Open sjacks26 opened 7 years ago
These are additional resources/documents that were posted in the #far-right Slack channel surrounding this discussion: http://www.voxpol.eu/download/report/VOX-Pol_Ethics_Politics_PUBLISHED.pdf https://aoir.org/ethics/
I am waiting to hear back from some contacts I have in psychology research and will report back if/when I hear anything!
I finally had a chance to read the VOX-pol report. Most of that is geared towards law enforcement and other government action, and they don't specifically talk about identifying individuals at all.
One thing they mention (pp. 47-49, roughly) that might be useful for us is having different standards for, on the one hand, accounts that are propaganda/spokesperson accounts, versus, on the other hand, accounts that are individuals who might express some passing affinity with the far-right but haven't intended for their account to be primarily about that affinity with the far-right. That's super abstract, so here's an example: we might want to treat David Duke's social media presence differently than some person who has posted or commented once or twice on pepe images out of thousands of posts. We prob don't care about pointing out all of David Duke's social media accounts, but we might not want to connect that other person's Facebook account to Twitter or LinkedIn or whatever else might be more professional. It might be difficult to operationalize the difference between those two types of accounts, though.
Before we get too far along in things like community detection, we want to have some internal D4D guidelines for content attribution (doxxing, basically). We have a document with some preliminary language, but this is very much a work in progress, and we would appreciate everyone's thoughts. Feel free to edit/comment on the doc, or comment here.