How the late internet turned the orienting response into a business model
Three hundred thousand years ago, you crouch at the edge of a clearing.
The afternoon is hot and nearly still. Insects rasp somewhere beyond the trees. Your simian hands (paws, really) are occupied, but a patch of grass suddenly moves against the wind.
Your head turns before you have decided to turn it.
The grass separates. A face emerges: low eyes, flattened ears, the pale flash of exposed teeth. Your gaze fixes upon it. The muscles in your neck tighten. Whatever occupied your mind a moment earlier disappears.
There is no internal debate about whether this deserves your attention. Attention has already been taken.
Psychology calls this the orienting response: the involuntary redirection of our senses towards something novel, sudden or potentially significant. Eyes move. The head turns. The body prepares. Processing resources are withdrawn from whatever we were doing and concentrated upon the interruption. Researchers have described it as a kind of circuit breaker—an automatic signal that something outside our present focus may matter more than whatever is inside it. The response can be triggered before we fully understand what the stimulus is.
That encounter did not create the reflex. It reveals the evolutionary bargain behind it.
An animal that ignored every rustle did not remain alive for long. Neither did one that spent all day investigating every movement in the grass. Survival required a system capable of interrupting itself quickly, assessing the change and deciding whether to flee, fight, approach or return to the task at hand.
Attention was never designed to tell us what is important in some abstract or philosophical sense. It tells us what might be important before enough information exists to know.
Movement, brightness, novelty, faces, danger and social disturbance receive priority because they have often carried consequences. A sudden expression may signal hostility. An unfamiliar sound may reveal a predator. A violation of group expectations may threaten trust, status or cooperation. The mind notices first and interprets second.
These are not bugs in human cognition. They are part of an extraordinarily useful system for remaining responsive to an uncertain world.
Then we built an uncertain world that could study the system.
When the Internet Still Waited
The early internet was slow, awkward and frequently ugly. A modem announced your arrival with a sequence of electronic shrieks. Images appeared line by line. Video was an event rather than a background condition. You typed an address, followed a link or searched for something you had already decided to find.
It was not an innocent paradise. There were pop-up advertisements, scams, pornography, spam, flame wars and sensational headlines almost from the beginning. Human beings brought their appetites and cruelties online as readily as their curiosity.
But the structure of the experience was different.
The early web was composed largely of destinations. You visited a website, read its pages and eventually reached the end. Forums had boundaries. Blogs were published at human intervals. Email waited in an inbox. Even chat rooms required you to enter a particular place.
The web was somewhere you went.
Its slowness imposed a form of discipline. Waiting was irritating, but it also created a moment in which intention could reassert itself. Was this page worth loading? Did you really want to download that file? Were you still interested enough to continue?
Friction is usually described as a technical defect, and much of it was. Slow connections excluded people, constrained communication and made valuable information difficult to access. Yet friction also created stopping points. The early internet could educate, connect and astonish us, but it could not occupy every unclaimed second of the day.
It gave us extraordinary libraries, archives, communities and tools. People learned obscure skills from strangers, found others who shared interests no one in their physical lives understood, published work without asking an institution for permission and communicated across distances at negligible cost. The promise of the internet was not simply more information. It was greater intellectual and creative agency.
High-speed access expanded that promise. It made open courses, video calls, collaborative software, digital art, independent publishing and vast public repositories possible. Broadband was not itself the villain. It was the removal of a constraint.
But once the constraint disappeared, a new commercial question emerged: how much of a person’s life could now be moved online?
The Internet Stops Waiting
Broadband removed the delay. Smartphones removed the location. Notifications removed the need to remember. Infinite feeds removed the end.
The internet stopped waiting to be visited and began following us.
This was the decisive transition. The late internet was no longer primarily organised around pages that people intentionally sought. It was organised around streams that arrived continuously and selected themselves. The user no longer needed to ask a question, choose a destination or even know what they wanted. The feed would decide what came next.
High-speed access made this technically possible, but the advertising model gave it direction. When a service earns money by placing advertisements in front of people, every additional minute of attention becomes potential inventory. The commercial objective is no longer merely to satisfy the user. It is to prevent the user from leaving.
