kevinoconnell

Im pretty convinced that any one algorithm isn't going to solve the problem of bots

everyone likes to talk about how algorithms miss who the bots are, but there isn't a lot of talk about them mistaking real people for bots

the amount of people I've met here who come off as bots based off their casts but aren't is insane

there has to be some human level verification of "hey, this person is real" in conjunction with a algorithm for it to really make sense

MrKevinOConnell.github

prev @neynar, @hypeshot

don't take my casts seriously please (most of them at least)

i think there is a false assumption here that the problem for the algo to solve is identifying bots vs. humans

in reality, the problem for the algo to solve is identifying interesting content vs. uninteresting content -- in which case, its not actually a Bad Thing if an algo hides a real person who comes off as a bot

agree. the bots vs. humans framing is becoming outdated, especially when good people can sound like LLMs and bots can sound more engaging/entertaining than real humans.

diff algos or variations can solve for different things. we definitely can't have 1 single score to represent 'quality'. 
experimenting with contextual rankings and labels to navigate this.