Hjalmar Gislason is about to launch the first directly human edited search engine. No machine algorithms. No link moderators. Any person can suggest a search result and the keywords(tags) by bookmarking them at Spurl. Then, they index the pages. The more people that suggest a certain web page with a certain keyword, the more relevant it is. The better the contributor's reputation (think slashdot), the more relevant the result.
In Hjalmar's words, here is what they are doing:
> What we are doing is that we are
> using the massive amount of human information that we now have in the
> Spurl.net database. There are roughly 1.3 million pages there with
> some sort of annotation (category, tags, description, snips from the
> text, custom titles) and more. We give this information extra weight
> in the search while we also search "machine information" (i.e. the
> page text, title, url and other things that any bot could get as
> well). Finally, we use the number of users that have spurled a page as
> an indication of it's "importance".
Duh!
People stopped innovating the concept of the online directory when the OpenDirectory died and when Yahoo 1) decided search was more important and 2) decided the directory was too expensive to maintain with all of their PAID moderators.
Duh!
Well, the re-emergence of the directory is here. What most people are calling social bookmarking is really just a better human edited directory of websites. But, as I pointed out before, searching del.icio.us, furl and spurl isn't that easy. That will all change in a week or so. When spurl launches their website.
Why am I excited? Think about this...
How many websites do you blog? How many websites do you bookmark? If you are a blogger, I bet your answer is that you bookmark more sites than you get to blog. You may intend to blog them, but you don't get a round tuit. So, therefore, you are creating more metadata that is stored at del.icio.us, spurl or furl, than you are making for google, yahoo or ask.
So, isn't it natural that del.icio.us, furl and spurl will have better search results? (eventually)
I've been given a sneak peak of what spurl is about to launch. And I can vouch that the next generation of search is arriving.
With Spurl's search engine, there won't be 1 million search results for any keyword (yet), but the two pages that a given query returns, will be very relevant. There won't be search results for too many non-technical searches, but it'll be link rich for us tech-geeks. Unfortunately, I cannott point people to the demo. But, if you want a sneak peak, leave a comment here for him.
Pete & Hjalmar,
I'd love to see what's happening with spurl search.
Posted by: Shimon Rura | February 20, 2005 at 11:22 PM