Every specialism in this Part — emotions, the body, everyday life, the life course, applied work — is being remade, right now, by the same force: the migration of human life onto digital platforms. So this topic takes it head-on. It asks what is genuinely new about a social world that is increasingly mediated (lived through screens and platforms), recorded (leaving a permanent data trail), and sorted (shaped by algorithms deciding what you see) — and, just as important, what only looks new.
That double question is the discipline of this lesson, because the digital invites two opposite errors, both lazy. One is breathless novelty — the assumption that everything is unprecedented, that the old sociology is obsolete, that we need entirely new theories for an entirely new world. The other is dismissive continuity — the assumption that it is all just the same old human behaviour with new gadgets, nothing to see here. The truth, as usual in this course, is neither: some things about digital life are genuinely, structurally new and demand new concepts, while others are ancient patterns in new clothing that the classic perspective explains perfectly well. The skill is telling which is which — and a sociologist who can do that sees the digital world far more clearly than either the evangelist or the cynic.
The flu that the data couldn't predict.
When "big data" first arrived, it came with a thrilling promise: that with enough data, we would no longer need theory — the numbers would speak for themselves. The showcase example was a system that tried to predict flu outbreaks faster than health agencies could, simply by tracking how often people searched for flu-related terms online. At first it worked beautifully, spotting outbreaks well ahead of official statistics. It looked like the future: no surveys, no theory, just the vast exhaust of human behaviour revealing reality directly.
Then it failed — badly. Over time the predictions drifted far from the truth, at one point overestimating flu prevalence by roughly double. Why? Because the data was not a neutral window onto reality. Search behaviour was shaped by the platform itself — by media panic driving people to search when they were not sick, by the search engine's own changing suggestions nudging what people typed, by the fact that searchers are not a representative sample of the population. The "data" was not raw human behaviour; it was behaviour already shaped by the platform that recorded it — and a model that ignored that was measuring its own reflection.
This failure became a founding lesson of digital sociology, and it cuts against the breathless-novelty error directly. More data does not mean more truth. Data is never raw; it is produced — by specific platforms, with specific designs and incentives, capturing specific slices of specific populations. The dream that big data would let us skip theory and read reality straight off the numbers turned out to be exactly backwards: the more data we have, the more we need the sociological questions — where did this data come from, who is in it and who is missing, and how did the platform shape the very behaviour it recorded? The flu that the data couldn't predict is the reason digital sociology exists as a critical discipline rather than a cheerleading one.
The plain-words core: what is actually new
Let us name, carefully, the things about digital life that are genuinely structural changes — not just new gadgets, but new conditions that alter how social life works. Four stand out.
One — Datafication. More and more of social life is turned into data — recorded, quantified, stored, and made analysable. Your movements, purchases, messages, glances, hesitations, and relationships increasingly leave a machine-readable trail. This is genuinely new in scale and kind : for the first time, vast swathes of ordinary human behaviour are continuously captured as data by default. Datafication changes what can be known about people, by whom, and to what end — and it is the raw material of everything else in this topic.
Two — Platform mediation. More and more interaction happens through platforms — privately owned digital spaces (for search, social contact, shopping, work, dating) that structure the interaction they host. A platform is not a neutral pipe; it is designed , with rules, defaults, and incentives baked in, and those design choices shape what you can do and how (what sociologists call affordances — what the design makes easy, hard, or impossible). When your social life runs through a platform, the platform's design becomes part of the structure of your social life.
Three — Algorithmic sorting. More and more of what you see is chosen for you by algorithms — automated systems deciding which posts, results, products, prices, and people are shown to you, usually to maximise the platform's goals (engagement, sales, time-on-site). This is a new kind of power: not telling you what to think, but shaping the informational world you think within — a machine operating at the second face of power (11.1.3), setting the agenda of what even reaches your attention.
Four — Scale and speed. Interactions, information, and coordination can now happen at a scale and speed without precedent — a message to millions in seconds, a movement organised overnight (11.2), a rumour or a market panic circling the globe before lunch. Quantitative change this large becomes qualitative: some things are possible at digital scale and speed that were simply impossible before.
Going deeper: what only looks new
Now the other half of the discipline — the equally important recognition that much of what feels unprecedented about digital life is old , and yields to the classic perspective without new theory.
The needs are old; the medium is new. People online are doing what people have always done — seeking status, belonging, connection, information, love, distraction, a fight. The drives are ancient; only the arena is new. The teenager crafting an online image is doing Goffman's impression management (front stage, 12.1.2) with new tools; the influencer is a status-seeker in a new market; the online mob is a moral panic (10.3.2) at fibre-optic speed. Reach for the classic concepts first; they explain more of digital life than the novelty-merchants admit.
