The quiet rebuild

The conversation is almost entirely about what's breaking. I keep noticing something else.

What's actually growing

Kagi, the search engine, built something called the Small Web. It's a curated index of independent, non-commercial sites. Personal blogs, small project pages, people writing on their own domains. The index has been growing steadily, and thousands of those sites publish at least monthly.

This isn't hobbyist nostalgia. These are people who write regularly, on their own terms, on infrastructure they control.

They're not alone. Marginalia is an independent search engine that prioritizes non-commercial writing over SEO-optimized filler. Blogroll.org curates over a thousand blogs by hand. Personalsit.es collects personal websites into a browsable directory.

These are small projects, but they're solving the same problem from different angles: making independent sites findable.

The discovery gap was always the strongest argument against publishing on your own domain. If nobody can find you, why bother? That argument is weaker than it was a year ago.

And RSS never actually died. It stopped being fashionable. Podcasts kept it alive as infrastructure, and now a growing number of readers are returning to feed readers as an alternative to algorithmic timelines.

I read most of my web through Reader, an RSS app. The protocol is the same one it was twenty years ago. It still works.

The social layer nobody talks about

Discovery is one problem. Conversation is another. When you publish on a platform, the platform handles replies, likes, and shares. When you publish on your own site, that layer disappears.

Except it doesn't have to. The W3C standardized Webmentions in 2017. It's a protocol that lets independent sites notify each other of replies, mentions, and interactions. No central server. No algorithm. One site telling another: I wrote something about what you wrote.

I run Webmentions on my personal blog. The volume is small. A handful of mentions, mostly from the kind of sites you'd expect: builders who care about HTML and CSS, people like the ones behind nerdy.dev, who run their own domains and publish on their own terms. It's a small circle, and it will probably stay that way.

But the principle shapes how I think about publishing. Writing for my own domain first. Letting the social layer be an open protocol rather than a platform feature. Treating conversation as something that happens between sites, not inside a silo.

I want to be honest about the limits. Webmentions adoption is niche. The infrastructure is mature, well-maintained, and standardized, but the community using it is small.

Most people will never configure a Webmentions endpoint. That's part of the observation, not a counterpoint to it. The tools are ready. The crowd hasn't arrived yet.

We've seen this before

My first website was hand-coded in Notepad. I spent hours on sites like k10k.net, following links from one personal page to the next. That was how the web worked. Everyone had their own page. Links were the social fabric. Then platforms solved real problems: discovery, identity, interaction at scale.

People moved to where the people were. The independent web got quiet.

Now the platforms are failing at the things that made them useful. Discovery is buried under AI-generated noise. Identity is diluted by bot accounts. Interaction is shaped by algorithms optimizing for engagement, not conversation.

The reasons people moved to platforms are eroding.

The cycle is familiar. What's different this time is that the infrastructure for an independent web is more mature than it was during any previous turn. The standards exist. Webmentions, RSS, microformats.

The discovery layer is being built in real time. The cultural momentum, people tired of algorithmic feeds and generated noise, is providing the pull.

This isn't the first time the web has decentralized. It's the first time the tools are this ready for it.

Waiting for nobody

I don't know if this becomes the next web or stays a quiet corner of it. Most people will keep using platforms. Platforms aren't going anywhere.

But the assumption that you need them to be found, to have conversations, to exist on the web, that assumption is weaker than it used to be. You can publish on your own domain and still have conversations across sites. You can subscribe to writers directly through RSS without an algorithm deciding what you see. You can own your words, your replies, your links.

That's not a technical argument. It's a design choice about where your work lives.