Atlas of Unfulfilled Warnings: Onboarding the Future

User eXperience design began as a way to make software easier to use. Over time, it became a method for shaping behaviour at scale. Through experimentation, data collection, and psychological insight, UX learned how to guide attention, reduce resistance, and encourage participation among millions of people at once. All it needed was iteration to prove how effective it could be.

It has been so effective that entire industries now exist to measure how users respond, where they pause, and what persuades them to continue. Interfaces are refined to keep people engaged, to smooth decision-making, to remove friction where hesitation might occur. This practice influences people all over the world to make decisions.

But, what happens when these same techniques leave the screen?

As digital systems begin to govern more of everyday life the logic of UX does not disappear. It follows. The research that once optimised clicks can just as easily optimise compliance. The methods used to guide users through platforms can guide citizens through systems. When interfaces become infrastructural, design choices acquire political weight without ever announcing themselves as power.

This essay examines how onboarding becomes a mechanism for normalising participation in systems that extend beyond the digital. Without force, and with great ease. Not through commands, but through invitation. The question to ask ourselves is, what kind of future do they teach us to accept?

I. The First Experiment: Before the Screen

Long before the first login form flickered to life, the world had already learned how to test us. The science of persuasion did not begin with code, but with colour, placement, and tone. Shop windows became laboratories of attention. Billboards learned to borrow the language of desire. The earliest advertisers were not artists but experimenters, recording which words stirred hunger, which faces lingered in memory, which shades of red could make an appetite quicken.

In 1928, Edward Bernays wrote that the conscious and intelligent manipulation of habits and opinions was “an important element in democratic society.”^[1] He called it the engineering of consent. Behind the phrase lay something new: a belief that the public mind was measurable, steerable, and, given the right sequence of cues, predictable.

The conscious and intelligent manipulation of the organised habits and opinions of the masses is an important element in democratic society. Those who manipulate this unseen mechanism of society constitute an invisible government which is the true ruling power of our country.

After the World Wars, persuasion became industrial. Governments had used posters and film reels to summon duty and hatred, to shape the chaos of emotion into collective will. When peace returned, the same techniques found a gentler market. Buy, not fight. Belong, not obey. The apparatus of propaganda dissolved into the language of commerce. Psychologists and copywriters collaborated on the same experiments. In the aisles of supermarkets, in the timing of jingles, in the quiet warmth of brand colours, data began to whisper through design.

By the middle of the century, consumer research had become a quiet discipline of power. The focus group replaced the rally. Observers watched through mirrored glass as volunteers discussed detergents and dreams. Their gestures were recorded, their pauses analysed. It was not coercion; it was calibration. The public was learning to be measurable. And the companies were learning that with enough measurement, control no longer needed force, it could flow through the invitation of choice.

Retail followed suit. Aisles stacked by order of natural eye movement. Music softened the mind. Lighting thickened in the corners where profit hid. It was experience design without the name: physical interfaces built to steer movement and feeling. When the web arrived, it did not invent manipulation; it inherited it. The browser was simply a new kind of store with infinite aisle length. Every click could be observed. Every hesitation became a signal. What had been studied in rooms and markets now occurred in real time.

II. The Age of Analytics: When Design Began to Listen

The earliest web pages were flat documents written in code that barely concealed its construction. Forms asked for information, buttons performed a single act, and each site felt honest. There was little to seduce, little to measure. The web was a library of pages. It was still built by generalists: a single mind could write the code, shape the structure. Function mattered more than feeling.

Once visits could be counted, once clicks could be traced, the web began to wonder what else could be learned. Amazon was among the first to realise that design could be treated as data. In 2006, an internal experiment^[2] tested a simple intuition: what if the shopper were offered recommended items at checkout? The feature had been considered intrusive and was initially forbidden. Yet the results overturned instinct. Testing revealed that its absence was costing the company a remarkable share of revenue. The experiment proved that the smallest change in sequence could alter the flow of commerce itself.

HTML evolved into CSS; structure acquired style. Soon style demanded interactivity, JavaScript, animation, and responsiveness. The web was no longer a static canvas. Front-end developers emerged, their task no longer merely to display information but to shape experience; layout became psychology and spacing became strategy. The same consumer research marketing techniques applied in real-time.

