4 ms·

Other than a lot of people preferring the reverse chronological feed, I think the biggest complaint I see these days is the perception that the algorithm is bei

Other than a lot of people preferring the reverse chronological feed, I think the biggest complaint I see these days is the perception that the algorithm is being manually manipulated to promote/not promote certain topics and/or people. Given that Twitter/Facebook/Reddit/Tiktok are the behemoths in the room that all have the power to sway public opinion by either promoting or suppressing information, I think they should be regulated so that any topic that was promoted or suppressed by the algorithm or manual human intervention should show itself on the content just like an ad does. "This content has been quarantined by [Twitter Staff / Automated Review] and is only available to those with the direct url." would be a perfect example of such a message. From there, users should have control over if they want to see such quarantined content or if they want a "safe" experience that Twitter curates for them.

mschuster91 4y ago
> I think the biggest complaint I see these days is the perception that the algorithm is being manually manipulated to promote/not promote certain topics and/or people. The scary thing is, this isn't a conspiracy myth. Back in 2014 FB got caught at experimenting at how algorithm changes affected the mood of users [1]. [1] https://www.forbes.com/sites/kashmirhill/2014/06/28/facebook-manipulated-689003-users-emotions-for-science/ https://www.forbes.com/sites/kashmirhill/2014/06/28/facebook...
axg11 4y ago
This is only possible for content that is flagged in a binary way. Most of these systems will also have a score for content and up/down rank the content based on the score. Using a score is actually a way to reduce the harm done by the algorithm. A low score will reduce the number of people that see content but not remove it completely, preventing false positive censorship.
hooande 4y ago
This is wildly unlikely. the content scope of twitter is huge, the data scale is huge and they have so many engineers that it's difficult to keep "manual human intervention" confidential. Even something as simple as a keyword search, ie "hunter biden laptop", would require searching tens of millions of strings per hour. Distributed across thousands of servers, integrated with whatever dev ops set up they have. It just seems like a huge pain, for little practical gain. Their scale is so huge that I doubt twitter has much manual control over what people see, even if they wanted it. It seems much easier to just multiply weights
jdrc 4y ago
Why would it be ok to quarantine legal stuff. If i subscribed to it i should be getting it. Otherwise Twitter is falsely advertising its service

mschuster91 4y ago

> I think the biggest complaint I see these days is the perception that the algorithm is being manually manipulated to promote/not promote certain topics and/or people. The scary thing is, this isn't a conspiracy myth. Back in 2014 FB got caught at experimenting at how algorithm changes affected the mood of users [1]. [1] https://www.forbes.com/sites/kashmirhill/2014/06/28/facebook-manipulated-689003-users-emotions-for-science/ https://www.forbes.com/sites/kashmirhill/2014/06/28/facebook...

axg11 4y ago

This is only possible for content that is flagged in a binary way. Most of these systems will also have a score for content and up/down rank the content based on the score. Using a score is actually a way to reduce the harm done by the algorithm. A low score will reduce the number of people that see content but not remove it completely, preventing false positive censorship.

hooande 4y ago

This is wildly unlikely. the content scope of twitter is huge, the data scale is huge and they have so many engineers that it's difficult to keep "manual human intervention" confidential. Even something as simple as a keyword search, ie "hunter biden laptop", would require searching tens of millions of strings per hour. Distributed across thousands of servers, integrated with whatever dev ops set up they have. It just seems like a huge pain, for little practical gain. Their scale is so huge that I doubt twitter has much manual control over what people see, even if they wanted it. It seems much easier to just multiply weights

jdrc 4y ago

Why would it be ok to quarantine legal stuff. If i subscribed to it i should be getting it. Otherwise Twitter is falsely advertising its service