Capture a minute of your feed. We read it on your device and show you the shape of what you're being served — what it's training you to be, feel & buy, and how far it's drifted from the middle.
📵 The platforms know exactly what they funnel you — it's what keeps you scrolling. They just never show you. This does.
Two easy ways: screenshot your feed 15–20 times as you scroll, or screen-record it for a minute or so. The more you capture, the sharper the read — a quick peek only shows a couple of directions.
Nothing about you is stored — ever.
Select them all at once — 15+ is ideal (more = sharper). Or tap to choose.
🔒 nothing about you is storedDeep is the default — an AI (Amazon Nova) reads the text for the sharpest result, then discards it (nothing is stored). Prefer to keep everything on your device? Pick Quick.
Your feed has a shape you can't see from the inside. doomscrollstats.com holds up a mirror: it reads the posts an algorithm serves you and shows what it's reinforcing — what it's quietly trying to make you be, feel, and buy.
It's free, a mirror not an alarm, and non-partisan by design — a hard-left and a hard-right bubble score the same. Your screenshots are read on your device (or, for the sharper read, by a cheap model on the server); nothing about you is stored.
Built because algorithms quietly shape what millions of us think, and almost no one gets to watch it happen. Now works on X, Instagram, TikTok and YouTube.
1. You capture your feed. Screenshots or a short screen-recording of your For You / home feed. Text feeds (X, YouTube, LinkedIn, Reddit, Facebook) are read on your device; image feeds (Instagram, TikTok) are sent to a model and immediately discarded.
2. It reads the posts. On-device: Tesseract.js (OCR) plus a small in-browser model. Sharper reads and image feeds use Amazon Nova (Micro for text, Pro for vision) on a server that stores nothing.
3. It scores & summarises. Each post gets a political lean (−1 to +1), emotional heat (0–1), a topic, and a rough "bot / low-effort" estimate. From those it writes the read: what your feed is, what it's trying to make you be/feel/buy, your blind spots, and which accounts shape it.
Extremity is measured symmetrically — a hard-left and a hard-right feed of equal intensity get equal-magnitude scores. The tool takes no side; it measures the shape of your feed, not whether you're right or wrong.
Classifying short posts and images is inherently uncertain. Account lean is solid for well-known accounts and a rough guess for small ones. The "% bot" figure and news-outlet positions are estimates, not facts. Treat every score as a conversation starter, not a verdict.
No account, no tracking cookies. Text feeds never leave your device. Image feeds are read once and discarded — nothing about you is stored. The code is open source.
A live, anonymous snapshot of what the algorithms are serving everyone who's checked. Only counts — never anyone's actual feed.
A calm way to see what their feed is really serving them — a conversation starter, not an inspection.
Run it on your own feed first, together. Seeing your own bubble before theirs disarms the whole thing — and it's genuinely surprising. It never says anyone's been "radicalised"; it just shows the shape of a feed.
Copy a friendly, low-pressure message to text them — curiosity works far better than a lecture.
Let the result start the conversation — not win an argument.
See how your bubble stacks up against a friend's. Non-partisan — it's about who's more boxed-in, not who's right.
Every social feed is shaped by an algorithm you can't see. doomscrollstats.com is a free tool that reads your feed — on X (Twitter), Instagram, TikTok, YouTube, Reddit, LinkedIn or Facebook — and shows you the shape of your filter bubble: what the algorithm is quietly training you to be, feel and buy, how much outrage and rage-bait it runs on, which accounts and viewpoints dominate, the perspectives it never shows you, and how much is AI-generated slop. It's non-partisan by design, reads your feed on your device (or via an AI that stores nothing), and is fully open-source.
Post it back where you found your feed — help a friend see their own bubble 🌍