The rise of analytics made the interface recursive. Tools like Google Analytics, Optimizely, Mixpanel were instruments of curiosity and calibration. Every interaction could be captured, compared, replayed. Each visitor became a data point. The designer’s work no longer ended at launch; it began there. Code was written to be watched. The website turned into an organism that adjusted itself by watching its own pulse.

A/B testing, once a method of market research, became the architecture of digital life. Two versions of a page, two tones of a message, two pathways to the same decision. The better performer survived. The weak were deleted; the persuasive endured. It was Darwin by design. What had once been an art of communication became a science of conversion.

By the end of that first digital decade, the web had become a listening architecture, an environment where every movement was measured, every preference predicted, and every hesitation turned into insight. What began as a tool for access had become an engine of adaptation. The interface had learned to hear desire, and to answer before we even spoke.

III. Empirical Evidence: Measurable Behavioural Effects

Every company, from the smallest subscription service to the largest platform, was quietly conducting experiments. The results were rarely published, but the patterns could be seen in the traces left behind. The surface of the Internet became a living record of its own behavioural science.

Some of these tests escaped into view. In a 2021 study from Princeton University^[3] researchers examined more than eleven thousand shopping sites and found that most contained some form of what they called a dark pattern (interfaces built to steer rather than to serve). Countdown timers, fake scarcity, pre-ticked boxes, guilt-tripped opt-outs: all silent cues designed to hasten agreement.

At the same time, controlled experiments showed just how measurable that shape could be. In one^[4] users exposed to “mild” manipulative designs were more than twice as likely to subscribe to a service; with “aggressive” versions, nearly four times as likely. The price of the product mattered less than the choreography of the page.

Regulators began to notice. France’s data authority^[5], the CNIL, found that the layout of a cookie banner could swing acceptance rates by vast margins. A bright “Accept All” button, easy to see and quick to press, captured far more consent than a muted “Refuse.” The difference was only design but its consequences reached into every file of personal data collected afterward. When governments started to fine companies for deceptive interfaces, they were acknowledging, perhaps for the first time, that design itself had become a form of governance.

The same logic operated in subtler places. In 2014, Facebook conducted a massive experiment across nearly seven hundred thousand users, quietly adjusting the emotional tone of their news feeds^[6]. When more positive posts appeared, people wrote in happier language; when the content skewed negative, their words darkened in turn. A few years earlier, a small “I Voted” message shown to sixty-one million Americans measurably increased voter turnout. Interface: No threats, no propaganda.

Beyond the headlines, the pattern held. Regulators forced Booking.com^[7] to remove false urgency messages, fining them for implying that rooms were nearly sold out when they were not. The U.S. Federal Trade Commission accused Amazon of designing a cancellation process so labyrinthine^[8] it effectively trapped users in Prime memberships, and later fined Google for concealing location tracking consent^[9] behind misleading flows. Each case revealed the same lesson: that what users saw, how they moved through a page, and what choices appeared easiest were not aesthetic questions but economic ones. The interface was a machine for extracting assent.

Taken together, these examples form a portrait of an invisible discipline where persuasion has been absorbed into infrastructure. The web’s most ordinary gestures, the casual click or the quick tap of a finger, are now sites of continuous testing. The results are seldom public, but the outcomes surround us: consent gathered, moods adjusted, purchases completed, loyalty earned.

IV. The Social Mirror: When Data Learns to Feel

Once platforms learned how to trace behaviour, they turned their attention to emotion. The invisible motivation. The logic was simple: the longer people lingered, the more they revealed. What they revealed could be catalogued, predicted, and, with time, induced. The social web, born as a promise of connection, evolved into a data lake that reflects the world’s feelings, with the ability to manipulate the experience.

In the early years, these systems seemed harmless. Their offer of a new vocabulary, likes, follows, and shares, were a small pulse of recognition. Every pulse was a data point, and together they formed an emotional map of humanity. The platforms learned that outrage travelled faster than affection^[10], that sorrow was sticky^[11], and that affirmation kept the gaze steady^[12]. Algorithms began to select for what sustained attention, not for what nurtured it.

What propaganda had once achieved with posters and radio, algorithms could now accomplish invisibly, in real time, across continents.

