I am delighted every time that there is a press release which announces a new star in the searchengine-sky. If the mentioned searchengine is German, my glee becomes immeasurable. The time for this has come again, WeFind (“the new searchengine for the web 2.0”) has started. To cut a long story short, the quality of the results has not caused me to go and download the WeFind-background and install it on every PC. This being the case, I do not really want to keep writing about WeFind itself, but the fundamental problems of metasearchengines.

Metasearchengines work by directing the query to a number of other searchengines, gathering the results and processing the results in a potentially meaningful way. Metasearchengines were rather popular in the Internet's onset because each of the numerous small searchengines was only crawling a small portion of the web and through the help of metasearchengines you were able to get at least partly fitting results. Today, metasearchegines are nearly without relevance anymore – on the one hand, Google's and Yahoo's indexes are large enough, which means that usually no important site will be left out and on the other hand, the quality of the Metasearchengines is unable to keep up.

For the most part, this is caused by the fact that while Google, for example, has 200 different signals to construct a nice ranking, metasearchengines have to make do with noticeably less. Like others, WeFind acknowledges that they are usually using an OpenSearch-interface from their “searchpartners” – and these are not releasing any more informations than are usually seen through the browser. Sometimes we have the chance to see how hard it is to meaningfully weight results form more than one source by Google itself. The Universal-Search also queries more than one index (web, pictures, news, videos, etc) and Google is faced with the problem of piecing together the different rankings – which often goes awry, even though Google has access to all the ranking-factors for every index.