Summary

  • FOLDOC’s 1994 introduction encouraged online searches so frequently requested missing definitions could be identified and added.
  • The project’s FAQ explains how failed internal links feed a missing-term list, while the editor distinguishes gaps, misspellings and misleading matches.

The broken link was taking attendance.

At FOLDOC, the Free On-line Dictionary of Computing, following an internal reference could lead to a term with no definition. The site’s explanatory FAQ gives a disarmingly practical reason for keeping such links: clicks told the editor which absent entries people wanted. A dead end could earn its keep.

That logic was already visible in a 1994 introduction preserved by the project. It encouraged readers to search the dictionary on its web server, so frequently requested missing terms could be identified and supplied with definitions. A file called Missing collected unsuccessful lookups. Readers were invited to inspect it and contribute an answer.

The dictionary had found a way to turn coming up empty into an item of work.

From a few definitions to a server

Editor Denis Howe’s first-person history begins with about half a dozen definitions copied from a magazine for personal use in 1985. After arriving at Imperial in 1990, he discovered anonymous FTP and other collections of computing definitions. He combined material, including the Hacker Jargon File, with his own collection.

The resulting file went on an FTP server on his desktop machine, advertised through Usenet. Gopher then let readers search the dictionary without fetching the whole thing. Around March 1994, Howe adopted the web, gaining cross-references, feedback forms and links to resources elsewhere.

Each step changed the exchange. A file could travel to a reader. A query sent to a server could tell the editor what that reader had come looking for. The 1994 request to use the web version made that return journey explicit.

Please leave the ordinary dead ends alone

The FAQ explains the routine in more detail. Internal cross-references act like typed searches. When either fails, the requested term joins the list of frequently sought missing entries. Howe can then decide whether a link was mistyped or a definition deserves to be added.

His explanation for choosing the popular requests is cheerfully finite: there are many gaps, and he has limited time. The system helps him spend that time where readers are asking for it.

Accordingly, the FAQ asks people to spare him routine reports that an internal reference finds nothing, except for obvious misspellings. The click has already lodged the request. Sending another message about the same empty cupboard gives the person stocking it another message to read.

Other failures need different treatment. A broken link to an external site should be reported, preferably with a better destination. A search that lands on an unrelated entry also deserves a report: the matching system can return a partial match when the exact term is absent. Finding the wrong answer creates a different repair job from finding no answer.

Somebody still has to write the entry

The missing list gave contributors a useful place to start. Howe’s account says submissions were usually edited extensively before inclusion. The reader supplied a need, another reader might supply knowledge, and the editor decided what became part of the dictionary.

It was spare-time work, carried on alongside other obligations. A list of demands could help choose the next task; the task still needed doing.

FOLDOC’s small bargain was unusually candid. If the dictionary failed you, that failure could help set its agenda. Even its dead ends had been assigned office duties.