That would be subject to criticisms of bias though. It would be interesting if the citations were found credible by different clusters of users + sources.
-
-
Replying to @kimmaicutler @vr00n and
There's no way to avoid accusations of bias in this, no matter what you do :/.
1 reply 0 retweets 3 likes -
Replying to @dswillner @vr00n and
Of course. But you wouldn’t want a small cluster of mostly Americans making a whitelist of credible sources.
1 reply 0 retweets 2 likes -
Replying to @kimmaicutler @dswillner and
Bit hard for me to follow the thread but yes, what you’re describing is possible for linked and memetic image content. From our experience you need a starting ontology of reliability but it should have emergence built in to assess edge cases.
3 replies 0 retweets 3 likes -
Replying to @krmaher @kimmaicutler and
That’s going to make some folks uncomfortable but you want a learning system. Building it at scale would be difficult for a paid network but if you want it to work through crowd participation you have to have some reversible and transparent permissions hierarchies.
1 reply 0 retweets 1 like -
Replying to @krmaher @kimmaicutler and
Otherwise it’s easily open to brute force gaming. There’s a lot about designing anything like what you’re saying that would need to be quite a bit of trial and error, no silver bullets.
1 reply 0 retweets 0 likes -
Replying to @krmaher @kimmaicutler and
But I like it conceptually. The tech, surface area, and scale for training all exist. It baffles me that we don’t see more visible efforts in this regard, especially from a company that’s famously iterative. It’s like that “what if we build a better world...?” cartoon.
1 reply 0 retweets 2 likes -
Replying to @krmaher @kimmaicutler and
Or maybe they’ve experimented with something like it as part of their newsfeed signals efforts, and it didn’t work and was deprecated.
Not a lot of transparency, so hard to say.1 reply 0 retweets 2 likes -
Replying to @krmaher @kimmaicutler and
Well they hired Accenture who hired eyes from the Global South & Eastern Europe to moderate global content. There's a hyperlocal element here that's consistently missed in these solutions and could be meaningful is explored further. https://www.npr.org/sections/alltechconsidered/2016/11/17/495827410/from-hate-speech-to-fake-news-the-content-crisis-facing-mark-zuckerberg …pic.twitter.com/fxYZXwlkZE
1 reply 0 retweets 0 likes -
I know, I helped hire them! The difficulty is there’s a big complexity difference between “do you see porn” and “is this true and well sourced”. Also the article dramatically understated the speed of work. At least when I did moderation it was more like 1.5 seconds.
2 replies 0 retweets 3 likes
I think the prompt or act of forcing the original publisher or major sharers of info to add evidence and citations would change behavior, incentives. I don’t think you’d need human mods check each one, let alone more than a fraction of them.
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.