Inequality reproduces online, it does not dissolve. The early dream that the internet would be a great leveller — everyone equal, identity irrelevant, hierarchy abolished — was wrong (12.3.2 develops this). The inequalities of Part 8 reproduce in digital space: who has access, who has the skills, whose content is amplified, who is harassed off the platform — all track existing lines of class, race, and gender. The digital did not escape the social order; it carried it in .
Community and identity online obey familiar sociology. Online communities form, police their boundaries, generate norms and insiders and outsiders, and produce solidarity and exclusion — exactly as offline groups do (Part 2). Online identity is performed, negotiated, and shaped by audiences, exactly as offline identity is. The medium adds twists, but the underlying sociology of group and self is continuous.
The genuinely new twists — where old patterns get new properties.
The most interesting cases are neither purely new nor purely old but old patterns given new properties by the medium — and these are where digital sociology does its sharpest work. Two examples:
Context collapse. Offline, you present different versions of yourself to different audiences — one self to your boss, another to your friends, another to your grandmother — and these audiences are naturally separated by time and space (12.1.2 on the managed self; Goffman's separated stages). Online, on a single platform, all these audiences can collapse into one : a single post is seen by boss, friend, and grandmother at once. The old sociology of self-presentation still applies — but the platform has removed the separation of audiences that made managing multiple selves possible, creating a genuinely new predicament (danah boyd's context collapse ). Old pattern (managed self), new property (collapsed contexts).
Persistence, searchability, and scale of the trace. Offline speech is ephemeral — said and gone. Online speech is persistent (recorded), searchable (findable years later), and scalable (spreadable to millions). A careless remark that would once have died in the air can now surface a decade later before a vast audience. The sociology of reputation and stigma (Goffman again) still applies — but persistence and searchability give it new and harsher properties. Old pattern (reputation), new property (a permanent, searchable, scalable record).
This is the pattern to look for : not "everything is new" and not "nothing is new," but which old sociological pattern is operating, and what new property the medium has given it. That question — asked case by case — is the whole method, and it is far more illuminating than either the evangelist's "this changes everything" or the cynic's "there's nothing new here."
The methods question: computational sociology and its traps
The digital does not just give sociology a new object (online life); it gives new methods — the tools of computational sociology , which uses the vast new data and computing power to study the social world in ways not possible before.
Computational sociology uses large-scale digital data and computational methods to study social life. Its main tools:
Big data / digital trace data — analysing the enormous behavioural records left by digital activity (posts, clicks, movements, transactions).Network analysis at scale — mapping the structure of relationships among millions of people, testing how things (information, behaviour, disease) spread through social ties (a rigorous version of the network ideas behind 11.2.2's recruitment finding).Simulation / agent-based models — building artificial societies of simple "agents" following rules, to watch how large-scale patterns emerge from individual actions (1.3.2 on emergence, made computational).Machine learning — using algorithms to find patterns in, or make predictions from, social data. These are genuinely powerful, opening questions that surveys and interviews never could. But they come with traps that the flu story already flagged, and that the discipline has learned to watch for.
The four traps of computational sociology — why more data is not automatically more truth.
The "n = all" illusion. Big data feels like it captures everyone , so representativeness seems irrelevant. It is an illusion: digital data captures only those on the platform, behaving in trackable ways — a specific, skewed slice of humanity. The people not on the platform, or behaving in untracked ways, are invisible, and conclusions drawn as if the data were "everyone" are systematically biased toward whoever the platform over-represents.
Data is produced, not found. As the flu case showed, digital data is shaped by the platform that captures it — its design, its incentives, its own algorithms nudging behaviour. You are often measuring the platform's effects, not the underlying social reality. The data is a mirror with the platform's fingerprints all over it.
Correlation without mechanism. Machine learning excels at finding patterns and predicting, but a pattern is not an explanation (Part 7). A model can predict without telling you why , and a prediction with no mechanism is fragile — it breaks, silently, when the world shifts (as the flu model did), because it never understood what it was tracking.
Bias in, bias out. Algorithms trained on historical data learn the historical patterns — including the injustices . A hiring or policing or lending model trained on a biased past will reproduce and automate that bias, now wrapped in the false authority of "the algorithm" and the "objectivity" of data (12.3.2 develops this). The math launders the prejudice.
The unifying lesson is the one the flu taught: computational methods amplify sociology's power only when they are guided by sociological questions. Without theory — without asking where the data came from, who is missing, and what mechanism is at work — big data does not deliver truth; it delivers confident, large-scale error. The numbers never speak for themselves; they speak in the voice of whoever collected them and whatever shaped them.
Two studies of the same online conversation.
Imagine two researchers study political talk on a large social platform.
