Almost every website lives on a server in a data centre, or passes through one. Link Node does something different: it serves a site straight from the owner's own computer to each visitor's browser, peer-to-peer. Here is how that works, and what it is good and not so good for.
The problem: home computers are hidden
A home computer sits behind a router. The router shares one public address among every device in the house and, by default, lets no one from outside start a connection in. That is good for safety, and it is why hosting at home has traditionally meant port forwarding, a fixed IP or dynamic DNS, or giving up.
How Link Node gets through
Browsers already know how to connect directly to another computer behind a router: it is how video calls work. The technology is called WebRTC, and Link Node uses it to serve websites.
Your computer runs Link Node, which keeps a small connection open to the Link service and has a permanent address such as https://58310427.mdoc.us.
A visitor opens that address. The Link service sends their browser a small page, the same for every site, that starts a connection to your computer.
The Link service passes a few setup messages between the two sides, the way a phone exchange connects a call. Your router lets the reply through because your computer started talking first. Where the router allows it, Link Node also asks it for a port mapping.
The browser and your computer now talk directly over an encrypted link, and your pages, images and files travel along it. The browser also checks that the computer answering is really the one behind that address, using a key that never leaves your machine.
The Link service never sees your pages. It knows which addresses exist and which are online, and it introduces visitors to them. That's all.
How it compares
Ordinary web hosting keeps a copy of your site on the host's servers, always on, fast everywhere, and in someone else's hands.
Tunnels and reverse proxies let a home machine serve a site by sending every visit through the tunnel provider's servers. It always connects, but every byte passes through a third party.
Peer-to-peer sends every visit straight from your computer to the visitor. Nothing is uploaded and nobody in the middle carries your content, but your computer has to be on and reachable.
The honest limits
Your site is online when your computer is. When it sleeps, visitors see your offline page: one small static file that the Link service is allowed to keep, so it can show it while you are away. For an always-on site, run Link Node on something that stays on, such as a Raspberry Pi.
Some networks can't connect directly. On a few strict networks (some offices, some mobile carriers, certain routers on both ends) a direct connection is impossible. Link deliberately has no relay, so those visitors see a clear message rather than your content passing through Link's servers.
Speed depends on your upload. Each visitor is served by your home connection. That is plenty for a personal site, a portfolio or a small shop page, but not for sudden crowds.
Static files only. Link Node serves files; nothing in the folder runs on your computer. Pages can still use JavaScript in the visitor's browser, so galleries, maps and small apps work, but there is no server code.
Who it is for
People who want their own place on the web without renting one: a homepage, a portfolio, a page for a club or a small business, a way to share files with friends. And anyone who likes the idea that their website lives with them. The core is open source, so you can read exactly what runs on your machine.
Try it: Link Node is free, with no account. Download it for Windows, Mac or Linux, or run the open-source command-line node on a Raspberry Pi.