Archie

~ the last words ~

archie is a name i learned backwards. it turned up in a footnote about where the phrase search engine came from, long after i had already spent years typing into search boxes without wondering who had gone first. the thing it did was over before i owned a modem.

alan emtage was a postgraduate student at McGill University in Montreal, originally from Barbados, and part of his job was knowing what software sat on which public FTP server so other people could go get it. he wrote scripts to do the boring half of that job. log in to a list of anonymous FTP sites, pull down the directory listings, keep everything in one file you could search. released on 10 september 1990, that pile of scripts became Archie, the name being the word archive with the v knocked out.

you did not visit it. you telnetted into an Archie server and typed a fragment of a filename, or you mailed the server and it mailed you back. what came out was a list of hosts and paths. then you opened an FTP session yourself and went and fetched the thing. two steps, both manual, and it was still an enormous improvement over asking someone on a mailing list.

for a few years it was the only way to find anything, and it got used like the only way to find anything. at its height Archie was reckoned to account for half of Montreal's internet traffic, which says as much about how small the internet was then as about how popular Archie had become. in 1992 Emtage and Peter Deutsch formed Bunyip Information Systems with McGill and sold a licensed version, and mirrors went up around the world.

the internet i joined in 1998 had already forgotten all of this. by then finding something meant a web engine reading the words inside pages, and i never once had to know a file's exact name before going to look for it.

~ the record ~

~ what it could not do ~

Archie knew what a file was called and nothing else. no contents, no description, no ranking. if the author had named it readme.1st you were finished. searching meant guessing the naming habits of strangers, which is a skill people genuinely developed and then lost overnight.

the fixes came in layers. Gopher organized documents into menus, Veronica searched those menus, WAIS went after the text inside documents. then the web arrived with a crawler that could follow a link on its own, read the page it landed on, and index the words. against that, an index of filenames was not a smaller product. it was a different century.

~ the actual end ~

nothing shut down. there was no notice and no date. work on Archie stopped through the late 1990s, Bunyip dissolved in 2003, and the servers went off one by one as the admins running them retired or moved on. this is how a protocol dies: not with an announcement, just with the last person who had a reason finally not renewing it.

~ the long tail ~

one server outlived the rest by two decades. archie.icm.edu.pl, at the University of Warsaw's computing center, stayed up for historical reasons long after there was any practical use for it, and it kept answering until around 2023. that machine turned out to matter twice, because the source code was still sitting on it.

in 2024 a computer history project called The Serial Port went looking for a working Archie, found none anywhere, tracked the code down in Warsaw and got Alan Emtage's blessing to bring it back. there is a running Archie again, at archie.serialport.org, indexing what is left of the public FTP world. that is not the service being alive. that is a museum with the lights on, which is more than most of the things on this site got.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »