Hate Algolengua: How Online Speech Evades Social Media Algorithms

Hate Algolengua: How Disguised Speech Bypasses Algorithms

Platforms do not remove hostile speech; they force it to disguise itself. This is the thesis Newtral presented in a report by Luis Mejía García, shared by Ana Pastor. The text describes an algolengua—a parallel vocabulary of nicknames and circumlocutions—designed to evade automated filters targeting insults and generalizations. The warning backfired on its author.

What is 'algolengua' and how does it evade moderation?

The mechanism is easy to state but hard to stop. When an offensive term enters a platform’s blacklist, speakers do not abandon the idea; they change the word. Phonetic distortions, euphemisms, and letter games emerge, unrecognized by automatic filters but instantly deciphered by the receiver. This replacement dictionary, according to the report, allows messages to spread with less risk of blocking.

Comments repeat the idea that using codes to bypass censorship is nothing new: «human ingenuity will always be above censorship», one message summarizes.

The Counterargument: Media 'Neolengua'

The most repeated response inverted the frame. If there is a language of hate, the argument goes, there is equally a language of power: one that chooses euphemisms to soften uncomfortable realities and then feigns scandal when the public does the same. Several messages claim media outlets have been manipulating vocabulary for «half a century», arguing the complaint sounds like double standards. The accusation targets not the phenomenon, but the accuser.

The same package includes specific criticisms of journalistic practices: the use of anonymous sources, the reformulation of events with kinder terms, and the sense that official vocabulary changes according to the narrative’s convenience. These are debatable method criticisms, but criticisms nonetheless.

Why Many Read 'Algorithm' as Synonym for Censorship

The headline’s phrasing itself is the trigger. If one must bypass an algorithm, many readers reason, it is because that algorithm acts as a customs checkpoint: calling it moderation sounds like makeup. The equation «algorithm = censorship» repeats endlessly, shifting focus from vocabulary to the infrastructure pursuing it. It is also noted that the account sharing the warning has comments closed, a detail used as proof of incoherence.

From Irony to Insult: The Dominant Tone of Response

Much of the response does not discuss the thesis. It attacks the author. Personal comments, gratuitous disqualifications, and nicknames abound—material offering no data yet occupying the bulk of the volume. This mix is revealing: alongside reasonable objections about the double standard of language coexist explosions of anger that unintentionally confirm the phenomenon the report described.

One message rescues a previous episode from the channel to ridicule its verification method. It is revival, not analysis, but explains why the brand carries antibodies.

The 'Ethnic Substitution': A Thesis the Material Does Not Support

One message elevates the discussion, claiming immigration trinc a planned ethnic substitution. The assertion is launched without a single supporting data point and does not correspond to any documented demographic process: it is a conspiracy theory circulating online, not a verifiable fact. It must be separated from legitimate criticism of institutional language, because discussing how things are named is one thing, inventing a plan no one has proven is another.

What to Expect from This Race

It is predictable that vocabulary will continue mutating faster than filters, and each banned term will spawn two substitutes. Platforms will update their lists, speakers will bypass them again, and the gap between norm and usage will widen further. Whether this race is won by automated moderation or human ingenuity remains an unanswered question today.