No, it isn't impossible at all. However, it would be very subjective- like that Freenet overlay software that eliminates SPAM and child pornography with ease, such a system would be more based on subjective webs of trust and optional lists people could subscribe to, used to determine what you see.

The sender anonymity part could just be having every node act as part of a mixnet- wait until you receive X messages and/or Y time has elapsed, shuffle messages, send them out to randomly chosen nodes, repeat this for 3 hops and then your message is propagated.

I've wanted to make something like this for ages since 4chan has gone so downhill. Perhaps it should happen then. It would probably be easier to adopt than something as complex as Freenet FMS for sure.

technically possible, yes. but what do you think people would use it for and what problem would it solve?

how could it be abused by trolls? both local and nation state funded?

I would most of all hope it would solve the problem of people asking such questions, by making them perfectly moot. Folks would be forced to settle for wistful speculation about what the world might have been had the thing not been created.

so you want to take actions that force a state of regret upon the world for you having taken them?

the human condition sucks sometimes, but wow, that's pretty dark.

you could also consider, you know, actions that make your life or the lives of others better. it's likely a more rewarding endeavor.

Trolls would need to avoid being too overtly inflammatory, lest they get blocked by kneejerk reaction.

> What problem would it solve

Well, we could have a bunch of discussion boards again, ones that aren't at the permanent whims of profiteers or atrocious moderation. And strong anonymity even against global passive adversaries. Clearly there is a case for it, I know a lot of people who would find lots of joy in such a thing!

Freenet was also slightly slower than molasses in Antarctica. You had to really want to post there. The product of a messageboard is roughly:

sum[user_i in users](opinion(user_i)•engagement(user_i))

It's the second factor that causes all of the problems. But much like fast-food companies get a large fraction of their revenue from a small fraction of their customers, social media companies have trouble disempowering highly engaged users when functioning as a profit-oriented business.

It's interesting to consider what would happen with a "less moderated" messageboard where an individual's daily engagement is capped (supposing you solve the multiple account problem). It doesn't seem likely that we'll see that any time soon, though.