Archie

The world's first program that searched for files across the whole internet on its own: in 1990 a student at McGill University in Montreal wrote a script that went through anonymous FTP archives around the world and compiled a list of what was stored there, a job that had been done by hand before. Two years later the authors tried to sell access to the service to companies, but with the arrival of the web and Google the program was no longer needed and died in the late 1990s.

In brief

Archie was the world's first program that searched for files on the internet on its own, in place of a person: in 1990 Alan Emtage, a graduate student at McGill University in Montreal, wrote a script that went through thousands of anonymous FTP archives around the world and compiled a single list of what was stored there. Before that, the only way to find out which of hundreds of servers held the file you needed was by hand or on someone else's tip. In 1992 Emtage and his boss Peter Deutsch turned the tool into one of the first internet companies, Bunyip Information Systems, and tried to charge providers for a license. A few years later the web arrived, and with it Google and AltaVista, and indexing nothing but file names on FTP no longer made sense: Bunyip shut down in 2003. Neither Emtage nor Deutsch got rich, because back then it occurred to nobody to patent the idea.

How it started (the founders)

Alan Emtage was born in 1964 in Barbados, into a family that encouraged curiosity: his aunt, a teacher, took him to listen to BBC science programs and to explore Carlisle Bay. In 1981, on a trip to Great Britain, he got a Sinclair ZX81, his first computer. In 1983, after winning a Barbados Scholarship, he enrolled at McGill University in Montreal; he first considered meteorology and geology but chose computer science as the safer career in the middle of the 1983 recession. After earning his bachelor's degree in 1987, he stayed on for a master's and at the same time took a job as a systems administrator in the faculty's IT department, and that is where he ended up reporting to Peter Deutsch, who by then had spent several years pressing the McGill administration for a proper internet connection (the first line to Boston cost about $35,000 a year).

Emtage's job was to track useful free software on anonymous FTP servers around the world by hand; Deutsch himself jokingly called him the department's resident rat. In June 1990, by his own account, Emtage got tired of logging in to every server manually: "Rather than spending my time logging on to FTP sites... I wrote some computer scripts that would do the same thing, and much faster too." The program was named Archie, from "archive" without the "v." When Deutsch saw the internal tool, he suggested making it public; the reaction of colleagues: "Everybody was like, 'Oh my God — of course! Why didn't we think of this?'" A third co-author, Bill Heelan, worked on the first implementation together with Emtage, but it was Deutsch, as the boss, who after the fact became the public face of the project and, two years later, the president of the company.

Year-by-year timeline

1986: Peter Deutsch becomes systems manager of the Computer Science faculty at McGill and starts pressing the administration for internet connectivity fact
1989: Emtage is given the task by Deutsch of tracking useful software on FTP servers by hand; Emtage himself later dates the idea for Archie to this year fact
1990-06: Emtage writes the first scripts for automatically crawling FTP archives, for his own convenience fact
1990-09-10: public release of Archie on the host archie.mcgill.ca; the source of the date is a Usenet post by Deutsch himself (comp.archives, 1990-09-11), which, in his words, set off an avalanche of requests and led to the public front-end fact
1992-01: Emtage and Deutsch found Bunyip Information Systems in Montreal. Caribbean Beat magazine calls it the first internet startup in history, but UUNET was founded earlier, in 1987 (see the UUNET dossier); some sources (written well after the events) give 1990, but Deutsch's primary 1993 interview and the institutional biographies agree on January 1992 estimate
1992-10: the first commercial version of Archie comes out, completely rewritten software; licenses are sold to host providers, while end users on most Archie servers keep free access fact
1993: Archie handles about 100,000 queries a day on ~25 servers around the world; Bunyip has six people (five full-time, two part-time) fact
1994: the index covers close to 1,000 file archives around the world; RFC 1635 on anonymous FTP comes out, co-authored by Deutsch and Emtage fact
1996: Emtage leaves Bunyip over disagreements about raising outside investment; in his words, Montreal lacked the ecosystem of Silicon Valley fact
mid-1990s: Archie is pushed out first by Gopher search tools (Veronica, Jughead), then by web search engines (AltaVista, Yahoo!); work on the software stops by the end of the decade fact
2003: Bunyip Information Systems is officially dissolved fact
2017: Emtage becomes the first person from Barbados and the Caribbean to be inducted into the Internet Hall of Fame fact
2023 → 2024-05-11: the legacy server at the University of Warsaw is switched off; the retro enthusiasts of The Serial Port, with help from the same university, launch a new public server, archie.serialport.org, indexing about 30,000 files

