Quaero

~ the goodbye ~

Two heads of state stood up in April 2005 and announced a search engine. Jacques Chirac and Gerhard Schroder, at a French-German ministerial conference, said Europe was going to build its own answer to Google. Eight and a half years later the program closed and not one person outside the consortium had ever run a search on it, for the simple reason that there was nothing running to search with.

The name is Latin for I seek, which tells you the register the whole thing was pitched in. Chirac brought it up again in August 2005 at the launch of the new state agency created to fund exactly this kind of project, and again in his New Year address in January 2006, where it got real public attention. By April 2006 it was one of the agency's first selected programs. This was a serious commitment of political capital to a piece of software that did not exist.

What it actually was, underneath the speeches, was a research and development consortium. Thomson led it, with France Telecom and Exalead alongside. Behind those sat a long list of public institutions, running from INRIA and IRCAM out as far as the national library and the French defense procurement directorate. The stated goal was multimedia and multilingual indexing rather than a web search box, which is a real research agenda and a bad match for the sentence everybody kept repeating about competing with Google.

There is no honest distance to write here, because there is no distance. Nobody used Quaero, in Turkey or in France. The impulse behind it is legible enough from where I sit: this country had its own portals and its own search attempts in the same years, built by people who could see that the index of your own language was going to be somebody else's asset. What is unusual about Quaero is not the wish. It is that the wish was announced by a president before anyone had written the crawler.

The program ended on 31 December 2013, on schedule, the way a funded research program ends rather than the way a product dies.

~ the chronicle ~

~ what it was supposed to do ~

The technical idea was better than its reputation. Quaero was not aiming at keyword search over text, which was already solved and already owned. It was aiming at search over audio and video: transcribe the speech and index the transcript, then let people find a clip by what happens inside it. LTU Technologies supplied image work that let you search with a picture instead of words, matching shapes and colors against an archive.

Search by image and search by sound both arrived, roughly a decade later, inside products built by the company Quaero was aimed at. That does not make the research wrong. It makes the framing wrong. A ten-year multimedia indexing effort with a national library and a broadcast archive attached to it is a defensible use of public money right up until somebody puts it on television as the thing that will beat Google.

~ the germans leave ~

On 18 December 2006, at an IT summit in Potsdam, a state secretary in the German economics ministry announced that Germany was doing its own project instead. It was called Theseus and it pointed at semantic web and knowledge management rather than at video.

The split was over what search meant. German engineers wanted a text engine and French engineers wanted a multimedia one, and there was a second complaint underneath that which is more interesting: some of the Germans thought the project had turned into an anti-Google exercise rather than something with its own reasons to exist. The Economist reported in January 2007 that Quaero had been scrapped, which was wrong on the facts and right about the mood. The French half carried on alone for another seven years.

~ ninety-nine million euros ~

The European Commission cleared 99 million euros of French state aid for the program on 11 March 2008, three years after the announcement and fifteen months after Germany had gone. State aid approval is a slow instrument aimed at factories and airlines, and running a search project through it produces a timeline that has nothing to do with how fast the thing you are chasing is moving.

The critics said this at the time and were not subtle about it. Le Canard enchaine noted that the budget was a rounding error next to what Microsoft and Google were each spending, and the Financial Times ran a quote calling the whole thing misguided and unnecessary nationalism. In IEEE Spectrum, Nick Tredennick wrote that going at Google with well-funded entrepreneurs would be foolish and that attempting it with a multigovernment collaboration was beyond description. The article was titled with a question about the Latin for delusional.

~ what came out of it ~

Research programs do not leave a domain that stops resolving, they leave papers and hires. Quaero produced both, across speech transcription and image retrieval, spread through a couple of dozen labs and companies that mostly still exist. Exalead was bought by Dassault in 2010 while the program was still running. Theseus ran its own course in Germany and ended the same way.

What did not survive is the premise. The idea that a continent could decide to have its own search engine, fund it like a rail project and get one at the end, was tested here as directly as anybody has ever tested it. The answer was no, and the reason was not the money. It was that a search engine is a service that has to be running in front of people, getting worse and better every week, and Quaero was a program with milestones.

@}-->--

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

« previous  ·  random  ·  next »