Filtering the visibility of spam on the app/clientside, or on relays by preventing the spam from being stored in the first place, using simple follow/social graphs is far better than an explicit reputation system that scores pubkeys on random parameters to gauge how “trustworthy” they may be.

IPFS and others have historically tried to score peers to scale P2P networks; it devolves into an overly complicated mess of *arbitrary* rules.

Reputation systems don’t scale just like how Mastodon style governance doesn’t scale: due to argument surface about what classifies trustworthiness/who’s allowed in the fishbowl.

Relay operators integrating the social/follower graphs of people they trust to prevent the storage of spam will probably scale better than we think, especially if a lot of people begin to run their own small relay.

Social graphs become more powerful as more users join. It’s a cumulative network effect, where the social side of the network effect accelerates the usefulness of the technological social graphs: https://youtu.be/a99ry70CnRs?si=2dRZMK1GjBwVB-cP

We’re going to work on integrating this minimalist approach into the H.O.R.N.E.T Storage Nostr Relay @npub1h0rnetjp2qka44ayzyjcdh90gs3gzrtq4f94033heng6w34s0pzq2yfv0g with @npub1g53mukxnjkcmr94fhryzkqutdz2ukq4ks0gvy5af25rgmwsl4ngq43drvk new social graph package: https://github.com/mmalmi/nostr-social-graph

Shortcuts are futile; slow is smooth and smooth is fast. 🐢 Solving spam or any other challenges on nostr by compromising on its original values is ultimately a fruitless pursuit.