The following years revealed that the experiment was not an anomaly; it was the business model. Twitter discovered that conflict drove engagement^[13], and engagement sold advertising. YouTube’s recommendation engine quietly learned^[14] that users who watched sensational or extreme content stayed longer, feeding a slow pull toward radicalisation, that the company later acknowledged but could not easily unwind.

This ability to steer collective feeling soon escaped the laboratory. The feeds of billions became a stage, each platform learning what blend of emotion kept its audience returning. During elections^[15] outrage was currency. During pandemics, anxiety sustained the scroll^[16]. Each refinement of the algorithm deepened its sensitivity until emotion itself became both input and product. When the data learned to feel, it also learned to sell.

The Cambridge Analytica^[17] scandal was perhaps the first time the public saw the mirror crack. It revealed that the emotional architectures of social media could also be used to segment voters, to micro-target fear. Yet even after the outrage, the infrastructure remained. The experiment had become the medium.

V. When the Interface Stops Being Digital

Systems are already mature by the time regulation is a concern. Data protection authorities issue fines, mandate disclosures, and require consent mechanisms, but none of this interrupts the underlying process. The collection continues onward. The interfaces remain. The behavioural models improve. Regulation formalises accumulation of data, marking the system as real, but it does not stop the system from running.

Much of this regulation focuses on consent. Users are informed. Notices are displayed. Checkboxes are added. Agreement is made visible, documented, and repeatable. Yet, consent is no longer a moment of decision so much as a gesture of passage; a light agreement to documents few read, presented at moments when refusal would interrupt access, delay progress, or exclude participation.

This is already a familiar online pattern. Platforms learned early on that compliance could be achieved through design. Lengthy and legal documents are placed behind a single checkbox saying “I agree”. Defaults are set in advance. Progress through a system depends on agreement, and agreement is framed as continuity not as a choice.

It does not matter that individuals understand every implication, but that our behaviour becomes predictable. Each interaction produces data. Each agreement extends the record. Profiles thicken. Patterns stabilise. The system learns what people do, how they move through decisions when guided gently, consistently, and at scale.

These techniques did not remain online.

Tools that were once governed by physical presence and institutional discretion, like identity verification, access to services, and movement through public spaces, are being integrated into digital systems. The logic of being online has followed. Interfaces replace counters. Accounts replace credentials. Participation is managed through portals, apps, and automated checkpoints that borrow directly from the design principles refined online.

The same research that optimised user flows now optimises civic interaction. The same emphasis on friction removal now shapes how people encounter authority. What once guided users through platforms begins to guide citizens through systems. The transition is not marked by a declaration or a break. It is marked by familiarity.

In this context, consent becomes ambient. It is no longer asked once, but continuously reaffirmed through use. Each interaction implies agreement with a system whose scope expands quietly over time. Data collected for one purpose becomes relevant to another. What began as convenience becomes infrastructure. What was once optional becomes expected.

VI. The Future That Learned From Us

The internet was the perfect test bed to examine behaviour. Online events of human interactions are hidden behind the scrolls and clicks, recorded real time, and examined at scale. Millions of interactions, observed and refined in real time, teach designers how people respond when choices are framed and when defaults are set.

This kind of information, this kind of possible manipulation, will not remain in the domain the internet. The techniques that have been use to shape attention, reduce resistance and normalise participation will not be bound to screens alone. All this learned knowledge will not vanish when the the digital world gives feedback to human actions, it will become more consequential.

As digital identity, surveillance infrastructure, and automated decision-making expand into everyday life, these learned lessons will be applied by systems that manage them. Everyday life is ripe for manipulations by numbers and algorithms beyond human decisions. Movement through public systems will begin to looks like movement through online platforms. Access is mediated by interfaces. Participation is managed through accounts. Compliance is shaped through flow and mass acceptance.

This is a pattern already visible.

Where behaviour can be influenced, it will be. Where data can be accumulated, it will be. And where systems can guide outcomes without ever appearing to command them, that will be put into place.

The success of UX was never that it persuaded people to act against their will. It was that it taught systems how to make certain actions feel like the natural continuation of personal decisions. And, as this process is implemented into our public systems, and data is accumulated behind the scenes, the effects will feel more and more familiar.