Disinformation
The deliberate creation and spread of false or misleading information to deceive audiences and achieve political objectives
Disinformation
Disinformation refers to false or misleading information that is deliberately created and distributed to deceive audiences and achieve specific political or strategic objectives. The concept originates from Soviet-era “dezinformatsiya,” with a formal disinformation department established within the Soviet intelligence apparatus in the early 1920s. The term has been documented in political and intelligence contexts since that period and gained renewed prominence in American political discourse in the 2010s, particularly following the 2016 U.S. presidential election and investigations into foreign interference campaigns. Digital platforms are widely described as having expanded both the reach and speed of disinformation campaigns.
1. Una afirmación en tu feed
Alguien a quien sigues comparte una publicación con una afirmación. Nadie la ha verificado todavía, y nada en la publicación indica si es cierta. Decide si la compartes y observa a las personas que te siguen.
2. Se propaga, y le sigue una corrección
Amplía la vista a cuarenta personas. La afirmación se propaga desde la cuenta en la que la viste, y desde ti si la compartiste. Cada persona decide una sola vez, la primera vez que le llega. Unas rondas después, un verificador en el otro extremo determina que la afirmación es falsa, y la corrección se propaga con la misma regla. Quien recibe primero la corrección no comparte la afirmación. Pulsa reproducir.
3. Una etiqueta
La misma red y el mismo comienzo. Activa la etiqueta: una vez verificada la afirmación, la publicación lleva una etiqueta, y en este modelo quienes la ven etiquetada la comparten con algo menos de frecuencia. Compara con el paso 2: un anillo discontinuo señala a quien ya no la comparte.
4. Una pregunta sobre la exactitud
Ahora, en lugar de una etiqueta, se pregunta a todos qué tan exacta es una publicación antes de compartirla. Nadie tiene que verificar antes la afirmación, así que en este modelo actúa desde la primera ronda. Actívala y compara con el paso 2.
5. Lo que encontraron los estudios
Los pasos 2 a 4 muestran cómo podría funcionar cada respuesta en una red simplificada. Estos estudios midieron partes de ello, y sus resultados incluyen efectos pequeños y desacuerdos.
Tu feed
Compartida por alguien a quien sigues
Aún sin verificar
Los verificadores la calificaron como falsaSe añade una vez verificada la afirmación
Antes de compartir: ¿qué tan exacto es esto?
Personas que te siguen
- Cuenta 1 No la ha visto
- Cuenta 2 No la ha visto
- Cuenta 3 No la ha visto
- How far false stories spread. A 2018 study compared about 126,000 true and false stories on Twitter, classified by six independent fact-checking organizations. False stories spread farther, faster, deeper and more broadly than true ones, most of all for political news. Automated accounts spread true and false stories at the same rate, so the authors attributed the difference to people. Leer el estudio
- Corrections (taking a share back, which the model leaves out). A 2014 study of rumors on Facebook found that a reshare that received a comment linking to a fact-check was more likely to be deleted, while large rumor cascades accumulated hundreds of such comments and kept spreading. Leer el estudio
- Labels (step 3). In a 2020 preregistered experiment, "Disputed" and "Rated false" tags lowered how accurate people judged false headlines to be, by relatively modest amounts, with "Rated false" having the larger effect. The tags did not change how people judged unlabeled headlines. Leer el estudio
- Labels on only some posts (step 3). A separate 2020 study found the opposite on that last point: labeling only some false headlines led people to rate unlabeled false headlines as more accurate, an effect that disappeared when some true headlines were marked as verified. Labels also modestly lowered how accurate people judged the labeled false headlines to be, and the study found the same effects in decisions about which headlines to consider sharing. Leer el estudio
- Accuracy prompts (step 4). A 2021 study found that whether a headline was true had little effect on whether people said they would share it, although it strongly affected whether they judged it accurate. Prompting them to think about accuracy increased the quality of news they then shared, in survey experiments and in a field experiment on Twitter. Leer el estudio
- How large the prompt effect is (step 4). A 2022 analysis of 20 experiments by the same research group found that prompts reduced intentions to share false headlines by about 10% relative to control. Leer el estudio
- An independent replication (step 4). A 2021 preregistered replication of one of the group's earlier studies, by other researchers, found no significant effect in its first round and a small, significant effect after collecting more data. Leer el estudio
Step 1: someone you follow shares a post making a claim that nobody has checked, and you choose whether to share it; if you do, the people who follow you see it. Step 2: in a network of forty people, the claim spreads from where you saw it, and from you if you shared it; a few rounds later a correction starts from a fact-checker and spreads by the same rule, and people it reaches first do not share the claim. Because the claim started earlier, it ends up reaching more people than the correction, whichever choice you made. Step 3: with a label added once the claim has been checked, the spread is the same until the check, and fewer people share the claim overall; dashed rings mark who no longer shares. Step 4: with everyone asked how accurate a post is before sharing, fewer people share the claim from the first rounds on, and the correction spreads as before. Step 5: a study found that false stories spread farther and faster than true ones on Twitter; studies of warning labels and accuracy prompts found modest or small effects on how people judged or would share false headlines, and researchers disagree about whether labels make unlabeled false headlines seem more accurate.
