How do you deal with trolls going around filters? I'm sure most of us did that in the past, some of us were young teenagers in the time of auto moderated live chat.

Better filters. For instance crowd sourced filters that you can subscribe to and help update. Just like ad blocking technology does today to combat advertisers attempt to evade them.

That requires centralisation because either trolls will affect the filter or the list won't be updated.

Each filter comes from its own central location curated as seen fit by its owner, but there can be any number of filters distributed across the net. Each will win subscribers based on their reputation.

Again, this has all been tested in the wild. It works just fine for ad filtering today.

Not really. Adblock made deals with advertisers so people have to use Adblock Pro, and avoid Adblock awesome because it has adware. Same for uBlock.

edit: To make my point clear: it seems unfeasible to weed out bad actors without relying on good actors. Someone you place trust in (adblock list maintainers, p2p network filter maintainers) can make deals with who you try to avoid ('non-intrusive ads', mates). It's little better than letting the centralised host moderate, especially when the biggest moderators like Adblock are the ones we tell people to avoid. Otherwise it's just like PGP, perfectly designed to be never used.

That's irrelevant to the point, or actually makes it stronger. uBlock Origin continues on in spite of those compromised competitors and all of the subscriptions can continue to be utilized. I really don't understand what you're arguing against. The software can be replaced, a bad acting filter list can be ignored, and many good lists can be aggregated to counteract ad agencies attempts to bypass filters. And with the right software you can create your own local filters with a single click on the offending material.

We don't need a parental authority to protect us from seeing naughty images, we can take care of ourselves.

I was thinking more along the lines of it being technically impossible for the police to do anything.

The police can find the ip address of whoever's serving it, ask the isp for their name, and give a knock on their door.