Writing a spam filter is kinda fucking hard.
WOT is cool, but it's a band-aid that excludes newer users.
POW is also cool, but most Spammers can get around that super fucking easy, so I don't really see that as a proper solution.
Micropayments are also not a solution in my mind.
What I'm doing so far kind of seems to be working, but I also block my own posts from time to time.
Lol blocking your own posts could be problematic.
So far I havent blocked my own posts after some tuning this morning.
I guess I probably should have followed this up with what are other relay runners doing.
#asknostr
I think a combination between these methods together: pow + wot + paid relays we can get rid >90% of spammers
Can wot lead a censorship of new users? Maybe, but they'll have greater incentive to behave with a minimal level of politeness and respect in the platform. Not a perfect solution, but it's a good tradeoff when we are in a decentralized platform
With the way current web of trust relays work, if you don't have any followers, within the web of trust, you're not allowed to post on that relay.
So for instance, for my web of trust relay, I have a depth of two, meaning the people that I directly follow and then the people that my followers follow can post to the relay. But even that's not perfect because I was just recently checking my own web of trust relay to see what's getting posted there. And there's already some shit on there that I feel the need to nuke.
If you're a new user who doesn't have any follows, but you're posting interesting stuff, if people are only using Web of Trust relays, you might not get seen.
Paid wot relays could solve this problem, don't? Spammers and trolls hardly want to pay to post their useless shit
I mean, you can do a paid relay without Web of Trust that solves the problem. My paid relay has had zero spam. It's the free relay that is plagued with the reply guy spam.
Yeah, exactly as I said in my first comment. Multiple solutions approach (wot + paid relays + pow) and you can get rid of spammers in >90% rate. Not perfect solutions, better ones eventually can be developed, but this is what we have for today
I was thinking in a wot relay that allows new people to publish, this new people have a time span where their post will be visible, if they get engaging or similar they stay, if not goes away. There are more details, but does this makes sense?
yeah that does make sense. Keeps posts of established users, but then gives new users time to build a reputation. If no one is interacting with spam, it'll essentially be dropped.
Exactly I think its a better approach than paid relays, not because paid relays are bad idea, they are good, but people need to pay and also have btc, ln, cashu, xmr or whatever you want to use. It's a lot less friction, as well as pow, and the user doesn't need a voucher to publish. Then this can be combined with a robust collection of policies for easy discard. I wish I had the time to build it. If you want we can keep in touch to brainstorm/develop
What if I created 10 accounts...? Each one gives feedback to the others... interact a little... They add different people... Etc.... At that point, it will be hard to guess if they are botspam or not, or at least prevent it.
A simple opinion
None of these 10 account are able to decide nothing, they are not inside the wot of the relay
That would be a legitimate way of gaming that.
I've been wanting to kind of figure out a way of looking at npub behavior in general as a spam filter. So for instance, if you created all 10 of those accounts at the same time and they're only replies to where to each other Within a short amount of time of their creation, that could potentially be spammy behavior and could be blocked.
spam filtering really just is a game of whack-a-mole. There's always some way to game a spam filter.
That means the spammers should pay for 10 accounts in paid relays. This is very unprofitable for them. Paid relays drop the incentive for spamming to almost zero
A combination between non-wot paid relays + wot free relays seems good to me
