The goal for distributed moderation is not to prevent radicalization, but to prevent one rich man from controlling the content people see. As a user, I can opt into whether I want to see terfs or developers or movie spoilers online or not. I think it's important for users to have more tools and agency over their online experiences.
There are other factors in play though that I think help create less polarising and radical social networks - ability to create custom homepage feeds gives to more variety and less control by the platform maker to sway audiences. Non-commercial networks (which doesn't count Bluesky) have less incentives to juice engagement. More distributed networks have levers audiences can pull if the main developers steer the network in the wrong direction.
> since people on the extreme political spectrum will always try harder than the regulars
Imagine if anyone could create an alternate feed algorithm that penalises these try-hards which people could subscribe to and use instead of the default.