Six things are doing most of the work, and you can see all of them without trusting anyone's theory: no bottom to the page, a refresh that never returns the same thing twice, autoplay, red badges, streaks and counts, and a ranking system that shows you whatever held you last time. Each has a counter you can apply in under a minute. Those are at the end of each section.
No bottom to the page
Ordinary things end. A newspaper runs out. An episode finishes. The end is where you decide whether to continue, and deciding is the moment a habit can break.
The feed removes that moment. There is no page two, no last item, no natural place where a session concludes rather than gets interrupted. Sessions therefore end when something external stops them: someone speaks to you, the battery dies, you notice the time.
**The counter.** Put an ending back in by hand. A timer you can hear from another room. Getting up to fetch the phone from a shelf rather than a pocket. Using the browser version, which is worse at hiding its edges than the app is.
The refresh that is never the same twice
Pull down, wait, something new appears. Sometimes it is dull, sometimes it is exactly what you wanted, and you cannot tell which before you pull.
Unpredictable results hold attention harder than reliable ones. That is not a controversial claim about human beings and you have felt it at a bus stop, refreshing something you had already read. The design point is that the refresh is under your thumb, is free, and takes half a second.
**The counter.** Remove the fast route. Log out so the refresh costs a login. Turn off cellular data for that app so it only loads on wifi, which means it only loads in places where you had a choice about being.
Autoplay
The next video starts before you have decided to watch it. This converts a decision you would sometimes make into a decision you have to actively cancel, and cancelling is more work than drifting.
**The counter.** Turn autoplay off in the app's own settings. Most platforms still allow it, usually under playback or data saving. It is the single most effective menu change on this page and it takes twenty seconds.
Red badges and notifications
A red dot converts a glance into a session. That is its entire job: it makes an unopened thing feel like an unfinished one.
Notifications do the same from further away. Something arrives, you look, and you were not going to look. Most of them are not a person talking to you, they are a product asking for you back.
**The counter.** Badges off for everything that is not a human. Notifications on only for direct messages and calls. If you take one thing from this page and cannot face the rest, take this.
Streaks, counts and read receipts
Streaks manufacture a cost for stopping. Counts turn other people's attention into a score you can lose. Read receipts and typing indicators create obligation in what used to be a message you would answer when convenient.
None of these existed in the first version of any of these products. All of them were added, deliberately, and you can decide about each one separately.
**The counter.** Break a streak on purpose and see what happens. Turn read receipts off. Hide like counts where the platform allows it, which several now do because enough people asked.
Ranking, which is the quiet one
The feed is ordered by what has held people like you, which increasingly means whatever held you last time. It is very good at this, and it gets better the more you use it, which is why it feels personal. It is not personal. It is a measurement of your own past behaviour played back at you.
The practical consequence is that the feed you are trying to leave is worse than the feed you joined, because it has been trained on your worst evenings rather than your best ones.
**The counter.** Prune the inputs, which is far easier than beating the ranking. Unfollow every account you do not know personally. Most people's feeds empty out inside an hour and the people they actually care about are still there.
Where this argument runs out
You will read that the feed is engineered to addict you, and that you are up against a laboratory. Half of that is fair. The other half needs saying, because the despair version of this article helps nobody.
There is no recognised condition here. The American Psychiatric Association is explicit that its proposed research condition for gaming ["is limited to gaming and does not include problems with general use of the internet, online gambling, or use of social media or smartphones"](https://www.psychiatry.org/patients-families/internet-gaming), and that gambling disorder is the only behavioural addiction in the current manual.
Researchers reviewing how quickly this field has expanded warn that ["recent publications have suggested that nearly all daily life activities might lead to a genuine addiction"](https://pubmed.ncbi.nlm.nih.gov/26014667/).
