decaf

Decaf, for

social media.

A browser extension that strips the addictive mechanics from the platforms you keep open all day. The good stays. The slot machine leaves.

Firefox · Chrome · iOS beta · Android beta
Drained battery surrounded by anxious squiggles
Before drained · anxious · 17%
Full battery sprouting green leaves
After regrown · calm · kept
01, The problem

Your apps are designed

Short-form feeds, autoplay queues, infinite scroll, these aren’t features. They’re retention mechanics. The point isn’t the content. The point is that you don’t close the tab.

2:21 / day
Average daily time on social media, globally. That’s over 35 full days a year. (DataReportal, 2025)
205×
Times the average American checks their phone per day, once every 5 waking minutes. (Reviews.org, 2024)
81%
Of users check their phone within 10 minutes of waking up. Before they’re even out of bed. (Reviews.org, 2024)
02, How it works
i.

Full removal

Shorts shelves, autoplay sidebars, recommendation rails, deleted from the DOM, not hidden with CSS. They never render.

YouTube
ii.

Cooldown timer

Browse a set number of items, then a mandatory pause. You set the count and the wait. The overlay covers the feed; the timer counts down.

Instagram · TikTok · X · Reddit
iii.

Theater mode

Every video opens full-width. Combined with the nuked sidebar, you get a clean focused viewer, closer to a film than a feed.

YouTube
iv.

Anti-tamper

Opt in once. The overlay re-injects itself if you try to delete it via DevTools. You signed up for this. We hold the line.

All platforms · optional
03, Before & after

Drag to compare.
This is YouTube, decaffeinated.

Move the divider. Left is what YouTube ships. Right is what your browser actually renders with Decaf installed.

SHORTS
Sidebar removed · Shorts removed · Theater forced
Before
After
04, The science

This isn’t a vibe.
It’s a slot machine.

The mechanics that keep you scrolling aren’t a coincidence and they aren’t your lack of willpower. They’re the same intermittent-reward systems that B.F. Skinner used on pigeons in the 1950s, ported to your phone and tuned by teams of behavioral scientists. Decaf removes the levers. Here’s what the research says.

i.

Variable rewards hijack dopamine

Each scroll triggers a small dopamine release, not when you find something good, but in anticipation of finding something good. Unpredictable rewards produce stronger dopamine signals than predictable ones, the same mechanism that makes slot machines compulsive. Infinite scroll, autoplay, and pull-to-refresh are textbook variable-ratio reinforcement.

Sharpe & Spooner, 2025 · Clark & Zack, 2023
ii.

Attention spans collapsed

Dr. Gloria Mark at UC Irvine has been measuring how long people stay on a single screen since 2004 using objective computer logging. The number then was 2½ minutes. The number now is 47 seconds. Half the time, it’s shorter than 40. The same research links faster attention switching to higher measured stress on heart-rate monitors.

Mark, UC Irvine · 20+ years of data
iii.

The dose makes the poison

A 2022 meta-analysis of 55,340 adolescents found a clean linear dose–response, every additional hour of social media raised depression risk by 13%. The U.S. Surgeon General reports that teens crossing 3 hours a day double their risk of depression and anxiety. The current average for U.S. teens is 3.5 hours.

Liu et al., 2022 · HHS Surgeon General, 2023
iv.

Short-form is its own category

Brain imaging studies of heavy short-form video users (TikTok, Reels, Shorts) show reduced activation in the dorsolateral prefrontal cortex and anterior cingulate, the regions you use for executive control and impulse regulation. They’re also less sensitive to losses and make faster, more impulsive decisions. The 6-second loop isn’t neutral.

NeuroImage, 2025 · Frontiers in Human Neuroscience, 2025
47 sec
Average attention span on a single screen, down from 2½ minutes in 2004.
205×
The average American checked their phone 205 times a day in 2024, once every 5 waking minutes.
+13%
Increase in adolescent depression risk per additional hour of daily social media use.

None of this is news to the platforms. They have the same data, plus a great deal more. The slot machine is the product. Decaf takes it out.

05, Keeping your word

Try to
remove it.

When you start a cooldown, you mean it. Decaf takes you at your word and won’t let you wriggle out, even if future-you really wants to. The pause stays the pause. The promise sticks.

  1. i. The lock is off by default. You turn it on once, when you’re sure you mean it.
  2. ii. While the timer is running, the pause screen won’t close. Refreshing the page doesn’t end it. Closing the tab doesn’t end it.
  3. iii. The clever workarounds people use on other “block this site” tools don’t work here. Decaf is built to keep your word for you.
Try to remove it.
click the button →
06, FAQ

Frequently asked.
Honestly answered.

On desktop and laptop, yes, completely free, no accounts, no ads, no upsells. Mobile (iOS and Android) is a paid subscription when it ships, that’s what keeps the lights on so the desktop version can stay free forever.
Screen-time apps wait until you’ve already spent the time, then tell you about it. Decaf takes the addictive parts out of the page before you ever see them, so there’s less to resist in the first place.
Yes. You decide what changes. Some platforms can be cleaned up entirely, others can have a short cooldown after a few posts, and you can leave any of them alone if you want.
A mobile version is in private beta. Join the list and we’ll send you an invite when we open it up. Mobile will be a paid subscription, you can try it free for the first two weeks.
You might. That’s why there’s an optional lock, once you turn it on, the cooldown sticks until the timer runs out. Think of it as a promise you make to yourself when you’re thinking clearly, that gets honored when you’re not.
No. Pages load just as fast, often faster, since there’s less to render.

Ready to decaffeinate?

Free on desktop and laptop. Mobile coming soon, by subscription. No account required to start.