Content pfp
Content
@
https://warpcast.com/~/channel/buoy
0 reply
0 recast
0 reaction
pugson (Farcaster Professional) pfp
pugson (Farcaster Professional)
@pugson
i really wish someone at buoy would stop this from happening automatically
9 replies
0 recast
10 reactions
jj πŸ›Ÿ pfp
jj πŸ›Ÿ
@jj
Working on updating our spammy fids and adding pro users after the snap chain updates
1 reply
0 recast
0 reaction
pugson (Farcaster Professional) pfp
pugson (Farcaster Professional)
@pugson
ty. i got over 70 of these today πŸ˜΅β€πŸ’«
1 reply
0 recast
2 reactions
jj πŸ›Ÿ pfp
jj πŸ›Ÿ
@jj
Ok just updated it with https://github.com/warpcast/labels/blob/main/spam.jsonl Let’s see how it does
2 replies
0 recast
1 reaction
Colin Charles pfp
Colin Charles
@bytebot
checked this updated label, and still see spammy users (just from the pugson screenshot), e.g. {"provider": 9152, "type": {"target": "user", "fid": 500449}, "label_type": "spam", "label_value": 2, "timestamp": 1746237255} or {"provider": 9152, "type": {"target": "user", "fid": 548688}, "label_type": "spam", "label_value": 2, "timestamp": 1746237258} turns out spam labels might need another update to double check if their X accounts are suspended!
1 reply
0 recast
1 reaction
Jordan pfp
Jordan
@ruminations
interesting, these folks are below our spam threshold
1 reply
0 recast
1 reaction
Colin Charles pfp
Colin Charles
@bytebot
spam label updated - still 2 for these two for example - and more importantly, very low neynar scores.
1 reply
0 recast
1 reaction
Jordan pfp
Jordan
@ruminations
ah they have low neynar scores? interesting
1 reply
0 recast
2 reactions
Jordan pfp
Jordan
@ruminations
i guess a good approach would take into account network influence/reputation and activity, i.e. if you had a low score & lots of casts, that's suspect. or low score and low engagement rates and lots of casts.
1 reply
0 recast
2 reactions
Colin Charles pfp
Colin Charles
@bytebot
i would also look at the casts. tokenise and if matches certain tokens - that can help mark spam. so, look for words like "Sweet" - token id: 83184, then look for more like stylish, lit, etc. or just look at the one word casts (in replies) to also take a closer look. sample set size: fid 500449 i am sure we will beat the spammers - whack a mole - but we will win
1 reply
0 recast
1 reaction