Archie

What Archie was, and what it actually indexed

Archie was a searchable catalogue of the files sitting on public anonymous-FTP servers — the machines that, before the web, held nearly everything the internet had to offer in the way of software, documents and data. You gave Archie a filename or a fragment of one, and it told you which hosts held a file of that name and in which directory. That was the whole product.

Archie indexed file listings, not document content. This is the single fact most often got wrong about it, and it is worth stating flatly. Archie read nothing inside the files it catalogued. It ranked nothing. It summarised nothing, and it had no notion of what any file was about. If you did not already know roughly what the thing you wanted was called, Archie could not help you.

Nor was it a web search engine. It could not have been: Archie went live on 10 September 1990, roughly a year before the World Wide Web became publicly available, and it never indexed a web page at any point in its life. The distinction matters more than it sounds, because it is the reason two apparently contradictory statements are both true — Archie was the first internet search engine, and Archie was not the first web search engine.

How it worked

The mechanism was simple enough to describe in four steps, and understanding it explains both what Archie could do and what it could never do:

1. Archie maintained a list of public anonymous-FTP sites. That list was curated by people.
2. On a schedule — roughly monthly — the server connected to each site and pulled down full recursive directory listings: hostname, path, filename, size, date.
3. Those listings were merged into a central database of strings.
4. Users matched against that database by exact filename, substring or regular expression, depending on which client and which server they used.

There was no crawler in the modern sense, because there was nothing to crawl. A web crawler discovers new pages by following links out of the pages it already has; Archie had no links to follow, so it went only where its operators told it to go. There was no link graph, no relevance ranking and no full-text retrieval — all three arrived later and elsewhere.

One companion piece added a descriptive layer: a separate database called Whatis held short, human-written descriptions of software packages. It was hand-curated metadata rather than anything derived from the FTP listings, and it is the only part of Archie that told you what something was rather than where it was.

You could reach Archie four ways: a local archie command-line client; a direct Telnet session to a public Archie server; an email query — send a message containing find <string> and get the results back by mail, which was the standard method for anyone whose connection was mail-only; and, from 1993, web gateways. The best known gateway was ArchiePlex, an HTML form front end written by Martijn Koster, who also wrote ALIWEB and the robots.txt standard.

Was Archie the first search engine?

Archie is routinely called “the first search engine”. That is defensible with one word added, and indefensible without it. The accurate formulation is: Archie was the first internet search engine. Every other claim to a “first” in this story is a claim to a different title, and the answer you get depends entirely on which qualifier you attach to the question.

The claimants, with what each one actually claims:

• Archie — 10 September 1990. First internet search engine. Indexed FTP filenames, not content.
• Veronica and Jughead — 1991 to 1993. The same idea applied to Gopher: they searched Gopher menu titles, not content.
• World Wide Web Wanderer (Matthew Gray, MIT) — June 1993. The first known web robot, which built an index called Wandex. Gray's stated purpose was measuring the growth of the web, and he does not claim it was a search engine.
• W3Catalog (Oscar Nierstrasz, University of Geneva) — 2 September 1993. Wikipedia's timeline calls it “the world's first web search engine”. It worked by mirroring and reformatting existing hand-maintained link lists: searchable, but not a crawl.
• ALIWEB (Martijn Koster) — announced November 1993, launched May 1994. Often called the first web search engine. It was built from index files submitted by site owners; there was no crawler.
• JumpStation (Jonathon Fletcher, University of Stirling) — 12 December 1993. The first web resource-discovery tool to combine all three essential features — crawling, indexing and searching — and, per Wikipedia, the first that behaved and appeared to the user the way current search engines do. It indexed titles and headings rather than full text, and had no ranking.
• WebCrawler (Brian Pinkerton, University of Washington) — 20/21 April 1994. The first web search engine to provide full-text search: the first that indexed the words inside pages.

So: first internet search engine, Archie. First web robot, the Wanderer. First web search engine, W3Catalog by date and JumpStation by function. First full-text web search engine, WebCrawler. State the qualifier, or the sentence means nothing.

One earlier claimant occasionally circulates and should be set aside: WHOIS, from 1982. It is a directory lookup protocol for registration records, not a search engine over files or content, and Wikipedia's timeline of web search engines does not list it.

1989 or 1990? The launch date is disputed

Sources disagree on when Archie began, and the disagreement is worth printing rather than resolving. Wikipedia's Timeline of web search engines gives 10 September 1990 as the date Archie went live, and Wikipedia's article on Archie says the original implementation was written in 1990. McGill University's own bicentennial page and the Internet Hall of Fame both date the creation of Archie to 1989.

A third figure muddies things further: Wikipedia's article states that Archie “first appeared” in 1986, which appears to refer to precursor work at McGill rather than to any public service.

The sensible reading is that development ran across 1986–1989 and the service went live to users in September 1990, which is why the institutional accounts date it to the work and the timeline dates it to the launch. Anyone citing a single year should say which of the two they mean and whose account they are following.

Who built it, and who paid for it

Archie was created and run at the McGill University School of Computer Science in Montreal. Alan Emtage wrote the original implementation, with Bill Heelan and Peter J. Deutsch credited alongside him. It began as a university service project with no revenue model at all.

From 1992, Emtage and Deutsch founded Bunyip Information Systems to commercialise it, selling a supported Archie server to institutions that wanted to run their own. The Internet Hall of Fame describes Bunyip as “the first company dedicated to Internet information services” — a claim worth noting, though it is the Hall of Fame's own framing rather than an independently established fact. Advertising was never involved at any stage, in any form.

Emtage was inducted into the Internet Hall of Fame in 2017, cited as having “conceived of and implemented Archie, the world's first Internet search engine”.

Independent mirrors run by universities worldwide are how the service scaled without a business behind it. The last of them, at the Interdisciplinary Centre (ICM) of the University of Warsaw, outlived every other Archie server by decades.

What happened to Archie

There is no single shutdown date. Archie faded rather than closed.

• 1991–1992: Gopher's rise produces the analogous tools Veronica and Jughead, which search Gopher menu titles the way Archie searches FTP filenames.
• 1993: ArchiePlex and other web gateways put an HTML form in front of Archie — the first time most people saw it in a browser.
• 1996: version 3.5 is released. It is the last.
• Late 1990s: development ceases. Full-text web search engines make filename search of FTP archives a niche concern, and the archives themselves start moving onto the web.
• Until 2023: a legacy Archie server keeps running at ICM, University of Warsaw — the last one anywhere.
• 11 May 2024: The Serial Port, a computer-history preservation project, recovers the Archie 3.5 beta source from that Polish server and launches a working public instance as a museum exhibit, publishing the source and documentation.

That revival is why “Archie is back online” circulates. It needs a date attached. Checked on 19 August 2026, the Serial Port instance displays “Our Archie search service is currently offline”; the operators have said the Sun SPARC virtual machine running the server is down and that the Archie software “is rather complex and not fun to work with at times”. Anyone repeating that Archie is running again should re-check before doing so.

What it was good at, and what it was not

Archie's strength was narrow and real. If you knew the name of a file — a specific piece of software, a specific document — Archie would find you a host that had it, quickly, across a network that had no other way of answering that question. In 1990 that was transformative.

What it was bad at was everything else discovery-shaped. It could not tell you what a file was, what a topic contained, or which of forty hits was the current version. Using Archie was a skill: you had to guess filenames, and guessing badly returned nothing. Nobody should be pointed at Archie today as a working tool, and the revived instance is a preservation exhibit rather than a service — currently a closed one.

One usage figure circulates and should be handled carefully. Wikipedia states that at its peak Archie accounted for 50% of Montreal internet traffic; McGill's own page says instead that Archie “attracted half of web traffic in Canada” until the mid-1990s. Those are different claims — one city against one country — and McGill's phrase “web traffic” is anachronistic for a service that predates the web. Treat the figure as an oft-repeated illustration of how small the early internet was, not as a measured statistic.

Common misconceptions

“Archie was the first web search engine.” No. It was the first internet search engine. It predates the public web and never indexed a web page.

“Archie searched the contents of files.” No — filenames and directory paths only. The separate Whatis database held human-written package descriptions, and that was the only descriptive layer Archie had.

“Archie crawled the internet.” Not in the modern sense. It connected to a maintained list of known FTP hosts and downloaded their directory listings. It did not discover new hosts by following links, because in 1990 there were no links to follow.

“Archie is named after the comic-book character.” Emtage has explicitly denied it: the name is “archive” minus the “v”. The Gopher-era successors Veronica and Jughead leaned into the Archie Comics joke afterwards, which is where the retroactive assumption comes from.

“Archie was brought back in 2024, so you can use it.” An instance launched on 11 May 2024, but as of 19 August 2026 it is offline.

Frequently asked questions

Sources

• en.wikipedia.org/wiki/Archie_(search_engine)
• en.wikipedia.org/wiki/Timeline_of_web_search_engines
• 200.mcgill.ca/history/creation-of-the-first-internet-search-engine/
• internethalloffame.org/inductee/alan-emtage/
• historyofinformation.com/detail.php?id=101
• en.wikipedia.org/wiki/JumpStation
• en.wikipedia.org/wiki/WebCrawler
• archie.serialport.org/