Where everybody went

I’ve had that account since 2011. I’m not deleting it, and I’ve stopped opening it, which I suspect is how most of these end rather than with a decision.

What I’m interested in isn’t the corporate story, which everybody has read enough about. It’s what’s actually been lost, professionally, for people who do this job, because I don’t think it’s the thing people are naming.

What it actually was

I wrote a post in 2019 about a monthly meetup at a brewery that quietly stopped happening, and about how the thing it gave me wasn’t the talks. It was 20 minutes afterward, sitting next to somebody from a different agency, finding out that their Tuesday looked like my Tuesday.

That function moved online. Not deliberately, and not all at once, but by about 2016 the place where I found out how other people actually work was a timeline.

Everything I know about when a feature shipped, I learned there first. The arguments about utility classes that made me change my mind. The person who explained why my grid layout was wrong at 3 pm for no reason and no payment. The specific, concrete knowledge that everybody else’s build process is also held together with tape.

And a fair amount of employment, indirectly. Two jobs in my extended circle came out of it. Nobody put it on a résumé.

It was the long table, at national scale, open at 11:30 at night, and it took about six weeks to close.

The commons, and the sorting

Here’s the property I think mattered most, and it’s the one that’s hardest to rebuild.

It was one place. However enormous and however sprawling, the web industry had a single commons, which meant you encountered people you hadn’t chosen. The accessibility specialist who kept correcting everyone. The person who works on browsers and finds your complaints funny. The one who does government work and has completely different constraints.

In 2019 I wrote that the trouble with online communities was that the sorting is too good: you find your people and then you only see your people, whereas at a table you talk to whoever sits down. I stand by that, and I’d add that a big messy shared timeline was the closest the internet got to a room. The sorting was bad enough to be useful.

What’s replaced it is worse on exactly that axis, and I’ll come to that.

What the federated version is and isn’t

I’m on one of the Mastodon servers now, like everybody else, and I’ve been there long enough to have opinions.

The good is genuinely good. It’s chronological, so what I see is what the people I follow posted, in order, rather than what an algorithm decided would keep me there. There’s no advertising. Moderation is local and done by a person you could plausibly email, which is a completely different thing from an appeals form. And there’s no company whose fortunes I have to care about.

The loss is discovery, and it’s not a small loss. Because there’s no central algorithm surfacing things, you see what you have deliberately chosen to see and essentially nothing else. Which is exactly the sorting problem again, only now it’s structural rather than incidental.

So the version of me who found out about CSS Grid in 2017 because it drifted past would not find out about the 2027 equivalent, unless I’d already followed the right people, which requires knowing who they are, which is the thing I’d have used the timeline to learn.

I don’t think that’s fixable by a feature. It’s what happens when a commons becomes a set of rooms.

Rented land, and a caveat

The obvious lesson is the one everybody’s been posting: don’t build your professional identity on land you don’t own. A decade of writing, connections and reputation living inside a company that can be sold on a whim, and then is.

I do believe that. This site is eight years old, it’s a folder of text files, it would survive any company on earth going under, and everything on it is still where I put it.

But I’d like to put the counter properly, because “just have your own website” is advice that sounds better coming from someone who already has an audience.

The reason platforms mattered is that they were how people without a network got heard. If you’ve got a blog and nobody knows you exist, you have a diary. The timeline was a distribution mechanism, and it was free, and it did not care where you worked or who you’d met. That’s the same argument I made about public code in 2018: the thing that looks like a leveling device usually is one, for somebody.

So the honest version isn’t “own your platform and you’ll be fine.” It’s that ownership protects what you’ve made and does nothing at all about reach, and the reach problem is now harder for someone starting out this year than it was for me in 2014. I don’t have an answer to that and I’m suspicious of people who sound like they do.

What I’m actually doing

Two things, both unglamorous.

I’ve gone back to a feed reader, which I abandoned in about 2013 when the one everybody used was shut down. I’ve got somewhere north of 80 sites in it now, built up over a few evenings of following links from people’s blogrolls, and it is better than the timeline was. Long-form, no reply counts, no arguments, and I read things properly instead of skimming.

It’s a worse discovery mechanism and a much better reading one, and I’ve decided I’ll take that trade and solve discovery by hand, slowly, by following links.

And I’m writing here more, which is partly the point of the above and partly that I notice I’ve been drafting sentences in my head all year with nowhere to put them.

Anyway, six weeks. I’d have said that thing was a permanent feature of my working life, and it turns out it was a product, and products end.

The long table also ended, mind you, and I did fine. I’d just quite like to stop learning this particular lesson every few years 😔