Here are a few hard questions we wrestle with on integrity that I am curious to get the larger design community’s take on (eg. what principles or frameworks would you use, are there reliable patterns, opportunities for win-wins?):
1/ Given imperfect detection, what’s the right balance to strike between: a) false negatives (under-enforcement on bad content and accounts) vs. b) false positives (over-enforcement on good stuff)?
2/ How should we force-rank tradeoffs between a) free expression, b) safety and dignity, c) procedural fairness, d) privacy, e) community empowerment / devolution of centralized tech co. power?
3/ How should we prioritize problems across: a) their prevalence (how widespread they are), b) perceived intensity, c) potential to encourage off-platform harm, d) disparate negative impact on more at-risk populations?
4/ How should we balance fast responsiveness to emerging and exigent threats vs. proactively investing to get out ahead of new risks and building more scalable solutions?
5/ Take one problem space, bullying and harassment… what should we prioritize between: 1:1, many:1, 1:many; coordinated, uncoordinated; spiking/surging vs. serial abuse; observable (you can clearly see it) vs. contextual (you need to know the situation)?
I realize there aren’t obvious or general answers to these questions, and more context and data is necessary, etc. But very open to thoughts, questions, ideas!
