I take issue with this experiment's use of the term "shadowbanning". The simulation they describe tested the effects of showing a tweet to others based on the v

I take issue with this experiment's use of the term "shadowbanning". The simulation they describe tested the effects of showing a tweet to others based on the viewers' preferences. > the network might choose to show the content of a moderate user to a relatively right-leaning connection ... but block that same content from the timeline of a left-leaning connection That's not shadowbanning because the tweet is still being shown to some people. That's normal algorithmic sorting with down ranking. A true shadowban would result in a tweet not being shown to anybody. What they describe is more like de-listing from the global timeline but still being shown to followers. Now you're just talking about search bubbles. What's strange is they describe this as "neutral". > At first blush, the banning appears to impact every user more or less equally. How is that equal? What the viewer sees depends on their cohort. It's the same kind of manipulation that we've known about for years. The unique issue with a shadowban is how it affects the creator. Specifically a shadowban is intended to counteract ban evasion using multiple accounts. If you're going to make a conclusion about whether or not shadowbanning should be allowed it should be done by comparing silent versus explicit bans.