In social media, algospeak is a self-censorship phenomenon in which users adopt coded expressions to evade real or imagined automated content moderation. It allows users to discuss topics deemed sensitive to moderation algorithms while avoiding penalties such as shadow banning, downranking, or de-monetization of content. A type of netspeak, algospeak primarily serves to bypass censorship, though it can also reinforce group belonging, especially in marginalized communities. Algospeak has been identified as one source of linguistic change in the modern era, with some terms spreading into everyda
In social media, algospeak is a self-censorship phenomenon in which users adopt coded expressions to evade real or imagined automated content moderation. It allows users to discuss topics deemed sensitive to moderation algorithms while avoiding penalties such as shadow banning, downranking, or de-monetization of content. A type of netspeak, algospeak primarily serves to bypass censorship, though it can also reinforce group belonging, especially in marginalized communities. Algospeak has been identified as one source of linguistic change in the modern era, with some terms spreading into everyday offline speech and writing. The term has been used more broadly to include any language change driven by digital usage.
== History == The term algospeak–a blend of Algorithm and -speak—appears to date back to 2021, though related ideas have existed for longer. In 2018, the internet researcher Emily van der Nagel coined the terms Voldemorting and screenshotting, two strategies social media users use to avoid giving attention to objectionable figures or attracting algorithmic attention from unwanted audiences. The term Voldemorting references the fictional character from the Harry Potter series, also known as "You-Know-Who" or "He-Who-Must-Not-Be-Named", and involves obfuscating the referent of a post by avoiding directly mentioning a name or a term. Screenshotting refers to the act of sharing screenshots instead of machine-readable text.
Discovered by embedding cosine similarity (sentence-transformers MiniLM, 384-dim).