Oddbean

sad, that client devs seem to think mute words will stop a stalker or bots.. how naive they must be i guess.. everyone is on a different journey, sometimes it's strange to think, "oh, yeah, for the first time they are discovering what its like to be the target of intensive spam? weird." anyway.. we got shit for this, and it aint "nos.social blocked the bot" -or- "mute the word GM" my god y'all.. dont mute GM

But they did stop it. At least in the user's own space. Mute words just make sense to a whole lot of people. Sure, there are way better options, but the simplest that gets the job done to that specific user always wins.

it didnt though. not only did it not stop replyguy, but it'll never work against the same thing or a slightly evolved thing targeting individual pubkeys or groups.. what happened was a combination of big relays finally filtering, the mute lists (yes) and these webs of trust/ paid relays.. clients need fine grain relay controls.. gossip has a relay setting for marking "spam safe" where it will pull more notifications than from a relay that isnt marked. i was not specifically talking about any one client this is just something i notice over a very long time period. i used to attempt to block things like replyguy too, on the relay side, the content blocks DO NOT WORK against an evolving bot. the same thing happens over and over. sometimes to individuals, sometimes to the wider network. many will be leaving nostr in the future because of this, they have no control over replies other than just turning on "replies only for follows". and this is not gonna work cause follows cant grow past 1-2k pubkeys. mute words is handy, its useful. but its not a solution. i hope client devs realize that.

no, in theory wot relays, wot clients, lightning network.. thats what all my 'game theory' ideas keep coming back to as a solution. but i think im about a year ahead of everyone from running these relays and seeing all the spam all the time. you dont follow that many people so you may not see the endless misery of the users that it's not fixed for, gave up, etc.

i think maybe ive just been a huge target of past replyguys. so my main point was, client devs dont seem to be seeing what im seeing, thats the perfect way to attack nostr is attack the plebs, so client devs dont notice or dismiss as not a problem.. like brigades? i duno. i was forced into paid relays because bots ravaged my original relay till everyone left, despite all kinds of language parsing detection stuff. dont let the same thing happen to you by building in some smart selections upfront. works for some, is not gonna #grownostr. this shit has to work, its our strength and its what twitter/bluesky couldnt do..

I agree with you. I just don't think this particular instance is representative of an actual attack. Whatever we code for it can be circumvented by a smarter player. Not that we shouldn't code it, but let's be mindful of what we are solving for. Real bots with WoT (because people will gladly sell their nsec to a bot maker and restart their accounts) will be much more demanding than this simple attack. They are going to look, act and work as a real human. Every message in Nostr with immediately get 200 replies with products and opinions from people that are paying for it by nsecs of previous real users with WoT and so on. Any serious defense on this will need to include a complete understanding of the entire network or it will be something selected by the user, like just people that I follow.

did you read the thing about bluesky trust and safety team working overtime on their user influx? how theyre drowning in moderation? this is not the future of nostr i want to see. the only way i see this working is outbox, smaller relays, manageable relays, managed by small groups of people that are aligned in their values.. same for media links. amethyst is doing well, it takes time, but what other clients are doing is scary road to trust and safety megacorp. this is a large part of why i dont advertise much, i dont want to be a big relay nor do i want to use big relays because then im part of the problem and not the solution. having fine grain relay selection, easily doable in amethyst with a few more iterations. other clients, its not clear to me that they want to do any of this.

I did see it and I do agree its a massive problem and small relays are the solution. But small relays is just decentralized moderation. The moderation is still there, just run by thousands of relay runners instead of an organized central point. Which is better, but the infra to moderate in a decentralized way needs to be there. That's why you me relay ops interfaces are so crucial to the future of nostr.

A combination of paid relays, WoT and NIP-13 seems to be enough to block any kind of spam. All 3 increase the cost of a spam attack, each in it's own way. What we need is more clients implementing NIP-13.

ok great, so you limit the frequency, what happens? bots can go just as fast as a human meanwhile nothing changes other than humans are penalty. just look at the damus relay for example. 6 events a minute and no way to increase for legit use.