Two things are true at once and most writing on this subject picks one. The feed really is engineered to be difficult to put down, by people who have said so publicly. And the evidence that it is doing serious measurable harm to the average person is much weaker than the headlines suggest. Holding both is the only honest position, and it changes what you should do about it.
In 1954 a rat pressed a lever seven thousand times an hour and stopped eating to keep pressing
James Olds and Peter Milner, working at McGill University in 1954, implanted an electrode slightly off-target in a rat's brain. Given a lever that stimulated that spot, the rat pressed it again and again — up to seven thousand times an hour. It stopped eating. It stopped drinking. It crossed an electrified floor to keep pressing.
— from ENTRANCED, chapter: The Rat
The distinction in that quote is the one everything else rests on. Wanting and liking run on different circuitry. A system can drive you to pursue something relentlessly while delivering very little satisfaction when you get it, and from the inside that feels like appetite rather than compulsion.
Which is why the honest description of eleven o'clock at night is not enjoyment. It is almost. The next one might be the funny one, the useful one, the one that settles it. It is not, and that is exactly the reason the scrolling continues.
Infinite scroll was built in a single weekend in 2006, and the man who built it has apologised for it
Aza Raskin wrote the code for infinite scrolling in 2006. It took him a weekend. It is now on essentially every content platform in the world.
He has spent years since saying publicly that he regrets it, and has estimated the feature costs humanity in the region of 200,000 hours of attention every day. Whatever you make of the arithmetic, the admission is unusual and worth taking seriously: the person who built the thing is telling you what it does.
What he removed was the stopping cue. A page that ends makes you decide whether to continue. A page that never ends never asks. The decision to keep going gets made by default, which means it does not get made at all, and a behaviour that requires no decision is very hard to interrupt with a decision.
Pull-to-refresh added a lever you operate with your thumb. Notifications inverted the relationship so the machine initiates. None of these are accidents and none of them are secret — they are documented design patterns with names, taught and copied deliberately.
The mechanism is unpredictability rather than reward, which is why more content never settles it
Slot machines do not hold people by paying out. They hold people by paying out sometimes, on a schedule nobody can learn. In behavioural psychology this is called a variable-ratio reinforcement schedule, and it produces the highest and most persistent rates of response of any arrangement ever tested.
A feed runs the same schedule. Most of it is nothing. Occasionally something lands. The ratio is unlearnable, so the behaviour never extinguishes — there is no point at which your brain concludes that this is not working, because intermittently it does.
This also explains the specific frustration of it. A reward you can predict produces satisfaction and stopping. A reward you cannot predict produces searching and continuing. You are not failing to enjoy the feed. The feed is not built to be enjoyed; it is built to be continued.
The strongest objection comes from researchers who found the harm is far smaller than the headlines claim
Here is the part most pages on this subject leave out, including the earlier version of this one.
Amy Orben and Andrew Przybylski published a study in Psychological Science in 2019 using time-use diaries rather than the self-reported estimates most research relies on. They found the association between digital technology use and adolescent wellbeing was negative but extremely small — on the order of 0.1% of the variation in wellbeing. Their broader argument is that this field has a habit of overinterpreting tiny effect sizes and publishing them as alarms.
They are not cranks and they are not funded by anyone with an interest. They are among the most methodologically careful people working on this question, and their finding has held up better than a good deal of what it contradicts.
It is also worth saying that "social media addiction" is not a recognised clinical diagnosis. The World Health Organisation's ICD-11 includes gaming disorder; it does not include a general media or internet addiction. Anyone telling you otherwise is reaching.
Compulsive and harmful are two different claims, and running them together is why this never settles
The argument goes in circles because both sides are answering different questions.
The design critics are making a claim about mechanism: the feed is engineered to be hard to stop. That claim is well supported, partly because the engineers say so.
The sceptical researchers are making a claim about outcome: at a population level, screen use explains very little of who is unhappy. That claim is also well supported.
Both can be true. A thing can be genuinely difficult to put down and still not be the reason your life is hard. Recognising that is not a weakening of the case — it is what makes the case usable. If you believe screens are destroying you, every hour spent on one becomes evidence of personal failure, which is miserable and produces nothing. If you believe the feed is well designed and mostly a waste of your evening, you can address it as a design problem, which is a tractable thing to do on a Tuesday.
The one place the population-level finding does not help you is if you are the individual it does not describe. Averages conceal tails. If you have repeatedly decided to stop and repeatedly not stopped, the fact that most people are fine is not information about you.
The changes that hold are the ones that alter the environment rather than the person
Decisions made in the moment lose to a system designed by professionals to win in the moment. Decisions made in advance, which remove the moment entirely, do considerably better.
In practice that means moving the phone rather than resolving to ignore it, and removing apps rather than intending to use them less. The de-cluttering guide covers the specifics, and the free settings on your phone do most of it for nothing.
The second half is less obvious and matters more. The seeking system does not switch off because you deleted something; it goes looking for another lever. Anyone who has quit one app and quietly replaced it with another within a fortnight has met this. Something has to occupy the drive, which is the whole subject of what to do instead.
The useful question is not only how many hours you spend. It is also whether you can stop when you decide to. The twelve-question quiz helps you look at that, and the free course gives you a practical next step.
Sources: Olds & Milner, McGill University, 1954 · Aza Raskin on inventing and regretting infinite scroll, 2006 onwards · Amy Orben and Andrew K. Przybylski, ‘Screens, Teens, and Psychological Well-Being: Evidence From Three Time-Use-Diary Studies’, Psychological Science, 2019 · World Health Organisation, ICD-11.
Michael has spent much of his working life online—and much of the rest of it trying to stop that technology taking over. He has thrown a phone into a lake, locked devices in boxes and experimented with long periods offline. He writes about what helped, what failed and how to keep the useful parts of digital life.