And on the harm the design supposedly causes, all four UK Chief Medical Officers concluded that ["This research does not present evidence of a causal relationship between screen-based activities and mental health problems"](https://www.gov.uk/government/publications/uk-cmo-commentary-on-screen-time-and-social-media-map-of-reviews), while still advising a precautionary approach.
So: the features above are real, observable, and deliberate. The claim that they constitute an addiction machine that has taken your agency is a bigger claim than anyone has demonstrated. Both things being true at once is uncomfortable and it is where the honest ground is.
What actually beats design
Friction. This is the useful half, and it has trials behind it rather than argument.
A field experiment tested greyscale, which the researchers classify as design friction. It ["led to an immediate, significant reduction of objectively measured screen time compared with the control condition"](https://pubmed.ncbi.nlm.nih.gov/36577008/), and beat self-set time limits, which worked more slowly.
A randomised trial that blocked mobile internet for two weeks, while leaving texts, calls and desktop internet working, found it ["improved mental health, subjective well-being, and objectively measured ability to sustain attention; 91% of participants improved on at least one of these outcomes"](https://pubmed.ncbi.nlm.nih.gov/39967678/).
Notice what both interventions have in common. Neither asked anyone to want it less. They changed what was easy. That is the same move as every counter on this page, and it is the reason design is beatable by design rather than by character.
Honesty about the limits, from the same greyscale study: "we found no immediate causal effect of reducing usage on subjective well-being and academic performance." You get hours back. You do not get a different mood as part of the deal.
The three to do tonight
If you do nothing else: autoplay off, badges off for everything that is not a person, and unfollow every account you do not know. Ten minutes, and it changes what the product is allowed to do to your evening.
Then, if you want the version with the sleep evidence attached, [three changes that reduce pickups](/guides/social_media/how-to-stop-doomscrolling) is the practical piece. And because a quieter feed leaves a gap, [have something ready for the moment your hand reaches](/guides/social_media/what-to-do-with-your-hands-instead-of-scrolling).
None of this requires you to win an argument with an industry. It requires six settings and a shorter following list.
Common questions
**Is social media designed to be addictive?** It is designed to hold attention, and the features that do it are visible in the product. Whether that amounts to addiction is a different question, and there is no recognised diagnosis for social media use anywhere. You can accept the first without needing the second.
**Why is TikTok harder to put down than other apps?** Short items, autoplay and an unpredictable next thing mean the decision point that ends most sessions never arrives. The counters are the same as everywhere else: turn autoplay off, remove the badge, and put a timer somewhere you can hear it.
**If it is engineered, is it even my fault?** Fault is the wrong frame and it is also a trap, because it leads to either shame or despair and neither changes an evening. The features are deliberate and the counters are available to you. That is enough to act on without settling the blame.
**Does knowing how it works help me stop?** On its own, not much. People who understand the design get held by it too. What moves the numbers in trials is friction: greyscale, blocking, a longer route to the app. Use the explanation to pick your friction, not instead of it.
Sources
- [American Psychiatric Association, Internet Gaming](https://www.psychiatry.org/patients-families/internet-gaming). The proposed gaming condition excluding social media and smartphone use, and gambling disorder being the only behavioural addiction in DSM-5-TR.
- [UK Chief Medical Officers' commentary on screen-based activities and children and young people's mental health](https://www.gov.uk/government/publications/uk-cmo-commentary-on-screen-time-and-social-media-map-of-reviews). No evidence of a causal relationship, and the precautionary approach.
- [Digital Strategies for Screen Time Reduction: A Randomized Field Experiment, Cyberpsychology, Behavior, and Social Networking 2023](https://pubmed.ncbi.nlm.nih.gov/36577008/). Greyscale as design friction reducing measured screen time, and the null result on wellbeing and academic performance.
- [Blocking mobile internet on smartphones improves sustained attention, mental health, and subjective well-being, PNAS Nexus 2025](https://pubmed.ncbi.nlm.nih.gov/39967678/). The two-week block, the 91% figure, and the design that kept texts and calls working.
