Half a century before the web, a lawyer in Brussels built a machine for answering any question in the world. It ran on index cards, and it worked.
April 2012
In 1895 Paul Otlet and Henri La Fontaine started writing down, on standard cards, every book and article they could find. Not a library — a library holds the things it owns. This was meant to be an index of everything ever published, wherever it happened to sit. By the end of the first year they had four hundred thousand entries. By 1934 there were around fifteen million cards, in rows of wooden drawers in Brussels.
To find anything in that heap they needed an address system, so they built one: the Universal Decimal Classification, grown out of Dewey’s, nine classes and some seventy thousand subdivisions. Its real invention was not the numbering. It was a small set of symbols for joining numbers together — this subject in relation to that one, in this place, in that language, at that date. A card could therefore sit in more than one place at once, and the connection between two subjects became something you could write down. In a system of shelves, that is close to heresy. In a system of links, it is the whole idea.
You could write in with a question
From 1896 the institute ran a service that is hard to read about with a straight face today: send in a question by post or telegram, pay a small fee, and receive an answer compiled from the cards. By 1912 it was handling more than fifteen hundred queries a year. If your question would return more than fifty items, they warned you first.
His unit was never the book. He cut publications down to facts on cards, and let the relations between them do the work.
A search engine, then, with a results limit and a fair-use policy, staffed by people, running on paper, thirty years before the first computer.
It had a building too. The Palais Mondial opened in 1919 in the Cinquantenaire in Brussels, renamed the Mundaneum in 1924: part index, part museum, part headquarters of a fairly utopian idea about knowledge holding the peace together. La Fontaine had won the Nobel Peace Prize in 1913. These were not men with small ambitions.
1934: the screen on the desk
In his Traité de documentation of 1934, Otlet described what should come next. He called it the radiated library and the televised book: workstations — “electric telescopes” — connected by telephone and television lines to the collection. You would put your question to the system from home and see the answer appear on a screen in front of you, several results at a time, with sound alongside the text, and readers able to annotate and exchange what they found.
Read that again with the date in mind. Not a metaphor, not science fiction dressed up: a design, published, by the man who already had the index and had been answering the public’s questions from it for forty years.
Why he is not quite the father of the web
It is tempting to crown him, and the crown does not fit. Tim Berners-Lee’s web won for reasons that were precisely the opposite of Otlet’s design.
Otlet’s system had a centre. One institution, one classification, one building in Brussels, and an editorial staff deciding what an entry meant. The web has no centre and no editor: anybody publishes, nobody approves, and the mess is the price of the reach.
Otlet’s system had to be right. Every item classified correctly, by trained hands, forever. The web assumes everything will be wrong somewhere and routes around it.
What he did have, decades before anyone else, was the part that still causes us trouble: a link that carries meaning. Not “see also”, but this is a consequence of that, this contradicts that, this measures that. It is the ambition behind what Berners-Lee later called the semantic web, and it remains largely unbuilt. Otlet’s mistake was not being wrong. It was being right by hand.
What he understood that we keep forgetting
The document is not the unit. He took books apart into the facts and statements inside them, because that is the size at which a question gets answered. A century later, most organisations still publish the container and leave the reader to dig — a report, a brochure, a file. The unit has always been smaller than the thing we publish.
Structure is a service, not paperwork. The classification existed so a stranger could find something, not so the collection could look tidy. Every menu written in the language of the organisation chart gets this backwards.
People want an answer, not a shelf. The Mundaneum’s most modern feature was that you could ask it something and a human being would reply. That is still what a visitor wants from a website, and still the last thing most of them offer.
How it ended
In 1940 the Germans took the Mundaneum’s rooms for an exhibition of Third Reich art and destroyed a large part of the collection in the process. Otlet died in 1944, out of fashion and out of funds. After the war the story of information science was told with American names in it, and the Belgian with the drawers was left out of it for half a century.
What survived was moved, argued over, and eventually rescued. Since 1998 it has been a museum in Mons, an hour from Brussels: rows of drawers, and the cards still in them.
Worth the trip, if only for the uncomfortable thought on the way home. He had fifteen million cards and the plan for a screen on every desk. We have the screens.
Read the originals
- Alex Wright, Cataloging the World: Paul Otlet and the Birth of the Information Age — Oxford University Press, 2014. The full story, and the only one that treats Otlet as a serious thinker rather than a curiosity.
- “The Internet Before the Internet: Paul Otlet’s Mundaneum” — JSTOR Daily. Twenty minutes, if the book is too much.
- The Mundaneum, Mons — mundaneum.org. What is left of it, an hour from Brussels.
- Peter J. Bogaards, InfoDesign — hand-picked links on information design and architecture since 1997, and an archive worth an afternoon.