Social media no longer just reflects reality; it quietly constructs it through invisible systems of algorithms, attention and influence
Image Source: Pixabay
In recent years, social media has not merely reflected public opinion but has actively shaped it. From elections and political polarisation to misinformation during global crises, viral outrage campaigns, cancel culture and even the emotional tone of public discourse, platforms such as Instagram, X, Facebook, TikTok and YouTube increasingly influence how societies perceive reality itself. What trends online often begin to feel true offline. The line between public sentiment and algorithmically amplified sentiment has become increasingly blurred.
In 1976, Don Henley, Glenn Frey and Don Felder of the band The Eagles wrote the famous line in Hotel California, “We are all just prisoners here, of our own device”, little could they have imagined how relevant it would become decades later, with the advent of mobile phones and social media. Today, most of us begin and end our day with social media. By continuously consuming content, we view social media as a random stream of posts, reels, videos, opinions and news. But that is just what appears on the surface. We tend to believe that our topics of interest and scrolling behaviour determine what appears on our feeds. However, that is only a part of the truth, and the real mechanism is far less visible. What shapes these feeds is actually governed by systems that decide what we see, when we see it and how often we see it. It is this invisible layer, hidden under the hood, that has the power to influence our interpretation of the world and the events that occur in it. This issue warrants an examination of that invisible layer of social media—the hidden mechanisms of algorithmic curation, engagement engineering and attention manipulation that increasingly mediate modern human experience, often without users fully realising their influence.
What trends online often begin to feel true offline. The line between public sentiment and algorithmically amplified sentiment has become increasingly blurred.
Less than a decade ago, information online was sequential, chronological and user-driven. Today, it is non-linear, real-time, system-driven, and largely algorithmically curated. Platforms are designed not just to host content but also to optimise engagement. Any content that keeps users scrolling, reacting, and sharing makes a stronger economic case for amplification than for accuracy. Hence, information is no longer just consumed but selected and amplified based on predicted behavioural response.
While it may sound abstract, its effects are visible all around us. Every day, one can observe how quickly some content gains traction. An informative post may pass unnoticed, while something provocative or emotionally charged travels more quickly across networks.
The phenomenon is not accidental. Multiple studies over the years have demonstrated that false, sensational or emotionally charged information tends to spread faster and more widely online than factual content, largely because it provokes stronger emotional reactions such as outrage, fear, surprise or anger. A widely cited 2018 study by researchers at the Massachusetts Institute of Technology (MIT), which analysed nearly 126,000 news stories shared on Twitter over a period of eleven years, found that false news spread significantly farther, faster and more broadly than truthful information. The researchers observed that falsehoods were more likely to evoke surprise and disgust, emotions that increase the likelihood of sharing and engagement. Similarly, during the COVID-19 pandemic, several studies examining the spread of misinformation across Facebook, WhatsApp, and YouTube found that emotionally provocative content such as conspiracy theories and misleading health claims consistently outperformed factual public health communication in terms of reach and engagement. The pattern has also been visible during elections, communal tensions and international conflicts, where algorithmically amplified misinformation often travels faster than verified reporting.
This dynamic is not merely a by-product of user behaviour. Platform designs today are meant to reinforce this phenomenon. Features like infinite scroll, autoplay and algorithmic recommendations are optimised to maximise attention. Over time, these systems learn user preferences and begin curating increasingly personalised feeds. The bottom line is engagement, not necessarily accuracy. In many cases, this creates a self-reinforcing loop where the more we engage with certain types of content, the more of it we are shown.
Less than a decade ago, information online was sequential, chronological and user-driven. Today, it is non-linear, real-time, system-driven, and largely algorithmically curated.
What’s easy to miss is how this plays out at a broader level. Research has shown that even minor changes in how content is presented can significantly influence user behaviour, perception and emotional response. Studies conducted by social media platforms themselves have demonstrated that altering the order, frequency or emotional tone of posts appearing in a user’s feed can affect not only what users engage with, but also how they feel and behave online. One of the most widely discussed examples was Facebook’s 2014 ‘emotional contagion’ experiment, in which researchers manipulated the emotional content appearing in the feeds of hundreds of thousands of users. The study found that users exposed to negative content were more likely to produce negative content themselves, while those exposed to positive content responded in a similarly positive manner.
