Which of your posts broke out?
Every post a pane of glass. The ones that got far more likes than the posts around them keep their colour.
Example: Experimental History, as of 25 Oct 2026
Each colour is a year, the oldest at the neck. Tap a pane to see its post. Point at a pane to see its post.
370,000 words. A pane is as big as its post is long; each colour is a year, the oldest at the neck. In colour: the 8 that broke out, the biggest in gold. Tap a pane to see its post. Point at a pane to see its post.370,000 words. A pane is as big as its post is long; each colour is a year, the oldest at the neck. Standing: the 8 that broke out, the biggest in gold. Tap a pane to see its post. Point at a pane to see its post.
- 2026 22 posts
- 2025 27 posts
- 2024 31 posts
- 2023 33 posts
- 2022 29 posts
Breakouts
Experimental History8
of Experimental History’s 142 posts broke out.
On this Substack, breaking out takes 5 times the likes of the posts around it, and 10 likes at least. In gold: “Good conversations have lots of doorknobs”, with 4,881 likes where a typical post around it got 65: 75 times as many.
1 of the 142 is a video and isn’t counted.
- Last breakout
- “The Decline of Deviance”, October 2025, 25 posts ago.
- The longest
- “Ideas aren’t getting harder to find and anyone who tells you otherwise is a coward and I will fight them”, 5,236 words.
- Then and now
- A typical post got 65 likes among the first 10 and 680 among the last 10.
- Length
- Posts over 2,800 words got about twice the likes of shorter ones.
Head count is by Julián Martin, who writes Huliwood.
The posts that broke out 8
How it decides
Likes are the hearts a post has on Substack, and they follow the size of the list, so each post is set against the eight published around it (fewer at either end), and a typical one of those (their median) is its yardstick.
Posts go up and down anyway, and on some archives far more than on others: on one, half the posts land within 30% of their yardstick; on another, within 85%. So the line is set by each archive’s own ups and downs: on a steady archive it is about two and a half times, and five times the posts around it is a hit on any archive. A breakout also needs 10 likes, since 3 likes against 1 is a good week, not a breakout.
Stood up, a pane is as tall as its likes over a typical neighbour’s, on a log scale, and every breakout stands taller than every post that did not break out. A post at or under its neighbours lies flat; podcasts, threads and other writers’ posts lie flat too.
“Then and now” is a typical post among the first ten against one among the last ten, leaving out the last two weeks, whose likes are still coming in; it shows only once one side reaches 10 likes. The words are Substack’s own counts.
Free and paid posts reach different readers, so each is set against its own kind. A publication with several writers is judged on its own author, or on the writer with the most posts when the author signs fewer than half as many, since a writer with a bigger or smaller following would pass for a hit or a miss. Podcasts, threads (Substack’s own, and posts named as open or discussion threads) and posts under 100 words stay out of it; they are still in the head.
Length is measured the same way, the posts over the middle length against the ones under it, with whichever of free and paid you publish more of. “About the same” means a gap under 1.1 times, or one that chance could make; “a few more” is a real gap under 1.25 times. It takes 30 posts with likes.
The archive is read from Substack’s public pages when you ask, and when you come back after drawing your own head here. The answers are kept for about an hour, so a shared link does not read it all again. Archives longer than 1,000 posts are drawn from the newest 1,000. The last address you drew as yours is remembered in this browser only, so the page opens on your head next time and says what changed.