(Before I go into sharing my thoughts, I wanted to +1 that Ao3's moderation is very much opinionated. The opinion is "don't like, don't read", but good luck trying to harrass someone in there.)
Here are my thoughts as someone who has worked in tech for a while and was part of many teams discussing moderation and safety:
(This will be a simplified explanation, the subject is more complex than that.)
Moderation is the closest word in the tech world for "police". Much like the police, both automated and human moderators reinforce an existing set of policies that they didn't define, and much like the police, their job is not to protect people, but to protect private property (in this case, to protect companies from liability).
That does not mean companies don't care! Some do.
Those policies are written by specific people with knowledge on what needs protection, and that's usually the bare minimum of what moderation needs to help contain.
Usually those companies are based in the United States, and their policies reflect US laws. As they expand to other countries, they adjust their policies in those countries as needed, so they're allowed to operate.
Defining policies becomes a much bigger problem when companies grow and become global spaces. That means companies are not only dealing with their own biases and internal ideological conflicts, but they're also dealing with different governments' policies and mandates.
A great example of that is censorship of content in China, or when Twitter employees in India got death threats from local government for refusing to change policy to favour a dictatorial government.
This idea of unlimited growth that became normal in tech tends to act as an imperialist force by reinforcing American policies, which as we know is geared towards corruption, individual liberty, and neoliberal capitalism. Their growth doesn't make them experts in dozens of different cultures and political beliefs, and usually they are not really trying to learn beyond what they are legally obligated to do.
But again, some companies do care more than others.
So what happens when you put literally the entire world in one place? Well, you're going to have all kinds of people: oppressors and oppressed. But how do you decide who deserves to be in a space like that? And who gets to define what's allowed?
The negative feedback loop problem
In a society that is broken, with social and economic inequality, with a select few at the top becoming richer as we all starve, do you expect the police to stop crime? Can they contain unlawful individuals? Can they protect us?
It's the exact same for moderation in tech spaces.
Donella Meadows wrote in "Leverage Points: Places to Intervene in a System":
“Reducing the gain around a positive loop — slowing the growth — is usually a more powerful leverage point in systems than strengthening negative loops, and much preferable to letting the positive loop run.”
If you imagine the growth in users and interactions as positive loops in the systems that make up tech products, moderation (and any other safety measures, like government regulation) is a negative loop trying to contain that positive loop.
Whilst investing in negative loops is paramount (because an unchecked positive loop leads to chaos), it can't be effective in changing the system. Because the speed of growth (relative to moderation) is exponentially faster. Regulation is a more sustainable negative loop, but how can you regulate something that is global in an effective manner?
Donella talks about self-organizing to change the goals of a system, then changing paradigms (and transcending them) as the most effective leverage points you have to truly change a system.
So maybe it's about changing who owns tech spaces. What would happen if we all owned it instead of corporations? And what would the goals of those tech spaces be? Would we be better at protecting people?
It's even more important to experiment and imagine different systems and structures (for instance: federated, not-for-profit social spaces), and be pragmatic about making them a reality, than it is to try to contain the positive loops of the current system.
If we want to protect trans people (or anyone else vulnerable in a system), we need to shatter the current paradigms into a million little pieces. Moderation will never stop hate and harrassment (maybe the most chaotic positive loop of the internet age), and in fact it will only make it worse (much like the police!).
It will turn us into beings that believe we can solve human problems with digital vigilance, based on policies that we didn't vote for, and that can be turned against us in the blink of an eye.
Anyway, read Mark Fisher!