Cómo funciona
- Basado en
- Tambuscio, M., Ruffo, G., Flammini, A., & Menczer, F. (2015). Fact-checking Effect on Viral Hoaxes: A Model of Misinformation Spread in Social Networks. Proceedings of the 24th International Conference on World Wide Web (Companion), 977-982. This model keeps their idea of a false claim and its correction competing through one network, with the correction stopping people from spreading the claim. It drops their gullibility and forgetting, and adds a label and an accuracy prompt.
- Qué supone
- Step 1: the claim's content is left out, so nothing about it suggests whether it is true.
- Steps 2 to 4: forty simulated people sit on a loose grid, each linked to their nearest neighbors, and whatever someone shares reaches the people linked to them in the next round. Each person decides once, the first time something reaches them, by comparing a fixed readiness of their own with the chance of sharing.
- The claim and the correction spread by the same rule, at the same chance of sharing. The only difference is timing: the claim starts at once, and the correction starts a few rounds later from one fact-checker on the far side of the network. Which of the two reaches more people depends on that head start and on where the fact-checker sits; on this network, the claim does.
- Anyone the correction reaches before the claim does not share the claim. Someone who has already shared the claim does not take the share back, and may still pass on the correction; the picture keeps showing them as having shared the claim.
- Your step 1 choice carries into steps 2 to 4, and the label and prompt do not change it.
- Step 3: from the round the claim is checked, a label lowers everyone else's chance of sharing it. The label studies mostly measured how accurate people judged headlines to be, and one also measured which headlines people would consider sharing; the model shows a label only as a smaller chance of sharing.
- Step 4: from the first round, a prompt lowers everyone else's chance of sharing the claim, and leaves the correction as it was. Each step turns on only its own response.
- Both effects are modest, roughly the size the studies reported or a little larger. A dashed ring marks everyone a label or prompt kept from sharing, so the change shows even when it is small.
- Qué deja fuera
- Who made the claim, and why: deliberate disinformation and honest mistakes spread by the same rule here.
- Whether a fact-check is right, and disputes over fact-checkers' judgments: here the check is correct by construction.
- Bots, coordinated accounts, ranking algorithms and spread across platforms, which the article describes; here only people pass the claim on.
- Taking a share back after seeing a correction.
- Belief: the model tracks who shares, not what anyone believes.
- Whether labeling some false posts makes unlabeled ones seem more accurate. Studies disagree, and step 5 shows both results; here a label affects only the post it is on.
- Qué no demuestra
- How large any effect is on a real platform, or which response works best. Step 5 summarizes what studies measured; the effects of labels and prompts they found were modest or small.
Key Characteristics
Intentional Deception: Unlike misinformation (false information spread without intent to deceive), disinformation is deliberately crafted to mislead.
Formatted Presentation: Some disinformation content mimics established news formats, incorporating design elements and attributed sources to appear credible.
Coordinated Distribution: Some disinformation campaigns use networks of real and fake accounts to amplify false narratives across multiple platforms simultaneously.
