is why you'll see most platforms that start with a free-speech maximalism approach coming up with a lot of nuance and exceptions over time. And those that don't turn into cesspools.
Most people are pretty great, but moderation has to be built for the worst-case attacker.
Which is ultimately what I object to.
Moderation is full of gray areas, and they are unavoidable.
What we have now is a system where, on many platforms, moderators often put their thumbs on the scale and decide that certain groups need more protection than others. Generalizing about or disparaging certain groups is ok, but the sensitivities of other groups are considered sacrosanct and must be deferred to.
Like I said, draw the line anywhere you like. If it applies to everyone equally, I am happy. I am fine with things that require subjective judgment, as long as that subjective judgment is behind a screen that conceals identity.