Is Going Crazy the Key to Traffic? Threads Lost Item Post Reveals Algorithmic Reward Paradox: When Malice Gets More Attention Than Kindness, Are Social Platforms Pushing Us Toward Conflict?

A Threads user posted under the 'lost and found' hashtag, saying they found a beautiful blue MacBook in a YouBike basket at Wende Station in Taipei Metro, and were hesitating whether to keep it or sell it. They wrote, 'After thinking about it, I decided to sell it, so I took it to a reseller right away,' ending with 'hehe, unexpected windfall.' This post received about 2,400 likes, 53 reposts, and 71 shares. Earlier, the same user posted that the laptop had already been handed over to the station's information desk. That post received 355 likes, 7 comments, 22 reposts, and 4 shares. The difference in likes is nearly 7-fold, and shares nearly 18-fold. The entire thread accumulated around 1 million views within twelve hours of posting. These numbers are significant because they turn a normally theoretical issue into an observable comparison. Two posts by the same user, at nearly the same time, on the same topic and event, differ only in narrative: portraying themselves as someone who intends to keep lost property, versus someone who returns it. A user screenshot both posts side by side, titling it simply: 'The current traffic gap.' This comparison image itself received 5,872 likes. Another user commented that this case is perfect for studying how absurd the Threads algorithm is, receiving about 10,000 likes. 'First time seeing someone use reverse psychology to find the owner,' received about 11,000 likes. The original poster replied: 'Because of this broken algorithm, it seems like you don't get traffic unless you go crazy,' which received 9,031 likes — one of the highest-engaged comments in the thread. A netizen summarized even more bluntly: 'Turns out 'crazy literature' is indeed traffic.' A second notable phenomenon is that criticism of the algorithm itself also became high-traffic content. The most engaged comments in the thread were all complaints or mockery of the algorithm. Criticism of the platform also gained massive interaction and could again become a signal exploitable by the recommendation system. This cycle reveals the stubborn nature of the problem: even resistance can be commodified. The poster later updated, saying they checked again with the information desk — the morning staff said they didn’t see the laptop when it was returned, suggesting the owner may have already claimed it. Three Studies Point to the Same Structure The observations made by Taiwanese users in the comments align closely with recent research. A 2021 study published in the Proceedings of the National Academy of Sciences (PNAS) analyzed 2.73 million posts on Facebook and Twitter, finding that content mentioning opposing political groups had a sharing rate about twice that of content mentioning one’s own group. Each additional opposing-group keyword increased sharing probability by 67%. A 2023 randomized controlled study in Nature Human Behaviour found that each additional negative word in a news headline increased click-through rates by about 2.3%. Platform design may amplify this effect. According to 2021 internal Facebook documents revealed by The Washington Post, the platform had increased the weight of emoji reactions in its ranking algorithm. Although mechanisms were later adjusted, internal research still found that high levels of angry reactions were strongly associated with misinformation, toxic content, and low-quality news. These three studies, from different data sources, all point to the same phenomenon: content that triggers strong emotions, conflict, or opposition tends to gain more user interaction and is more likely to be amplified by platforms. What’s Rewarded Is Performance, Not Behavior No one was proven to have actually done anything wrong. The laptop was returned, and the owner may have already claimed it. What happened was closer to a successful piece of performance art. And that’s precisely the issue. When 'performing malice' consistently yields higher rewards than 'reporting good deeds,' rational content creators learn to perform malice. This doesn’t require anyone to actually become worse — only for everyone to learn to add a bit of sarcasm, provocation, or a flaw that invites correction in their narratives. The original poster’s comment, 'Because of this broken algorithm, it seems like you don't get traffic unless you go crazy,' is an accurate description of this pricing mechanism by a content creator — and also a confession. Once this performance becomes the default syntax, the cost falls on third parties. Audiences cannot distinguish from narrative tone who is acting and who is genuine; long-term exposure to performed malice distorts perceptions of social reality. This is not a moral issue of a single user — the problem may lie more in the pricing mechanism. When a system can only measure 'how much reaction it triggered,' it systematically overvalues anything offensive. This 1 million-view thread will remain one of the most vivid public records on Taiwan’s internet about algorithmic reward structures. And its recorders are ordinary users who calculated the traffic gap themselves and posted the side-by-side screenshots. This pair of posts cannot prove that Threads’ ranking mechanism deliberately rewards malice, but it clearly presents a phenomenon worth studying: when the same event is framed as a moral conflict, it gains far more interaction than otherwise. Existing research also shows that negative language, political opposition, and strong emotions indeed more easily trigger clicks and shares. Anger prompts comments, absurdity prompts reposts, moral controversy prompts taking sides. As more people argue, the platform pushes the content to more users; as more users see it, the argument intensifies. Thus, while algorithms don’t directly create conflict, they establish a clear reward mechanism: the more exaggerated, the higher the traffic; the more extreme, the faster the spread; the more anger you provoke, the more visible you become. Over time, users learn these rules. Straightforward storytelling gets no attention? Make the story more outrageous. Normal discussion gets no interaction? Deliberately leave a provocative statement. Views not sharp enough? Divide the world into black-and-white. Everyone’s behavior is repeatedly reinforced by platform algorithms; perhaps we have already, unknowingly, been trained by social media to be more extreme, more divided, and less able to understand each other. Sources: Public Threads posts. https://www.threads.com/share/BAYCx_Oq9N/​ Rathje, S., Van Bavel, J. J., & van der Linden, S. (2021). Out-group animosity drives engagement on social media. PNAS, 118(26), e2024292118. Robertson, C. E., Pröllochs, N., Schwarzenegger, K., Pärnamets, P., Van Bavel, J. J., & Feuerriegel, S. (2023). Negativity drives online news consumption. Nature Human Behaviour, 7(5), 812–822. Merrill, J. B., & Oremus, W. (2021, October 26). Five points for anger, one for a 'like': How Facebook's formula fostered rage and misinformation. The Washington Post. Masnick, M. (2021, October 28). Let me rewrite that for you: Washington Post misinforms you about how Facebook weighted emoji reactions. Techdirt.