Bluesky says I have 2,558 followers. For two years I've treated that number as the verdict on my writing: that many people decided my posts were worth seeing, every day, and kept deciding. Next week I turn 21, and before the number ticks over I wanted to actually meet my audience.
So I pulled the list.
The neat thing about Bluesky is that the data isn't locked away. Every follow is a public record stored on the follower's own server, in a repository anyone can query. So I did. I resolved each of the 2,026 visible accounts to the server that hosts them, walked every follow record they'd ever written, and matched them against my own identifier. Two thousand accounts, spread across servers I'd never heard of, all of it public, all of it free, none of it authorised by anyone. This is what the AT Protocol is for.
The number was lying
2,558 is what the profile says. When you ask for the actual list of followers, you get 2,026. The gap isn't people who unfollowed — it's accounts the platform stopped showing. Taken down, deactivated, muted, blocked. The profile count is a running aggregate that never gets cleaned up. The list is filtered. Most of that missing ~530 are bots.
And the list itself isn't all people. 237 of my followers follow 20,000 or more accounts. Some follow over a million. One of my followers is a vegetable-garden planner that follows 1,088,063 accounts. A garden planner. Following every person it has ever seen. Charles Keener follows 843,351. Isidor follows 457,117. These are the accounts that turn "follow" from a decision into a reflex. Count them the way the platform does and you're counting the same thing twice — once for them, once for everyone else.
Another 250 of my followers have never posted. Not once. Some are bots, some are people who follow and never speak. The platform counts both identically, and so, until now, did I.
Strip those out, and the honest number of people who chose to follow me is somewhere around 1,700.
I'm not going to pretend 2,558 was a lie. It's the number the platform showed me, and it was real. It just wasn't a number of people. It was a count of relationships, and close to a third of them were made by scripts.
The people who are actually there
So who are the 1,700? That's the part I wasn't expecting.
621 of them — 31% — run their own domain as their handle. On a platform where the default is a shared .bsky.social address, a third of the people who follow me decided to bring their own identity. Not .bsky.social with a cool handle, their own domain, their own rules for their own identity. Twenty-six of those domains are already dead, though — the handle broke, and Bluesky renamed the account handle.invalid. I've written a whole essay about link rot. I apparently live in it.
There's Bryan Newbold — the person actually maintaining the AT Protocol this whole site runs on. He follows me. So do Alex and Jim Ray from the Bluesky team. Not because I asked. Because I kept posting.
There's Peter Morley, who has 166,000 followers and writes about the same things I do. There's Ben Werdmuller and Steve Klabnik, people whose words I read before I ever posted a single one. 216 of my followers have more than ten thousand followers themselves. 203 of the people who follow me, I follow back.
The numbers I keep coming back to, though, are the boring ones. 761 of my followers — 38% — had a Bluesky account before I did. And 819 — 40% — joined in the same autumn as me, October to December 2024. This is not a crowd that found me. It's a cohort that found each other.
I told myself the story that this audience was mostly my own ecosystem — the self-hosting, indieweb crowd who found me because I kept building the machinery of the protocol itself. Malachite, pulling my scrobbles back from a dead LastFM connection and pushing them to teal.fm. Bismuth. Tourmaline. A link shortener running on my own board. And it's true that the people with the most followers are exactly those people. But 69% of my followers are just people on a social app who read something I wrote and decided to stay.
That's the humbling part. I thought I'd built an audience of engineers. I'd actually built an audience of readers.
The growth wasn't compound interest
I wanted this to be the nice chart. Slow start, then the curve bends upward, and you can draw a line from every post to the next follower. That's not what happened.
The receipts, for the record:
eighteen days to my first ten followers
six months to fifty
200 on the 7th of November 2024
1,000 on the 21st of December 2024
1,500 on the 30th of December 2025
608 people followed me in November 2024. That is 31% of everything I have ever gained, in one month, because that's when everyone joined Bluesky. I went from 100 to 1,000 followers in nine weeks. Then it took a full year to go from 1,000 to 1,500. And then, when nobody was watching, 2026 picked back up — 480 follows in eight months, on pace for about half as much again as the whole of 2025. The curve is bending again, and this time it's not the platform's doing.
The most viral thing I ever did was sign up during an exodus. Well, sign up and then post nothing for three and a half months. My account is from April 2024. My first post is from the 2nd of August. I lurked for the entire summer, and a small handful of people followed me anyway. That's the closest thing I have to a talent.
The posts that "went viral" didn't
Here's where I have to correct my own record, and I have the receipts — the public API returns them and there aren't any to return.
The Sony piece, the one I'm proudest of this month, the one about the whole industry being okay with no discs at all? Four likes. No reposts, no quote-posts. It went nowhere.
The Mac-viruses post got two likes. The pentacle posts, the ones I thought would connect with people, got between zero and three each.
I said those posts went properly out past the ATProto crowd. They didn't. They got read by my normal audience — which is small — and my normal audience is exactly who was supposed to read them.
The actual strategy
I've claimed for two years that I have no posting schedule. I post when I've got something to post, I told people, which is most days, but "most days" isn't a strategy.
Let me correct the record with worse news.
8,880 posts in two years. Twelve a day. On 97.7% of days — I've missed seventeen days since I started posting in August 2024. That is not "most days." That is a schedule with a fancier name, and it is the only strategy I've ever had.
November 2024 — the month everyone arrived, the month I gained 608 followers — is also one of my three most prolific months on record. 493 posts in thirty days. Sixteen a day. I didn't plan that. I was just, apparently, unable to stop.
The median post gets one like. One. The most popular post I've written since this all started — the job-hunt post — got fifty. There was no point where a post "went viral" and everything changed. There was only the steady grind of a small number of people who kept coming back because I kept showing up.
Which is the actual answer to the question I set out to answer. I asked what 2,558 followers meant. The real number is closer to 1,700, and most of those aren't even there for the reasons I assumed. They're not there because of the viral posts that never happened. They're there because I posted twelve times a day, every day, for two years, and a thousand and a half people decided that was worth their time.
That's not a big audience. It's a real one. Real is better.