Archie FTP Search Tool Released
In 1990, Alan Emtage, then a systems administrator at McGill University, released Archie, a tool that automatically indexed filenames across publicly accessible FTP servers and allowed users to search that index by query string, creating the first searchable index of internet-hosted files.
Background
By the late 1980s, a large number of universities and research institutions had connected to the internet and were running FTP servers, the standard way at the time of making files publicly available for others to download. The files were there, and anyone with network access could retrieve them. The problem was finding them.
There was no index. If you wanted a particular piece of software or a dataset, you either already knew which server held it, or you asked around and hoped someone did. The collective store of publicly shared files across the network was, in practice, invisible. It was a library where every book was shelved in a different building, none of the buildings had a catalogue, and you could only get in if you already knew the address.
Some people maintained hand-curated lists of useful FTP sites. These circulated informally and went out of date quickly. The network was growing fast enough that no human effort could keep up with it.
What happened
Alan Emtage, Bill Heelan and J. Peter Deutsch were working as systems administrators at McGill University in Montreal when Emtage wrote the first version of what became Archie. The tool did something simple but, at the time, novel: it connected automatically to publicly accessible FTP servers across the internet, pulled down their file listings, and stored those listings in a local database. Users could then send a query to that database and get back a list of filenames matching their search, along with the addresses of the servers holding them.
The 1992 paper Emtage and Deutsch presented at ACM SIGUCCS describes the system in some detail. Archie did not fetch or read the files themselves, only their names and locations. That was enough. A researcher looking for a specific program no longer needed to know in advance where it lived. They queried Archie, got a list of candidate servers, and went directly to the right one.
What made it work as a practical service rather than a one-off experiment was the regular, automated re-indexing. Servers were polled on a schedule, so the database reflected the network as it actually was rather than as it had been months earlier. Other institutions later ran their own Archie servers, which helped distribute the query load as use grew. The design kept indexing and querying as separate steps, which meant each could be improved or scaled without disturbing the other.
Why it mattered
Archie demonstrated that it was practical to automatically harvest, index, and query distributed file listings across a network at scale, a conceptual precursor to the web crawler and search engine architectures that would follow. By separating the acts of indexing and querying, it established a design pattern that later search engines refined rather than replaced. Although Archie predates the World Wide Web and indexes only filenames rather than document content, it addressed the core information-retrieval problem of making an otherwise unnavigable distributed corpus discoverable.
People
Alan Emtage, Bill Heelan, J Peter Deutsch
Organisations
Sources
- Archie, An Electronic Directory Service for the Internet.ACM SIGUCCS.Primary source
- Brief History of the Internet.Internet Society.Secondary
Cite this page
AI Achievements. (1990). Archie FTP Search Tool Released. Retrieved 2026-08-22, from https://achievements.ai/milestone/first-search-engine-archie-launched
@misc{achievements_first_search_engine_archie_launched,
title = {Archie FTP Search Tool Released},
author = {{AI Achievements}},
year = {1990},
url = {https://achievements.ai/milestone/first-search-engine-archie-launched}
}