The shift persisted even after users left the algorithm
En el verano de 2023, casi cinco mil usuarios de X en Estados Unidos participaron sin saberlo en un experimento que revelaría algo incómodo sobre la naturaleza del pensamiento político en la era digital: los algoritmos no solo reflejan nuestras creencias, sino que las moldean. Un estudio publicado en Nature por investigadores europeos encontró que el feed algorítmico de X desplazó a sus usuarios hacia posiciones conservadoras de manera medible y duradera, incluso después de que dejaron de usarlo. En un tiempo en que las plataformas digitales se han convertido en la plaza pública de millones, esta evidencia plantea una pregunta que trasciende la política: ¿quién, o qué, está construyendo la realidad en la que habitamos?
- El algoritmo 'For You' de X no era neutral: promovía activamente contenido conservador y reducía la visibilidad de medios de comunicación tradicionales, inclinando la balanza política de sus usuarios.
- Lo más perturbador no fue el sesgo en sí, sino su persistencia: cuando los usuarios regresaron a un feed cronológico, sus nuevas actitudes políticas no desaparecieron con el algoritmo.
- La mayoría de los usuarios desconoce que está siendo clasificada, filtrada y dirigida, lo que convierte a los algoritmos en una forma de influencia invisible pero profundamente real.
- Los investigadores advierten que el estudio se limita a X y a un período de siete semanas, dejando abiertas preguntas urgentes sobre efectos a largo plazo y en otras plataformas como TikTok o Instagram.
- La comunidad científica y los reguladores enfrentan ahora la presión de responder: si los algoritmos construyen realidades políticas, ¿quién tiene la responsabilidad de supervisarlos?
Un equipo de investigadores de Italia, Suiza y Francia se propuso responder una pregunta que lleva años incomodando a los críticos de las redes sociales: ¿los algoritmos realmente cambian cómo pensamos sobre política, o simplemente nos muestran lo que ya creemos? La respuesta, publicada esta semana en Nature, resultó más inquietante que cualquiera de las dos opciones por separado.
El estudio siguió a casi cinco mil usuarios de X en Estados Unidos durante siete semanas en el verano de 2023, unos seis meses después de que Elon Musk adquiriera la plataforma. Los participantes fueron asignados aleatoriamente a dos experiencias distintas: el feed algorítmico 'For You' o el timeline cronológico clásico. Antes y después del experimento, los investigadores midieron sus opiniones políticas y monitorearon, mediante extensiones de navegador, qué contenido veían y cómo interactuaban con él.
Los resultados fueron contundentes. Los usuarios del feed algorítmico no solo usaron la plataforma con mayor frecuencia, sino que desplazaron sus posiciones políticas de manera medible hacia la derecha y comenzaron a seguir a más activistas conservadores. El algoritmo, lejos de ser neutral, dirigía activamente a las personas hacia un extremo del espectro político, amplificando contenido conservador y reduciendo la visibilidad de los medios tradicionales.
Pero el hallazgo más perturbador llegó después: cuando los investigadores trasladaron a esos usuarios de vuelta al feed cronológico, sus nuevas actitudes políticas no se revirtieron. El cambio había echado raíces. Esto sugiere que los algoritmos no solo influyen en lo que vemos en el momento, sino que reconfiguran nuestra orientación política de formas que sobreviven al propio algoritmo.
Los científicos son cautelosos con sus conclusiones: el estudio se centra únicamente en X y siete semanas es una ventana corta. Aún no se sabe qué ocurre en meses o años, ni si efectos similares se producen en otras plataformas. Pero la implicación es clara: los algoritmos que median nuestro entorno informativo no organizan simplemente el contenido. Construyen la realidad política que habitamos, y lo hacen en silencio, sin que la mayoría de los usuarios sepa siquiera que está siendo dirigida.
A team of researchers from Italy, Switzerland, and France set out to answer a question that has nagged at social media critics for years: Do the algorithms that power platforms like X actually reshape how people think about politics, or do they simply show us what we already believe? The answer, published this week in Nature, is more troubling than either option alone.
The study tracked nearly five thousand X users in the United States over seven weeks in the summer of 2023, roughly six months after Elon Musk acquired the platform then known as Twitter. The researchers randomly assigned participants to one of two experiences: some saw the algorithmically curated "For You" feed, while others browsed the chronological timeline—the feed as it existed before algorithmic sorting took over. Before and after the experiment, they surveyed participants about their political views. They also installed browser extensions to monitor what content appeared on each user's screen and how they interacted with it.
The results were unambiguous. Users assigned to the algorithmic feed engaged with the platform more frequently. More significantly, they shifted their political views measurably to the right. They were more likely to follow conservative political activists. The algorithm, in other words, was not neutral. It actively steered people toward a particular end of the political spectrum.
But here is where the findings become genuinely unsettling. When researchers moved users from the algorithmic feed back to the chronological timeline, their political views did not revert. The shift persisted. This suggests that algorithmic exposure does not simply influence what you see in the moment—it appears to reshape your underlying political orientation in ways that outlast the algorithm itself. The damage, if you want to call it that, had already been done.
The researchers analyzed the content flowing through both feeds and found the mechanism at work. The algorithm promoted conservative posts and conservative activists at much higher rates than the chronological feed. It simultaneously reduced the visibility of traditional news outlets. The algorithm was not reflecting the political composition of X's user base; it was actively amplifying one side.
The scientists are careful about their conclusions. They note that their findings apply specifically to X and that seven weeks is a relatively short window. They cannot yet say what happens over months or years, or whether similar effects occur on other platforms. But the implication is clear: the algorithms that mediate our information environment do not simply organize content. They shape political reality itself. They do not just show us what exists; they construct the world we inhabit.
This matters because for millions of people, social media has become the primary source of news and political information. The algorithms that govern these platforms are not transparent. Most users do not know they are being sorted, filtered, and steered. The study suggests that this steering is real, measurable, and persistent—a form of influence that continues to work even after you think you have escaped it.
Notable Quotes
Algorithms shape political attitudes significantly, and this effect endures even after removing algorithmic curation— Study authors in Nature
The Hearth Conversation Another angle on the story
So the algorithm doesn't just show you more conservative content—it actually changes how you think?
That's what the data suggests, yes. And the unsettling part is that the change doesn't reverse when you leave the algorithm behind. It sticks.
Why would that be? Why wouldn't your views snap back once you're looking at a chronological feed again?
The researchers don't fully explain the mechanism, but think about it this way: you spend weeks following conservative activists, reading conservative posts, engaging with that ecosystem. You're not just seeing different content—you're building relationships, following new accounts, training yourself to interact with that side of the political spectrum. When you switch back to chronological, those relationships and habits remain.
So it's not the algorithm itself that's doing the long-term work—it's what the algorithm made you do while you were using it?
Partly, yes. But it's also about what you've been exposed to. Your sense of what's normal, what's credible, what matters—that shifts. The algorithm didn't just show you conservative content; it showed you conservative content at the expense of traditional news sources. Your information diet changed.
The study was only seven weeks. Do you think the effect would be stronger or weaker over months or years?
Almost certainly stronger. Seven weeks is enough to shift your views measurably. Imagine what happens over a year, or five years. This could be reshaping entire political generations.
And this is just X. What about the other platforms?
That's the question no one can answer yet. But if X's algorithm works this way, it's reasonable to assume others do too—maybe differently, but with similar intent. The algorithm is designed to keep you engaged, and engagement often means amplifying the extreme, the partisan, the emotionally charged.