That objective shaped an entire generation of design.
Autoplay eliminated the decision to begin another video. Infinite scroll eliminated the signal that there was nothing left to see. Notifications manufactured urgency outside the application. Recommendation systems replaced deliberate navigation with predicted appetite. Likes, streaks and visible counters converted social participation into a series of small, measurable rewards.
None of this was inevitable. It was designed.
The phrase “the algorithm” can make responsibility sound abstract, as though a mysterious intelligence independently decided to colonise human attention. Algorithms do not establish business models. Companies, executives, investors and product teams do. They chose what to measure, which behaviours to reward and what the system should optimise.
The late internet was built around engagement because engagement could be sold.
The Industrialisation of Interruption
The forest did not measure which rustle kept your eyes fixed for half a second longer. The predator did not test twenty faces to discover which expression frightened you most efficiently.
Digital platforms do exactly this.
Every click, pause, replay, share, search, hesitation and abandonment can become evidence. A platform observes which images interrupt you, which subjects anger you and how much uncertainty you will tolerate before losing interest. It then uses that evidence to modify what you encounter next.
The system does not need a psychological theory of you. It only needs to notice patterns in your behaviour.
This is what distinguishes the platform economy from older forms of persuasion. A billboard cannot reconstruct itself after watching your eyes. A newspaper cannot quietly produce a different front page for every reader. A television programme may target a demographic, but it does not ordinarily modify the next scene according to whether one particular viewer appeared restless during the last one.
The modern feed is simultaneously the experiment, the laboratory and the product.
Clickbait is one visible result. Its techniques survive because they outperform restraint under the measurements that determine visibility. Alarm, accusation, mystery and emotional exaggeration are not merely stylistic failures. They are adaptations to an environment in which the first click is more legible than the reader’s eventual understanding.
Research using thousands of randomised headline experiments found that negative language increased clicks even when different headlines led to the same underlying story. In the dataset, adding a negative word to a headline of average length increased its click-through rate by approximately 2.3 per cent. That advantage seems small until it is repeated across hundreds of millions of impressions. At sufficient scale, a fractional improvement becomes an editorial commandment. The study appeared in Nature Human Behaviour.
The curiosity gap works in a similar way. A headline presents enough information to create a question while withholding whatever would allow the mind to settle it. The reader is not promised knowledge so much as relief from a deliberately induced incompleteness.
The efficient headline cannot be entirely empty. If it conceals too much, the reader has no reason to care; if it reveals too much, there is no reason to click. The art lies in creating exactly enough uncertainty to activate the orienting response without resolving it. Research suggests that excessive vagueness eventually becomes less effective. The gap must be engineered with precision.
Outrage is even more valuable because it does not end with a click. It generates comments, denunciations, arguments, screenshots and quote-posts. A person may hate a piece of content and nevertheless distribute it to thousands of others. Hostility towards the material becomes labour performed on its behalf.
The platform records the activity, not the approval.
How the Machine Trains Its Audience
The feedback loop does not merely determine what we see. It influences how we behave.
People notice which of their posts attract attention and gradually learn what the environment rewards. Certainty travels further than hesitation. Accusation receives more response than qualification. Moral anger attracts allies, opponents and spectators simultaneously.
Research into online social networks has found that receiving greater social feedback for expressions of outrage can increase a user’s subsequent expressions of outrage. The internet does not simply reveal an angry public. Its incentives teach people to become more effective performers of anger. Social reinforcement helps establish the emotional norms of the platform.
The algorithm is therefore neither a malicious puppeteer nor an innocent mirror. It forms one half of a coupled system. Human beings supply impulses. Platforms measure them. Ranking systems amplify the material that produces the strongest reaction. People observe what succeeds and adapt themselves accordingly.
The machine does not invent our worst tendencies. It identifies, rewards and trains them.
That does not absolve the person who spreads cruelty or falsehood. But it does expose the absurdity of treating the attention crisis as a collection of individual moral failures. A person choosing between two headlines is making a choice, yet the available options have already been selected and arranged according to predictions about what will be hardest to resist.
