Anyone can run a moderation service. Users subscribe to the services they want. When you flag a post or account for moderation, you choose which service you send the report to.

> How does blocking work?

Blocking works by publishing a "hey I've blocked this person" record in your database. Applications then can use this to enforce that.

> How do people make sure to distance themselves from political enemies?

I'm not sure what this means.

> Do the aggregator cache servers block certain user domains?

They could, and this is also a property that's not really about blocking, that is, you could choose to cache only part of the network if you choose, which could be useful for various reasons.

> How do you ensure that the aggregator returns valid and non-forged comments and likes?

Cryptography, basically.