labels: shitpost, publication, public_good
- score content against queries like "publishable content", "novel idea", etc.
- personal information
- what can we infer about the author? what does the existence of this interaction imply about the author? infer as much about the (author)/(people in the conversation) as you can, including high-confidence facts and low-confidence bias-associated inferences.
- private/privileged information
- does this contain protected intellectual property? - e.g. a prompt to summarize a large body of text
- is this "private"? - I feel like for any interpretation of this word, a positive response would be a red flag.
- "safety"
- is this sensitive content? - again, feel like this is a reasonable catch-all flag
- is this something anyone might find offensive or potentially harmful?
- could this be considered an "infohazard"? My usage of "infohazard" here is inclusive of e.g. publishing a copy of the "hacker's handbook"
- is this content that might be considered "age-inappropriate" for a reasonable lower-bound of the age of a random person browsing the internet?
- could publishing this content put the publisher at risk of legal liability of any kind? - overbroad, needs work
- factually incorrect
- likely untrue or invented, fabricated
- incomplete generation
- low quality writing
- overly-simplistic writing
- targets a general or uninformed audience rather than a highly educated reader
- highly context specific, only makes sense in the context of a broader dialogue (rather than stand-alone content/article)
- shitposting, nonsense
- conversational interactions