Calling this “giving people what they want” is evasive. Human beings want many incompatible things. We want stimulation and tranquillity, knowledge and reassurance, novelty and coherence. We want to keep scrolling, and we want to stop. A platform that optimises the first desire while obstructing the second is not neutrally expressing preference. It is choosing which preference to cultivate.
From Information to Captivity
The consequences extend beyond irritating headlines. When publishers depend on traffic, every subject must be made urgent. Thumbnails grow more grotesque. Titles become personal, accusatory and absolute. Complex events are compressed into moral confrontations because confrontation moves faster than uncertainty.
Creators face a brutal bargain. Refuse the prevailing tactics and risk becoming invisible. Adopt them and gradually lose the ability to communicate in any other register.
What begins as packaging eventually enters the work. The article is written to justify the headline. The video is restructured around its thumbnail. Every opening must contain a hook. Every quiet interval becomes a retention risk. The work can no longer unfold according to the rhythm of its subject because the possibility of departure governs every second.
This is how an attention market alters culture without issuing explicit commands. It does not prohibit subtlety. It merely makes subtlety more difficult to finance.
Audiences change in response. Constant exposure to artificial urgency weakens our sense of proportion. If every election will end democracy, every celebrity mistake reveals a moral emergency and every minor scientific study “changes everything,” genuine importance becomes harder to recognise.
We do not become exhausted because nothing matters. We become exhausted because everything is presented as though it matters immediately.
Attention separates from intention. We open an application for one purpose and surface twenty minutes later carrying information we never chose to seek. The experience may have been stimulating, but stimulation is not satisfaction. The platform succeeded according to its measurements even when we regard the time as stolen.
This discrepancy reveals the moral failure at the centre of the late internet. Platforms can measure whether we continued, but not whether continuing served us. They can observe reaction more easily than understanding, compulsion more easily than fulfilment and agitation more easily than wisdom.
So they optimise what they can count.
The Personalised Trap
Generative AI is likely to intensify the process. Traditional clickbait is optimised across groups: several headlines are tested and the strongest general performer survives. Synthetic media can produce effectively unlimited variations at negligible cost.
The next stage will not merely select the headline most likely to attract an audience. It will construct the headline, image, voice and emotional pressure most likely to attract you.
The attention economy is moving from mass persuasion towards personalised temptation. The feed will not need to discover what shocks everyone. It will need to know what shocks one person reliably, at one particular hour, in one particular mood.
At that point, “use your willpower” becomes an increasingly unserious response.
Turning off notifications, removing applications and keeping phones outside the bedroom can help. Individual habits matter. But asking each person to defend their concentration against systems improved through continuous experimentation is like treating polluted air exclusively as a breathing-habit problem.
The crisis is structural because the capture is profitable.
Reclaiming the Right to Look Away
We should not romanticise dial-up or attempt to make the internet technically slow again. Speed has delivered extraordinary benefits, and the early web excluded many people precisely because access was expensive, difficult and unreliable.
What we need to recover is not latency but boundaries.
Feeds can have endings. Autoplay can be disabled by default. Notifications can be reserved for communication rather than manufactured urgency. Chronological views can be treated as a right rather than a hidden setting. Users can be given meaningful control over recommendation systems. Publishers can depend more on subscriptions, public funding and durable trust than on the repeated sale of momentary attention.
Regulation must also address the incentives beneath the interface. Transparency about targeting and amplification matters, but transparency alone does not neutralise a business model. A system can explain perfectly clearly that it profits from compulsive use and continue profiting from it.
The early web enlarged human agency. The late web discovered that agency could be interrupted, measured and sold.
“Attentionpocalypse” is itself the kind of title this market rewards. It compresses a complicated history into catastrophe and supplies enough emotional voltage to make indifference difficult. The title performs the mechanism the article describes.
But we are not doomed. The attention economy was built through human decisions and can be altered by them. Its present form is neither natural nor permanent.
Attention is more than a private resource. What a society notices influences what it can understand, remember and change. A culture unable to direct its attention will eventually find that its priorities have been selected by whatever is most effective at interrupting it.
Three hundred thousand years ago, the movement in the grass deserved your full attention. The reflex protected a life.
Today the grass never stops moving.
The internet was once somewhere we went. Now it is the thing that decides where our mind goes next.