Lesser-known but significant facts

Emtage did not work alongside Deutsch as an equal but reported to him. The public image of three student co-authors simplifies the real hierarchy: Deutsch was the systems manager and Emtage's boss, and he gave him the task as a routine work duty rather than inviting him in as a co-founder from the start.
Growth was set off by a single Usenet post. Archie was conceived as a purely internal tool; everything changed when it was mentioned in passing on an open mailing list, and a flood of requests from outsiders poured in, forcing the team to bolt on a public front-end in a hurry.
The company name Bunyip was chosen out of nostalgia, not for marketing. Deutsch is half Australian; the bunyip is a creature from Aboriginal mythology and the hero of a children's book he read as a boy. The name, as he put it, simply stuck in his head.
The load was designed from the start as a federation rather than as one server for the whole world. Early on, Archie servers in different countries (Finland through FUNET, Australia through AARNET) crawled only their own region and exchanged ready-made copies of the data with each other: peer replication, not a naive crawl of the entire planet by every node.
The creators themselves rejected X.500, the official international directory standard, as overly complex and political, and chose instead to build their own lightweight protocol, WHOIS++, through the IETF, a deliberate bet on simplicity against the consensus heavyweight.

Legend vs. the record

Legend: Archie was the first search engine. But what exactly did it search? The wording is correct but incomplete: Archie indexed only the NAMES of files on anonymous FTP servers, with no full-text search, so the user had to know roughly what the file was called. Before Archie there was nothing at network scale for this task; people asked on mailing lists or went through archives by hand. Verdict: the first automated internet index, but a narrowly specialized one, not a search engine in today's sense.
Legend: Google did the same thing as Archie. Emtage himself describes the connection far more cautiously than the retellings do: "Archie was the great great grandfather of Google," that is, a very distant ancestor through a shared idea (automatically collect data, put it into an index, offer a simple query), not the same method at a different scale. Archie did no full-text search and did not rank results by relevance; that did not exist and could not have existed for a task defined as a list of file names. Verdict: the shared idea of automating search, yes, was inherited; the technology itself and the complexity of the task, no, those are qualitatively different levels.
Legend: the creator of the first search engine got rich. The record says the opposite: "At the time, nobody was making money off of the Internet, and we didn't patent any of the original ideas behind Archie... So the patents would have been where I would have made the money." The search industry today is worth hundreds of billions of dollars a year, and Emtage gets nothing from it, but he is philosophically calm about it: "I wouldn't change anything." Verdict: no patent, no stake, no regrets.

The first growth lever

The growth lever was not marketing or a launch but chance: Archie was conceived as a strictly internal McGill tool, and Deutsch says so in plain words: "we mentioned it in a Usenet posting once, got flooded with postings with people asking us to do searches. So we stuck a frontend onto it, and that was sort of it." One post on an open mailing list, and the team was building out an interface in a rush to meet the demand that poured in. The same pattern of one community thread setting off growth shows up today on Reddit; in 1990 the venue simply was Usenet.

The second, less obvious lever was that growth did not bring down the infrastructure. Deutsch stresses: "we were actually very aware of this from the beginning," and from the start the team built a federated network of mirrors. Finland (FUNET) and Australia (AARNET) crawled only their own region, and data spread between nodes as copies, not through every server crawling the whole world again. Without that foresight, the viral effect of a single post could have killed a lone server in Montreal, instead of producing a worldwide network of ~25 nodes and ~100,000 queries a day by 1993.

