Wikipedia Launching a New Search Engine – Wikiasari

Quick Pre-Christmas post – Frank Watson at Search Engine Watch reports that Wikipedia and Amazon are teaming up to launch a search engine early next year. The story originally broke in the London Times where Wikipedia founder Jimmy Wales was quoted as saying “Google is very good at many types of search, but in many instances it produces nothing but spam and useless crap.”

Later information now indicates that Amazon does not have direct involvement, and that this is a Wikipedia owned project.  Evidently, social human input will play a major factor in determining search results.  In addition, indications are that the focus of the search engine will be on producing quality results, even if it means sacrificing depth.

It will be interesting to see how this search engine does.  People will want depth of results.  I am also not sure that community based human input will do a good enough job in producing good results.  Certainly it will be hard to police this as the index does grow larger.

My belief is that it is possible to do a better job at search, and someday someone will do it. Success will depend on a well built mixture of algorithmic processing power and human review. You can only go so far with pure algorithms, and human review is not at all scalable.

But perhaps it’s possible to build an infrastructure that flags things for human review – either through algorithmic detection, or a sufficient number of complaints from the community. There is no question such an approach would be expensive in comparison the (near) purely algorithmic search.

This might make it difficult for the current search engines to pursue. They are all public companies with operating profit goals that really can’t be changed without blowing a hole into their current valuations. But it would be fun to see someone try it.


  1. Eric Enge says

    Hi Anghus – There is an article in TechCrunch that explains it. You can see it here:

    Basically, the top 3 results will be Wikipedia content, and the rest will be sites that are linked to from Wikipedia. By this means, the expectation is that the community will have decided that the sites are reputable.

    Kind of a neat idea, but I would be surprised if it does that well.

Leave a Reply

Your email address will not be published. Required fields are marked *