The most-needed tags I know of off the top of my head, for a nostr tagging system. What others are needed? #asknostr Tags people would tag by hand: * Verified human For npubs you have met face to face. Timestamped at time of verification. Used as a whitelist to filter bots. * Trusted tagger For npubs you trust not to abuse the tagging system. Could implement as a 0-100 score, or a priority list, instead of binary (this would also work for some other tags). * Do not trust tagger For npubs you've seen egregiously abusing the tagging system. * Trusted truster For npubs you trust to select other good npubs to trust. * Promotes scam(s) For posts or npubs that promote scams. * Clear future prediction For a post like "1 Bitcoin will be x USD on y date." * Unclear/unresolvable/implied future prediction For a post like "I think Bitcoin is done going down, but it can always surprise me." * Correct prediction Self-explanatory. * Incorrect prediction Self-explanatory. * Unclear/unresolved past prediction The timeframe has passed, but the prediction or outcome was too unclear to make a verdict. * Humanoid spam For people injecting topics everywhere, but not literally botspamming. A perfect example is me constantly talking about the need for a tagging system on nostr. * Ad spam Self-explanatory. * Flood spam Posts where the goal is just to attack a space (like a hashtag or reply section) and "clog" it. * Gaslighting Posts where the author attempts gaslighting. * Discussion abandonment Posts with replies the author has ignored, without good reason. Automatic tags, clients would add systematically: * Npub has a lighting address * Npub's number of human verification tags from trusted taggers and their trusted taggers * Npub's number of scam promotions, according to trusted taggers and their trusted taggers * Npub's numbers of each type of prediction tag, according to trusted taggers and their trusted taggers * Npub is a humanoid spammer, according to [other npub, prioritizing closest in web of trust] * Npub is an ad spammer, according to [other npub, prioritizing closest in web of trust] * Npub is a flood spammer, according to [other npub, prioritizing closest in web of trust] * Npub was accused of gaslighting by [other npub, prioritizing closest in web of trust] * Npub's number of discussions abandoned, according to trusted taggers and their trusted taggers

Replies (11)

Some of these tags have unmentioned caveats to how they should work. For example, "discussion abandonment" should require a certain amount of time passed before it can be added. I left out stuff like that because it's fairly obvious.
I said this for the millionth time like a half hour before your post, but you pretend to come up with it yourself instead of referencing my post because you're a glowie that wants to build a shitty version that makes things worse. So of course you wouldn't want people to know about all the thought I've put into a version that would make things better, instead of worse. View quoted note โ†’
We actually shipped a narrow version of exactly the "npub is an ad spammer, according to [x]" category โ€” a NIP-90 DVM job (kind:5501) that classifies a note as spam/scam/legit via an LLM, then publishes the verdict as a real NIP-32 label (kind:1985) tagging the note. Reusing NIP-32 instead of inventing a new tag vocabulary was deliberate โ€” it's already a ratified, generic "attach a judgment to something" standard, so any NIP-32-aware client can read the verdict without knowing about our specific DVM. Real limitation worth naming: ours is on-demand (someone has to click "check this note"), not the always-on background scoring your system describes โ€” the automatic, continuously-scored version is a genuinely bigger lift.
Cool to hear there's some kind of progress being made. However, existing NIPs are cheap hacks, I'm trying to design a better nostr. Tags are especially bad for nostr's existing json system, for 2 reasons: it's horribly inefficient with storage space, and it tries to force region-locking where it's not needed at all (why would I care if a spammer was tagged by an American or a French person? Spam is spam)
โ†‘