Around 2006 a designer named Aza Raskin helped create a feature most people now use every day without thinking about it: infinite scroll, a page that keeps loading more content as you near the bottom, so there is no natural stopping point at the end of a page. Years later he said publicly that he regretted how it came to be used, and he suggested it was costing people a vast amount of time every day. Infinite scroll is a good place to start because it shows the double nature of design: a real convenience that also removes the cues that once told us to stop.

Persuasive technology

In the late 1990s the Stanford researcher B. J. Fogg began studying how computers could change what people think and do, a field he called persuasive technology. His later behavior model holds that a behavior happens when three things meet at the same moment: motivation, ability and a prompt. If one is missing, the behavior does not occur. Make an action easier, or put a prompt in front of someone who is already motivated, and the action becomes more likely. Fogg's ideas have been used in exercise and savings apps, and they also influenced many people who built social media products; some of his students went on to work at major technology firms.

The hook

In his 2014 book Hooked, the writer Nir Eyal described a four-step loop shared by habit-forming products: trigger, action, variable reward and investment. A trigger may be external, like a notification, or internal, like boredom or loneliness. The action is the simplest response available, such as opening the app. The reward is variable: sometimes there is something interesting and sometimes not, which keeps people checking, the same principle that makes slot machines compelling. Finally the user invests something, such as a post, a follow or a profile, which makes the next trigger more likely and the product more valuable to that user. Over time, Eyal wrote, the goal is for internal triggers to do the work, so that users reach for the product when a feeling arises, with no outside prompt.

A catalog of features

Many familiar features can be read through these models. Notifications and red badge counts are external triggers. Pull-to-refresh, made popular by the developer Loren Brichter around 2009, works like a lever that might deliver something new. Autoplay starts the next video before the viewer decides, so continuing takes no effort while stopping does. Streaks, which count how many days in a row two friends have exchanged messages, turn a habit into an obligation that feels costly to break. Likes and read receipts draw on deep human concerns about belonging and reciprocity.

Some designs go further and work against what users intend. In 2010 the designer Harry Brignull coined the term dark patterns for interfaces that trick people into choices they did not mean to make, such as subscriptions that are easy to start and hard to cancel, or a privacy setting buried several screens deep. Regulators in Europe and the United States now treat some dark patterns as unfair or deceptive practices.

The insiders' critique and its critics

In 2013 Tristan Harris, then working at Google, shared a slide presentation with colleagues arguing that the company had a responsibility to reduce distraction and respect users' attention. He later helped found the Center for Humane Technology, which describes the industry as caught in a race for attention, in which any firm that holds back loses ground to rivals that do not. On this view the problem is structural, not a matter of a few bad actors.

Others push back. Eyal himself wrote a later book, Indistractable (2019), arguing that most distraction starts from inside, from discomfort we want to escape, and that people can learn to manage it. Researchers note that the word "addiction" is often used loosely: heavy use is not the same as clinical addiction, and most people use these products without serious harm. Designers point out that many sticky features are also helpful; nobody misses clicking "next page" twenty times. The fairest reading is that design tilts behavior without fully determining it. Features tilt the field and people still choose, though the tilt is steeper for some, including children and people who are already struggling.

Worked example: ten minutes on a video app

Sam opens a short-video app after a buzz announces that a friend has posted. That buzz is an external trigger. Opening the app is the easy action. The first two videos are dull; the third is hilarious, a variable reward. Sam likes it and follows its creator, a small investment that makes tomorrow's feed more tailored. Each video ends and the next begins with no decision required, and the feed never reaches a bottom. Ten minutes become forty. Using Fogg's model, Sam can redesign the situation. Turning off that app's notifications removes the prompt. Moving the app off the home screen lowers ability a little. Deciding on a reason before opening it, such as watching the friend's post and nothing else, gives motivation a clear target. None of this proves the app is evil. It shows that the loop is real and that its parts can be named and changed.