Spread Mechanisms
Disinformation spreads through multiple channels, including interpersonal sharing, broadcast media, direct messaging platforms, and algorithmically curated feeds.
Algorithmic Engagement: Disinformation is sometimes designed to trigger strong emotional responses that generate high engagement, which can lead some ranking algorithms to promote the content more widely.
Bot Networks: Automated accounts can rapidly share and promote disinformation, creating the appearance of independent, widespread support.
Cross-Platform Distribution: Content is distributed to reach audiences across multiple platforms simultaneously.
Common Techniques
- False Flag Operations: Fake content attributed to opponents
- Selective Editing: Real footage edited to change meaning
- Deepfakes: AI-generated fake videos or audio
- Astroturfing: Organized campaigns presented as spontaneous public support
- Source Laundering: False information cited through multiple sources
Documented Effects
Researchers and institutions have identified several associations between disinformation and changes in political behavior, including:
- Declining trust in institutions
- Increased difficulty distinguishing verified from unverified claims
- Documented instances of false information about voting procedures reaching voters
- Heightened political tensions
- Shifts in public discourse and decision-making
The extent and causal mechanisms of these effects remain subjects of ongoing research and debate.
Detection and Response
Efforts to combat disinformation include:
- Fact-checking organizations
- Platform content moderation
- Media literacy education
- Government disclosure requirements
- Academic research and monitoring
Measured Spread and Responses
Most measurements classify claims as true or false rather than by intent, so they cover false information whether or not it was spread deliberately.
Spread. A 2018 study in Science examined about 126,000 stories shared on Twitter from 2006 to 2017, classified as true or false using six independent fact-checking organizations. False stories spread farther, faster, deeper and more broadly than true ones in every category studied, and the difference was largest for political news. Automated accounts spread true and false stories at the same rate, so the authors attributed the difference to people rather than bots.
Corrections. A 2014 study of rumors on Facebook found that a reshare that received a comment linking to a fact-check on Snopes was more likely to be deleted, while large rumor cascades accumulated hundreds of such comments and kept spreading. Modeling work treats a false claim and its correction as competing spreads through the same network: in one 2015 model, the false claim died out entirely once the probability that people checked it passed a threshold.
Warning labels. In a preregistered experiment published in 2020, “Disputed” and “Rated false” tags lowered how accurate people judged false headlines to be, by relatively modest amounts, with “Rated false” having the larger effect; a general warning about misleading content also lowered belief in true headlines. A separate 2020 study found that labels modestly lowered how accurate people judged labeled false headlines to be, but that labeling only some false headlines led people to rate unlabeled false headlines as more accurate, an “implied truth effect” that disappeared when some true headlines were marked as verified; it found the same effects in people’s decisions about which headlines to consider sharing. The 2020 tag experiment did not find that tags changed how people judged unlabeled headlines.
Accuracy prompts. A 2021 study in Nature found that whether a headline was true had little effect on whether participants said they would share it, although it strongly affected whether they judged it accurate. Prompting people to think about accuracy, for example by asking them to rate a single headline, increased the quality of news they subsequently shared, in survey experiments and in a field experiment on Twitter. A 2022 analysis of 20 experiments by the same research group found that such prompts reduced intentions to share false headlines by about 10% relative to control. An independent preregistered replication of one of the group’s earlier studies found no significant effect in its first round and a small, significant effect after collecting more data.
Fuentes
- 01.
Vosoughi, S., Roy, D., & Aral, S. (2018). The spread of true and false news online. Science, 359(6380), 1146-1151. Study of about 126,000 stories shared on Twitter from 2006 to 2017 by about 3 million people, classified true or false using six independent fact-checking organizations. Source for: false stories spread farther, faster, deeper and more broadly than true ones in all categories, with stronger effects for political news; automated accounts spread true and false news at the same rate, so the authors attribute the difference to people. The study classifies stories by truth, not intent.
- 02.
