Quite difficult given that humans can't interact with the internet "directly", but only mediated through software.

We need gated communities. The idea of a public internet has run its course, it's now an ocean of noise.

Set up a book club, meet in the park, or a coffee shop.

I've been trying to will a web of trust style system into existence for a while now, I lack both the marketing skills and programming know-how to actually create it though =)

Basically a way to see on every web page whether an actual human (or more) in your network has vouched for the content to be written by a person.

A crazy thought I had is that agents without a link to human identity might need to be treated as illegal. That human identity would be blamed the for the agent's actions.

This raises a rats nest of issues, but will we be able to avoid this necessity?

I think you've just thought of CAPTCHAs? Unfortunately, AI have increasingly become better than humans at solving the tasks we throw at them for such tests.

This is an interesting problem to solve.

I wonder if it is possible at all to have anonymity without admitting bots.

And how would you do that without dystopian verification checks?

The reasons why Youtube and Discord are so gung ho on age verification might be because these companies that sell ads and data have a monetary incentive for distinguishing humans from bots for their investors and shareholders.

If I were to chose I'd rather have a bot infested internet than a mass surveillance dystopia.