When all hell broke loose on Facebook during the Cambridge Analytica era, people were focused largely on privacy, but equally important was the realization of its recommendation algorithms.
Facebook set policy on what was considered to be too violent, too sexual, and too everything else, and then moderated and removed based on whether or not content crossed the policy line. But as people got closer to the line, their content was shared more, and thus, they were more rewarded, leading to a platform full of the most extreme content allowed, where the most rewarded content sat right next to the boundary line.
One of the main reasons twitter, around that same era in the twenty-teens, was safer and more trusted, was because of how they managed, algorithmically, this approach to the policy line. Twitter explicitly acknowledged that some behavior did not violate the rules but still harmed conversations, and that instead of removing it, Twitter reduced its visibility. Essentially, they flipped the curve.
As content approached the prohibited boundary line, twitter demamplified it, removed it from recommendations, restricted its appearance in search, and used other methods of reducing incentive to spread.
This all being then, it’s quite interesting because of where we are now. Have you seen what’s on twitter these days? The policy line has been shifted way to the right and the most extreme content gets the most rewarded.
In addition to the extensive pornography and graphic war violence, twenty-eight-year-old Dalton Eatherly, known as Chud the Builder, is a quintessential 2026 example of how the platform changed under Elon Musk.
Eatherly used the platform to become nationally known by livestreaming himself walking through towns in the Deep South, directing extreme racial slurs at random strangers before challenging them to “do something” that would justify him defending himself, including with the gun he carried. The slurs were so outrageous and disgusting they were literally jaw-dropping.
People shared the videos largely in disbelief that someone could openly do this. Eatherly claimed he was exercising free speech, and Musk’s platform did not moderate the content, showing that it fell within acceptable policy boundaries. The more controversial and extreme Easterly became, the more he was rewarded. I watched in disbelief, convinced he was eventually going to kill someone or be killed. It was clear he sought confrontations and deliberately provoked people because the platform rewarded that behavior.
Eatherly eventually did shoot someone during one of these filmed confrontations and is now in jail facing attempted murder. Even afterward, the content remained on the platform, and Eatherly raised more than $100,000 from supporters for his legal defense.
Unfiltered discussions and videos of street confrontations, assaults, war and even killings are surprisingly abundant on Musk’s platform. As for the cultural impact, of course it provokes outrage, anger, and engagement, ultimately feeding money back into the platform.
Every boundary creates an incentive gradient. The platform determines one boundary, the law establishes another, and creators adapt their behavior toward whichever boundary yields maximum reward. Incentivemaxing. This is the era we are in now.