Download websites that use Etherpad

About Etherpad

Collaborative editing was a research problem long before it became a feature of every office suite, and Etherpad is one of the few pieces of that early work still running in public. The idea was that several people could type into the same document in a browser at once and watch each other do it, which sounds unremarkable now and did not in 2008. A Apache 2.0 licence is the reason the software is still around at all, because anyone may run a copy without asking permission and no vendor is in a position to withdraw it.

Our catalogue files Etherpad under Rich text editors, and the honest reading of that label is that a self-hosted writing surface does not fit buckets designed for analytics scripts and shop platforms. The label matters less than the deployment pattern. An instance goes up deliberately, on hardware or a host somebody controls, usually because the text being written should not sit in a commercial cloud account. Universities, newsrooms, activist groups, hosting cooperatives and internal engineering teams are the recognisable operators.

As of October 2026 we detect Etherpad on 2 sites, and the figure is a poor estimate of how many instances exist. Self-hosted software of this kind lives on subdomains, on internal hostnames that resolve only inside a network, and behind logins, none of which a public crawl of registered domains will enumerate. A fingerprint needs a reachable page. Anything mounted at an address we never fetch stays invisible to us, and for a tool whose selling point is keeping documents off the open internet, that invisibility is the point rather than a defect in our data.

What survives the caveat is still useful. The JavaScript front end means the instances we do find are identifiable from the page itself rather than inferred from headers, so a match is a strong one. A list of publicly reachable instances is a map of organisations that chose to run their own infrastructure instead of buying a hosted document service, and that procurement signal generalises well beyond this one editor. Anybody selling self-hosting, support contracts, or a privacy-positioned alternative wants that population in a file, joined to country and top-level domain, rather than checking domains one at a time and never learning how many there are.