How to stop doomscrolling

Most advice on this subject is a list of tips. This one has a different shape, because there is now trial evidence, and the evidence points somewhere the tips do not.

What doomscrolling actually is

The word arrived properly in 2020 and got a measurement instrument in 2022, when Satici and colleagues published the Doomscrolling Scale in Applied Research in Quality of Life. They define the behaviour as habitual, immersive scanning for timely negative information on social media newsfeeds — the important words being habitual and immersive.

That is what separates it from reading the news. Reading the news ends. Doomscrolling is a session that does not have a natural end, about material selected for how badly it lands.

Their data, across three studies with a total of roughly 1,250 participants, found doomscrolling correlated positively with neuroticism, fear of missing out and social media addiction, and negatively with conscientiousness, extraversion and agreeableness. The third study found that psychological distress mediated the relationship between doomscrolling and wellbeing — meaning the scrolling does not make you unhappy directly so much as it reliably raises distress, and the distress does the rest.

Why it is genuinely hard to stop

Three mechanisms stack, and they are not the same mechanism.

Negativity bias. Attention is preferentially allocated to threat. This is old, deep and not a character flaw; it is the reason your ancestors noticed the rustle in the grass. A feed ranked by engagement discovers this about you within a few sessions.

Variable-ratio reinforcement. Most of what you scroll past is unremarkable. Occasionally something is genuinely alarming, genuinely funny, or genuinely worth knowing. Rewards delivered on an unpredictable schedule produce the most persistent behaviour of any reinforcement pattern in the literature — it is the slot-machine result, and it is the single most-copied finding in consumer software.

Intolerance of uncertainty. This one explains the specific doom part. Scrolling bad news feels like information-gathering: if I understand the situation, I will feel less unsettled by it. It does not work, because the feed regenerates uncertainty faster than you can resolve it. Türk-Kurtça and Kocatürk, writing in Personality and Individual Differences in 2025, found that intolerance of uncertainty helps explain the link between trait anxiety and doomscrolling.

Put together: a system tuned to threat, rewarded unpredictably, chasing a resolution the medium is structurally unable to deliver. That is not a willpower deficit. It is a well-designed trap, and calling it a personal failing is both unkind and analytically wrong.

What the trial evidence shows

In February 2025, Pieh and colleagues published a randomised controlled trial in BMC Medicine. 111 healthy students, averaging 276 minutes of screen time a day, were randomly assigned to either cut smartphone use to two hours a day for three weeks, or continue as normal.

The intervention group improved measurably. Effect sizes were small to medium: depressive symptoms (η² = .109), stress (η² = .085), wellbeing (η² = .053), sleep quality (η² = .048). The authors argue this supports a causal relationship rather than a correlation, which matters, because most of the literature in this area is observational.

So: cutting screen time helps. Modestly, measurably, causally.

Now the part that rarely gets quoted. From the same abstract:

Screen time increased rapidly after the intervention and at follow-up the values were once again approaching the initial level.

Three weeks of successful reduction, and six weeks later they were nearly back where they started. Not because the participants were weak — they had just demonstrated they could do it — but because the thing that had been holding the behaviour in place was the study, and the study ended.

This is the most useful finding in the whole area, and it reframes the question. The problem was never can you cut down. Almost everyone can cut down for three weeks. The problem is what remains in place on week seven when nothing is watching.

What tends not to survive week three

None of these are useless. All of them share a failure mode: they depend on the state you are in at the moment they are tested.

• Awareness and tracking. Knowing your average is four and a half hours is a fact, and facts stop being interesting the second time you read them. Weekly reports are read attentively for about two weeks.
• Grayscale, notification pruning, home-screen rearrangement. Real friction, genuinely worth doing, and adapted to within days. You will find the app by searching for it, and then searching for it becomes the habit.
• Deleting the app. Works completely while it lasts. Reinstalling takes forty seconds and involves no decision you would recognise as one.
• Time limits with a free dismiss button. Covered at length in screen time limits not working — a limit that can be waved away without a passcode is a notification, and it arrives at exactly the moment you are least receptive to notifications.
• Resolving to do better. The resolution is made by a calm, rested version of you, and it is executed by a tired one at 1am. Those two people have different preferences, and only one of them is holding the phone.

The pattern: each of these gets cheaper to defeat exactly as the pressure to defeat it rises. That is backwards, and it is why the effect of almost every intervention in this category decays.

What tends to hold

Roughly in order of how much they buy you, and honestly labelled.

1. Make the escape cost something that does not fall at night. Seconds and taps are cheap at midnight. A fixed cost is not. This is the whole idea behind commitment devices, and it is the only category of intervention that gets stronger as your resistance gets weaker, because the price does not care how tired you are.

2. Give the session an ending that comes from outside you. The core design problem with a feed is that it has no bottom, so the session ends when you decide it ends — a decision you must make while the feed is still working. Anything that supplies an external end point does real work: a hard limit with teeth, a person expecting you, a room you leave the phone in.

3. Put a real block between you and the app, not a reminder. A block you must authenticate past beats a screen you can dismiss, and both beat a notification. If you use iOS Screen Time, set the passcode and turn on Block at End of Limit; most people who think limits do not work have simply never found that switch.

4. Handle the bedroom separately. Charging the phone in another room is the single most reliable structural change most people can make, and it works because it is not a decision you re-make at 1am — it was made at 10pm, by someone in a better state, and reversing it requires standing up.

5. Replace the slot, not just the app. The scroll occupies a specific slot: low-effort, infinitely available, no commitment. Deleting it leaves the slot open, and something will fill it. Deciding in advance what fills it is unglamorous and it works better than resolve.

6. Expect the reversion, and plan for it rather than treating it as failure. The trial says the return to baseline is the normal course of events, not evidence that you are uniquely bad at this. Anything you build should assume week seven exists.

The 1am problem, stated plainly

Every intervention above is trying to solve the same asymmetry. Your intentions are formed when you are calm. They are tested when you are not. And almost every tool in this category becomes easiest to defeat at precisely the moment your ability to resist defeating it is lowest.

Willpower is a depleting resource. A price is not. That is the entire argument for the thing we build.

Last Ferry gives the apps that eat your day one shared daily budget. When it runs out, they lock. You can still get past the lock — but it costs real money from a balance you topped up in advance: $0.50 for the first unlock of the day, $1.00 for the second, $2.50 after that. The ladder resets at midnight and never resets because you paid.

Most days you close the app and keep the money. That is the app working — and the reason we track the percentage of users who spend nothing as a success metric, on the same dashboard as revenue.

It is not a cure, it does not cover browsers, and no software survives a determined uninstall. What it does is make the cheapest moment to give in cost the same as the most expensive one.

Sources

• Satici, S. A., Gocet Tekin, E., Deniz, M. E., & Satici, B. (2022). Doomscrolling Scale: its Association with Personality Traits, Psychological Distress, Social Media Use, and Wellbeing. Applied Research in Quality of Life.
• Pieh, C., Humer, E., Hoenigl, A., Schwab, J., Mayerhofer, D., Dale, R., & Haider, K. (2025). Smartphone screen time reduction improves mental health: a randomized controlled trial. BMC Medicine, 23(1), 107.
• Türk-Kurtça, T., & Kocatürk, M. (2025). Beyond the Scroll: Exploring How Intolerance of Uncertainty and Psychological Resilience Explain the Association Between Trait Anxiety and Doomscrolling. Personality and Individual Differences, 235, 112919.