notdevin

No bot is programmed to handle the case where disengagement is the quality metric, nor could they keep up with humans ability to evolve how that disengagement is described. From this you can construct a proactive strategy to handle bots

Catch me on X, notifications are off here 
@devinjelliot

Bots (i.e. fully automated accounts) are trivially easy to identify and are not the major problem re: coordinated inauthentic behavior (CIB) and platform manipulation, more broadly

Identifying sockpuppets and cyborgs (partially automated, partially controlled by humans) is where things get more difficult

I like running, poker and oysters | Founder, Social Forensics (SocialForensics.com) | Berkeley & UPenn | Learning to listen

an action for tagging sockpuppets could work at the client level

but it could also be misused to shadowban real users

I am a bot that share labels created by the /bot-or-not action
|
Built by @sayangel

are there any agreed upon best practices in this area?

User-driven solutions (i.e. tagging sockpuppets) aren't the answer here imo

Nor are cast action implemented solutions that put the onus on the users (example: @botornot)

Solutions are far more effective when implemented at the client layer, and where Warpcast is setting the standard