Over time, these seemingly minor design interventions accumulate into powerful behavioural patterns, shaping not only what users consume, but also how they interpret information, form opinions and emotionally respond to the digital world around them. When applied across millions of users simultaneously, these effects become systemic and societal.
Coupled with the availability of large-scale behavioural data and advances in artificial intelligence, today it is possible to predict and influence user preferences with increasing precision. The Cambridge Analytica scandal brought this into public focus, highlighting how data-driven profiling could be used to target individuals with tailored messaging.
Another important, though often overlooked, dimension of the digital information ecosystem is how information travels through trust networks. Messages received from known contacts, including family members, friends or community groups, are often accepted with minimal scrutiny, particularly within closed digital spaces such as WhatsApp groups or private online communities. In such environments, the sender's perceived identity frequently serves as a substitute for independent verification of the content itself. Credibility becomes socially inherited rather than factually established.
This pattern has had serious real-world consequences. During the COVID-19 pandemic, for instance, misleading forwards circulated through trusted messaging groups contributed to panic, misinformation about treatments and distrust of public health institutions. Similar patterns have also been observed during elections, communal tensions and financial scams, where false information spreads rapidly because it originates from familiar and trusted sources.
Add to this the growing presence of artificial intelligence (AI)-generated content, and the information environment becomes even more difficult to navigate. Advances in artificial intelligence now allow the creation of highly realistic text, images, audio and video that can convincingly imitate authentic material. Deepfake technology, in particular, has emerged as a major concern. Today, fabricated videos and cloned voice recordings can be seen used in political misinformation campaigns, impersonation scams, and manipulated wartime content, as seen in Operation Sindoor.
The larger danger lies in the gradual erosion of trust itself. In a digital environment where fabricated content can appear authentic, even genuine evidence risks being dismissed as fake. Together, trust-based sharing networks and AI-generated synthetic media are creating an information ecosystem where emotional familiarity and technological sophistication are displacing factual verification.
The larger danger lies in the gradual erosion of trust itself. In a digital environment where fabricated content can appear authentic, even genuine evidence risks being dismissed as fake.
Social media today has globally enabled unprecedented access to information and participation in public discourse. But the shift of control from individuals to systems is not something that most of us have fully understood or internalised. We are no longer interacting with platforms to consume information. We are interacting with systems that are actively shaping what we see and what we believe.
This raises some simple but critical questions. If the flow of information is being curated, amplified and optimised at scale in ways that are not fully visible, how does the system decide what surfaces and what gets left out? Who takes responsibility for the social, political and psychological effects of these systems? And how much of their inner functioning should be transparent to those being influenced by them?
These concerns have increasingly prompted governments and policymakers worldwide to consider stronger regulation of digital platforms. The European Union’s Digital Services Act, for instance, seeks to impose greater transparency obligations on large technology companies by requiring disclosures around algorithmic systems, content moderation practices and online advertising. Several countries have also introduced or proposed laws to address misinformation, data privacy, deepfakes, and platform accountability. In India, debates around intermediary liability, digital misinformation and platform responsibility have similarly intensified in recent years.
The challenge is no longer merely about the content we consume online, but about understanding the invisible systems that shape that consumption in the first place.
Alongside regulatory efforts, there is also a growing global conversation around the idea of ‘accountable technology’, which is based on the principle that digital platforms and algorithmic systems should not operate without transparency, ethical oversight or public accountability. Advocates of accountable technology argue that platforms influencing billions of users cannot remain opaque systems driven solely by engagement metrics and commercial incentives. Instead, they must be subject to safeguards that balance innovation with democratic values, user rights and societal well-being.
Ultimately, the challenge is no longer merely about the content we consume online, but about understanding the invisible systems that shape that consumption in the first place. It may therefore be time to look beyond the screen itself and pay far closer attention to what lies beneath the surface.
Ashish Upreti is a serving Indian Army officer with over 25 years of experience in operations, crisis management and strategic communications.
The author acknowledges the use of ChatGPT 5.5 to conduct a preliminary literature survey and generate a draft outline for this paper, based on the author’s own research arguments provided to the model.
The views expressed above belong to the author(s). ORF research and analyses now available on Telegram! Click here to access our curated content — blogs, longforms and interviews.
Ashish Upreti is a serving Indian Army officer with over 25 years of experience in operations, crisis management and strategic communications. He has represented India ...
Read More +
PREV