The naïve computational researcher scrapes millions of posts, runs a model, and announces "what the public thinks" about an issue based on the balance of posts. The error is fourfold and fatal: the posters are not the public (n = all illusion — most people never post); the platform's algorithm decided which posts were amplified and thus which got made (data is produced); the model found what correlates with what but not why (no mechanism); and if it predicts future opinion, it will break the moment the platform changes its algorithm (correlation without mechanism). The confident finding is confidently wrong.
The sociologically-guided researcher uses the same data but asks first: who is in this data and who is missing? how did the platform's design and algorithm shape what got posted and seen? what mechanism might explain the pattern, and how could I test it? She treats the digital trace not as a window onto public opinion but as a record of behaviour on a specific platform under specific conditions — and draws far narrower, far sounder conclusions. She may discover less, but what she discovers is true.
Same data, opposite quality of knowledge — and the difference is entirely the sociological questions brought to it. This is why digital and computational sociology is not a replacement for the discipline but the discipline applied to new material with new tools: the perspective is what turns data into understanding rather than into large-scale illusion.
Why this matters
Because you live a large and growing share of your life inside these platforms, and the two lazy stories about them — it changes everything and it changes nothing — are both told to you constantly, by people selling either a revolution or a shrug. This lesson gives you the third, truer stance: ask, case by case, what is genuinely new and what only looks new — which structural condition (datafication, platform mediation, algorithmic sorting, scale) is actually operating, and which timeless human pattern (status, belonging, impression management, moral panic, inequality) is simply wearing new clothes, perhaps with a sharp new property. That question protects you from both the evangelist and the cynic, and it lets you see your own digital life with the same double vision — recognising the ancient drives in your online behaviour and the genuinely new conditions those drives now operate under.
And it arms you against the most seductive claim of the data age: that the numbers speak for themselves, that with enough data we can dispense with theory and read reality straight off the screen. The flu that the data couldn't predict is the standing refutation. Data is never raw, never neutral, never "everyone" — it is produced by platforms with designs and interests, capturing some people and missing others, shaping the very behaviour it records. The more of your world is datafied, sorted, and mediated, the more — not less — you need the sociological questions: where did this come from, who is in it, who is missing, whose interests shaped it, and what is really going on beneath the pattern? In a world drowning in data, the perspective that asks those questions is not obsolete. It is the only thing that turns the flood of data into understanding rather than into confident, automated, large-scale error.
The final lesson of this topic (12.3.2) takes the sharpest edge of all this — the turning of datafied life into a resource to be owned and exploited , the new inequalities the digital produces, and the automated systems that increasingly sort human lives — and asks who benefits, and who pays, in the datafied society.
Datafication — the transformation of ever more of social life into recorded, quantified, analysable data, at unprecedented scale and by default.
Platform — a privately owned digital environment that hosts and structures interaction through its design, rules, defaults, and incentives; not a neutral pipe.
Affordances — what a technology's design makes easy, hard, or impossible, thereby shaping the behaviour it hosts.
Algorithmic sorting — automated selection of what each user sees (posts, results, prices, people), typically to maximise the platform's goals; a form of agenda-setting power.
Context collapse — the flattening of multiple distinct audiences (boss, friends, family) into one on a single platform, undermining the audience separation that ordinary self-presentation relies on (danah boyd).
Computational sociology — the use of large-scale digital data and computational methods (big data, network analysis, simulation, machine learning) to study social life.
The "n = all" illusion — the false belief that big data captures everyone, when it captures only those on a platform behaving in trackable ways, biasing conclusions toward whoever the platform over-represents.
Try it yourself:
One — Run the new/old test. Pick something about your online life that feels distinctly modern (curating a feed, ghosting, going viral, doom-scrolling). Ask: what ancient human pattern is underneath (status, belonging, avoidance, vigilance)? Then ask: what genuinely new property has the medium added (scale, persistence, algorithmic amplification, collapsed context)? Almost everything decomposes into old-pattern-plus-new-property — and seeing both halves is the whole skill.
Two — Find the platform's fingerprints. Think of some "data" about people you have seen quoted (what a platform's users "think," "want," or "do"). Ask the flu questions: who is in this data and who is missing ? How did the platform's design and algorithm shape the behaviour being measured? You will usually find the data is a record of the platform , not of the public.
Three — Watch an algorithm set your agenda. For a day, notice that you did not choose most of what you saw — an algorithm did, to serve its goals, not yours. Ask what it showed you a lot of, and what it therefore kept off your attention. That is the second face of power (11.1.3) operating in your pocket: not telling you what to think, but quietly setting the world you think within.
What's next: 12.3.2, Surveillance Capitalism, AI and Digital Inequality — we follow datafication to its sharpest edge: what happens when your recorded life becomes a resource to be owned and sold, when automated systems begin sorting human beings at scale, and when the digital world's promise of equality gives way to new hierarchies of access, skill, and exposure. Who benefits, and who pays, in the datafied society.