I put this site in my own house
You're reading this off a small computer sitting in a room in my house. Not a host, not a platform, not somebody's edge network. A single board computer that used to do something else entirely, and cost me nothing because I already owned it.
The only thing I pay for is the domain name. About ten dollars a year. That's the whole operating cost of this website.
Why bother, when free hosting exists
Because a platform can decide it's done with you overnight and there is nothing you can do about it. Accounts get suspended by systems with no human in the loop. Services shut down. Terms change. If your work lives entirely somewhere you don't control, you're one automated decision away from losing it.
This domain is mine, this box is mine, and the words are on a drive I can physically hold. The social accounts point here. Not the other way around.
It's just files
There's no database here. No app server, no login, no comment system, nothing that runs. It's a folder of plain files, and the web server's entire job is to hand them to you.
That sounds primitive. It's the strongest part of the design. Nothing to crash at three in the morning, nothing to keep patched, nothing to break into. The whole thing lives in version control, so every change is a commit and every bad idea I have is one command away from being undone.
Getting it online without opening your house
Here's the part people get wrong, and it's the part that actually matters.
The obvious way to put a home server on the internet is to open a port on your router and point the world at your house. Don't. That publishes your home address to every scanner on the planet and it makes your router the front door.
The better way is a tunnel. The little computer makes an outbound connection to a service, the way your laptop connects to a website, and that connection stays open. Traffic comes back down the pipe it already opened. Nothing inbound. No ports open. My home address appears nowhere in public.
The tightest version of this: my web server doesn't listen to the network at all. It only answers requests coming from the same machine. The tunnel program sits on that same box and is the only thing that can reach it.
So there is exactly one door into this site, and it's a door that opens outward. Nothing on my home network can reach it either. That's not paranoia, it's just the cheapest way to have fewer things to worry about.
Side benefit I didn't plan for: I have two internet connections, and when one drops the tunnel just re-establishes itself over the other. The site doesn't notice. Try getting that out of a port forward.
The part that's a little unusual
I don't intend to leave this site alone. I want to rebuild it constantly, and I want my own machines doing it.
So the design got split in two. There's a locked layer, which is the colors, the type, the mask, and the waveform you keep seeing. That never changes. Then there's everything else, the actual shape of every page, and that's supposed to change every single time.
Both halves are written down in a file that any rebuild has to read first. A rebuild that changes the colors failed. A rebuild that keeps the same layout also failed. Identity holds, structure rotates.
Every version gets kept, and you can go look at all of them. Most sites bury their old design like it's embarrassing. Watching the thing mutate while staying recognizably itself is more interesting than any single version of it, and it's the closest thing to showing my work that a website can do.
What it cost
- Domain: about ten dollars a year.
- Hosting: nothing. Hardware I already had.
- Tunnel: nothing at this scale.
- Software: nothing. All of it open.
- Electricity: a few dollars a year for a computer that sips power.
Ten bucks a year for a site nobody can take away from me. That's the entire pitch, and it's why I'd rather run it on a computer in a spare room than get it free somewhere I don't control.
If a specific piece of this is what you're stuck on, say which one and I'll write that up in more detail. It's more useful to me to fix the gap you actually hit than to guess at it.