> I really long for a future where content is published in a freely available database. Free to read from and free to write to. Where a post is a content addressable piece of data signed by the author. And comment too. And where a like is a cryptographically signed message, "I like this /joe". And where the order of your "feed" is not adjusted by outside forces.
Spammers and griefers historically conquer that kind of environment. Usenet fulfilled virtually all of your freeness criteria and now it's a cesspool of spam and worse. Everyone worth talking to has left. Moderation is necessary for quality discussion, or selective membership, or more likely both.
You can make a Merkle tree of signed comments but the problem was never the integrity of the messages (I don't recall anyone messing with the contents of existing Usenet messages) or relationships between them (the worst practical problem was bottom-quoting).
The order of a "feed" is adjusted by the rate at which spam is generated; spam will always be the first result in the feed because that is spammers' objective and they have automated resources to achieve it.
Usenet does not have the data needed to filter spam. A system of cryptographically signed follows and likes would.
My client could calculate a "karma" value for each post. How many of my friends follow the poster? How many of my friends friends? How many of my friends friends friends? How many of my friends liked the post? How many of my friends friends. And so on.
This sounds very similar to PGP if everyone had chosen to use it comprehensively for every conversation. Adoption, proper key management, and ease of use were the killers there.
Social karma style scores are trumped by giant clusters of spambots building up a huge collective karma and then infiltrating existing networks at the edges by phishing or buying existing accounts/keys. Also, some people turn nutjob and then you need a way to go back and revoke all your permanently-recorded cryptographically signed likes and follows because anything less would allow censorship. Not to mention the cost of re-running pagerank over the entire internet population on every client for every new (un)like/follow. People who abandon their keys leave permanent positive karma sitting around for exploitation. If karma is only positive then pure abusers have no negative signal and can coast on these abandoned karma edges. If you allow negative signals (publicly signed "block"s, "this is spam" judgements, etc) then spammers have a new weapon to kill off every other high-karma key with negative edges to reduce its spam-fighting effectiveness. Don't forget that this kind of battle will also be waged by otherwise highly-rational political actors whose survival depends on curating high karma keys and cliques to bolster then.
What works is human moderation in highly functional online communities. Read the excellent posts by dang here about his ethos and practices. Human communities are too complex for algorithms or models to manage, at least so far.
Spammers and griefers historically conquer that kind of environment. Usenet fulfilled virtually all of your freeness criteria and now it's a cesspool of spam and worse. Everyone worth talking to has left. Moderation is necessary for quality discussion, or selective membership, or more likely both.
You can make a Merkle tree of signed comments but the problem was never the integrity of the messages (I don't recall anyone messing with the contents of existing Usenet messages) or relationships between them (the worst practical problem was bottom-quoting).
The order of a "feed" is adjusted by the rate at which spam is generated; spam will always be the first result in the feed because that is spammers' objective and they have automated resources to achieve it.