Friggeri, A., Adamic, L. A., Eckles, D., & Cheng, J. (2014). Rumor Cascades. Proceedings of the International AAAI Conference on Web and Social Media, 8(1), 101-110. Study of thousands of rumors on Facebook referenced by Snopes. Source for: a reshare of a rumor that received a comment linking to a Snopes article was more likely to be deleted, and large cascades accumulated hundreds of such comments while continuing to spread.
- 03.
Tambuscio, M., Ruffo, G., Flammini, A., & Menczer, F. (2015). Fact-checking Effect on Viral Hoaxes: A Model of Misinformation Spread in Social Networks. Proceedings of the 24th International Conference on World Wide Web (Companion), 977-982. Model of a hoax and its debunking spreading through a network like competing infections. Source for: in simulations, the hoax was removed from the network when the probability of fact-checking passed a threshold that depended on gullibility and forgetting rather than on the spreading rate. The model the page's interactive explainer adapts.
- 04.
Clayton, K., Blair, S., Busam, J. A., Forstner, S., & Glance, J. (2020). Real Solutions for Fake News? Measuring the Effectiveness of General Warnings and Fact-Check Tags in Reducing Belief in False Stories on Social Media. Political Behavior, 42(4), 1073-1095. Author list truncated to the first five of 17; Brendan Nyhan is the senior author. Preregistered experiment. Source for: 'Disputed' and 'Rated false' tags and a general warning each lowered the perceived accuracy of false headlines, by relatively modest amounts that generally did not vary with whether headlines were congenial to respondents' politics; 'Rated false' lowered it more than 'Disputed'; the tags did not change the perceived accuracy of unlabeled headlines, while the general warning also lowered belief in true headlines.
- 05.
Pennycook, G., Bear, A., Collins, E. T., & Rand, D. G. (2020). The Implied Truth Effect: Attaching Warnings to a Subset of Fake News Headlines Increases Perceived Accuracy of Headlines Without Warnings. Management Science, 66(11), 4944-4957. Two online experiments (5,271 and 1,568 participants). Source for: warnings modestly reduced the perceived accuracy of false headlines, but untagged false headlines were then seen as more accurate than in a control condition (the 'implied truth effect'), in accuracy judgments and in sharing decisions; also marking some true headlines as verified eliminated the effect.
- 06.
Pennycook, G., Epstein, Z., Mosleh, M., Arechar, A. A., Eckles, D., & Rand, D. G. (2021). Shifting attention to accuracy can reduce misinformation online. Nature, 592(7855), 590-595. Four survey experiments and a field experiment on Twitter. Source for: whether a headline was true had little effect on participants' sharing intentions despite a large effect on their accuracy judgments; most participants said sharing only accurate news was important; subtly shifting attention to accuracy increased the quality of news people subsequently shared.
- 07.
Pennycook, G. & Rand, D. G. (2022). Accuracy prompts are a replicable and generalizable approach for reducing the spread of misinformation. Nature Communications, 13, 2333. Internal meta-analysis of 20 experiments by the authors' group, 2017 to 2020, with 26,863 participants in total. Source for: accuracy prompts improved sharing discernment relative to control, mainly by reducing sharing intentions for false headlines by 10% relative to control.
- 08.
Roozenbeek, J., Freeman, A. L. J., & Linden, S. v. d. (2021). How Accurate Are Accuracy-Nudge Interventions? A Preregistered Direct Replication of Pennycook et al. (2020). Psychological Science, 32(7), 1169-1178. Independent preregistered replication of an earlier accuracy-prompt study on COVID-19 headlines. Source for: the first stage (701 participants) did not replicate the effect; after a second round of data collection (1,583 participants pooled), it found a small but significant effect.
Historial de la página
Revisada el 26 oct 2026Cambios recientes
- Revisada: 10 hallazgos — 7 corregidos, 3 abiertos Revisión con IA
- Revisada: 5 hallazgos — 2 corregidos, 1 descartados, 2 abiertos Revisión con IA
- Revisada: 1 hallazgo — 1 abiertos Revisión con IA
Related Dynamics
Sin Eventos en la Cronología
No hay eventos en la cronología actualmente asociados con esta entidad.
11 Nodes • 26 Connections