A friend of mine has been complaining about Facebook for about eight years now. The algorithm buried her posts. Then it started showing her things she never asked for. Then it started showing her ads that felt like they’d been listening. Every year or so she announces she’s done with it, and every year she’s still there, because where else is she supposed to put the stuff she writes?
She’s had a blog in her head for a decade. What she never had was a Saturday free to figure out hosting, DNS, a CMS, and whatever a “static site generator” is.
That’s the part that changed. Not the internet, not the platforms, not the vibes. The setup cost.
The barrier was never the idea
For most of the last twenty years, publishing on the open web has been technically possible and practically out of reach. You could learn HTML. You could learn enough to buy a domain, point it at a host, and get a page up. Many people did, of course. But it took a weekend of frustration before you had anything to show for it, and most people quit somewhere between “what’s a nameserver” and “why is my CSS doing that.”
So they went where it was easy: Facebook was easy. Instagram was easy. Medium and Substack were easy. And easy is not nothing — easy is why those platforms have billions of people and personal sites have a rounding error.
The tradeoff was that you rented your audience. You built something on land you didn’t own, and one day the landlord changed the terms.
Now the setup cost is close to zero. You can open Claude or ChatGPT, say “I want a simple website where I can post my writing, walk me through it from scratch, assume I know nothing,” and it will actually do that. It will help you pick a domain. It will explain what you’re buying and why. It will write the page. When something breaks, you paste the error in and it tells you what to do next. I think that’s a bigger deal than most of what gets written about AI.
What people are missing
The conversation about AI and the web has settled into one groove: the slop. The auto-generated articles, the fake product reviews, the SEO landfill, the search results that used to be useful.
That’s where the story usually stops, and the ending it implies is that regular people should give up on the web and go back to the apps.
I read it the other way. The flood of generated garbage is exactly the argument for more real pages made by real people about things they actually care about! Slop wins by default when everyone with something worth saying decides it’s too much hassle to put it somewhere they own. A personal site with fourteen posts about your neighborhood, your recipes, your hobby, your kid’s baseball season — that’s not competing with the slop. It’s the thing the slop is imitating.
The other worry I hear is that AI-assisted sites are somehow not really yours. I’d push back on that too. Nobody thinks a house isn’t yours because you didn’t mill the lumber. What makes the site yours is that it’s your domain, your writing, your decisions about what goes on it, and no one can bury your posts or change your reach overnight. Nobody’s grading the HTML.
What “build something” actually means here
I want to be specific, because “go make a website” is the kind of advice that sounds nice and gets ignored.
Ask an AI assistant to take you all the way through, end to end, in one sitting:
- Buy a domain. Your name is fine. It costs about fifteen dollars a year.
- Get a single page up at that domain with your name on it and something you wrote.
- Put a link on it to one other thing — your old blog, a friend’s site, a project you like.
That’s it. That’s a website. You can add posts later, or a newsletter, or photos, or a comments section you’ll regret. The first version should be small enough that you finish it, because the version you finish is the only one that matters.
If you get stuck, paste the error. If you don’t understand an answer, say “explain that like I’ve never done this before” and it will. The thing that used to end these projects — hitting a wall at 11pm with nobody to ask — mostly doesn’t happen anymore.
Why I care about this
I’ve spent most of my career helping advocacy organizations with their technology, and the pattern I see over and over is that the technology was never the variable. The variable is whether someone believed they were allowed to build the thing. Whole campaigns have run on a volunteer who decided a spreadsheet and a scrappy site were good enough to start with.
We’re about to find out what a lot of people do when the “I can’t build that” excuse stops being true. Some of it will be bad. Some of it will be strange and specific and wonderful in a way the platforms have been squeezing out for fifteen years. I’d rather have the mess.
So: pick a domain this weekend. Ask for help getting one page up. Then tell someone the address.
The open web doesn’t come back because a company decides to fix it. It comes back one page at a time, and there’s no reason yours can’t be one of them.