Has anyone actually catalogued troll reflexive replies? E.g. [Stmt] -> "It's a Republic not a Democracy."
-
Show this thread
-
This is sort of related to what I’m doing with the anti-abuse stuff, but I assume someone’s already done it. Like seriously, search for the phrase and it’s variants — it’s, um, an extremely accurate classifier for “bot or person I don’t ever want to grant my attention to.”
2 replies 0 retweets 9 likesShow this thread -
Replying to @generativist
More generally, maybe all common responses could be in just a few buckets: Abuse, Knee-jerk memes, Uninformative, or Legitimate short reply.
1 reply 0 retweets 1 like -
Replying to @robinc
Yea. I've been having a lot of trouble with this UI-wise. Like, I had a component that showed multiple options for users to say what *kind* of [unwanted] response it was, but it felt like too much to ask.
2 replies 0 retweets 0 likes -
Replying to @generativist
Perhaps a two-level control? Level 1: unwanted Level 2: what kind of unwanted Akin to Facebook likes having optional variants. Perhaps only a few level 2s would be enough to calibrate the "kind" model useful for downstream tasks. Though the level 1 model is more important.
1 reply 0 retweets 1 like -
Replying to @robinc
Yea. That's *sorta* where I ended up. 1. [Abuse] [Unwanted] [Acceptable] 2. if User.Settings.Expanded { secondTierClassification() }
2 replies 0 retweets 1 like
I'd prefer expanded for all, but I think it would lower the first tier volume or make people get sloppier. Gonna need some actual numbers to back any of this up though, so first: RELEASE SOMETHING!
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.