The paired story

The pair is WAIS. Archie and WAIS solved the same problem of 1990–1991, finding something on the internet before the web existed, in opposite ways. Archie indexed only the NAMES of files on FTP servers and knew nothing about what was in them, mechanically similar to a library catalog without annotations. WAIS (Thinking Machines, 1991) searched the contents of the documents themselves and was built from the start as a partnership of Thinking Machines, Apple, Dow Jones, and KPMG Peat Marwick (see the WAIS dossier). Their commercial fates diverged as well: Bunyip sold licenses to providers and was dissolved in 2003, while WAIS Inc. was bought by AOL in 1995 for $15 million in stock, and with that money Brewster Kahle built the Internet Archive (see the WAIS dossier). The pair illustrates the fork between searching by name and searching by content well, the same choice that faces any modern indexer of other people's content (see the parallels below).

Parallels today (projects from the catalog)

Awesome Indie (awesome-indie): the same mechanics of an index of other people's content as a product. A free curated list of other people's launches, ranked only by upvotes, where the only thing you can pay for is a convenient publication date. As with Bunyip, monetization is attached on the side and does not touch the bulk of users, most of whom keep using it for free. The difference in eras: Awesome Indie's growth is pulled by one channel (a launch on Product Hunt), while Archie's came from a chance Usenet post and a federated architecture designed in advance. view this project's dossier →
AI Directories (ai-directories): a direct parallel to the Bunyip model. A free curated index (200+ AI directories) plus a paid add-on where they do the submitting for you ($99–199 one-time), exactly the same split into "freely available" but not "free" that Deutsch talked about in 1993: the open index stays free, and people pay only for a specific additional service on top of it. view this project's dossier →

What a builder can take from this in 2026

An internal tool built for your own laziness is a legitimate source of a product. Emtage was not designing a search engine; he was automating a boring routine for himself, and that is exactly why the solution turned out simple and immediately useful to many people.
One post in the right place can replace an entire marketing plan. Archie's growth began not with a launch but with a chance mention on Usenet: the demand already existed, and all it took was one public signal.
Design for load in advance, not after the first server crash. Federated replication by region was thought through from the very start, not added in a panic after the first surge of traffic.
Freely available and free are different things, and it is fine to explain that to users directly. Deutsch refused to apologize for the move to licensing: "letting people starve is a crime," while charging a reasonable fee is not.
Being first does not guarantee money; execution and timing matter more. Emtage acknowledges that not patenting was a decision of its era ("nobody was making money off the Internet") and has no regrets: it is part of the same chain of decisions as the invention itself.

Discrepancies and what we could not verify

The year Bunyip was founded: 1990 vs. 1992. Caribbean Beat gives 1990; Deutsch's own primary 1993 interview, the Internet Hall of Fame, and Wikipedia agree on January 1992. We went with the stronger primary source, closer in time to the events, which gives 1992; the discrepancy is recorded rather than hidden.
The figure of 50% of traffic at the peak: the source is known, the number itself is not verified. It is not in the 1994 EFF guide. The Wikipedia footnote leads to a retrospective article by Deutsch himself, "Archie — a Darwinian development process" (IEEE, 2000). That source is primary, but the full text is behind a paywall and the available abstract does not contain the figure, so it remains an estimate. TechSpot (2024) has a different version of the same legend, half the traffic of all of Canada rather than of Montreal, which looks like distortion in retelling, not independent confirmation.
The text of the original Usenet post: the author and date were found, the message itself was not. The post has been located: Deutsch, comp.archives, 1990-09-11; Wikipedia cites it as the source for the release date. Google Groups did not return the text (429 twice, a JS placeholder once), and Wayback was not reachable from this environment.
What became of Peter Deutsch after Bunyip. Emtage's career is documented in detail (Mediapolis, Internet Hall of Fame), Deutsch's in no source at all; a targeted search found only the reason Emtage handled the promotion, namely that Deutsch had a family.
Deutsch's IEEE article (2000): only the abstract is available. It says the project began in 1987 as a task of connecting McGill to the internet more cheaply, and a search tool was not planned. This is already the third version of the starting year: Deutsch named 1986 in his 1993 interview, Emtage named 1989, and here it is 1987. The date the scripts were written (June 1990) does not change; this is memory drift about the beginning, not a contradiction of the facts. The full text is behind a paywall.