Sphere’s new search impresses, but not there yet

Tony Conrad, Sphere’s CEO, gave me a tour of Sphere.com today, a new blog search engine. He started out by saying that Mary Hodder is on their advisory board. That’s interesting because Mary has done some of the best thinking of the gestures that search engines could track.

They showed me what the engine is good for, what it doesn’t do, and where they are going. I’m intrigued, but don’t yet see that it’ll hold off Google or Technorati. But they are on the right track and might get there first. I’m holding out hope.

See, there are a few blog search types that I think most people will want to do:

1) Find the “big fish” in a specific blog community. This type of person is looking to join a new community and find some good blogs to read. Search a blog search engine for, say, “Scrapbooking,” and see what comes up. Most of the time it’s noise because most blog search engines just show you the latest 10 posts that were made with the word “Scrapbooking” on them. Watch, this post will show up on Feedster and Technorati and IceRocket within a few minutes. What you really want, though, isn’t a blog that rarely writes about Scrapbooking. No, you want a blog that ONLY writes about Scrapbooking. That’s where Mary’s social gestures come in. If a search engine was tracking all those gestures, it could find the most relevant results. Now, Sphere does do better than other engines, because it brings back blogs based on relevancy, rather than just who just published. The problem is they are only tracking links and title tags so far. Yes, it’s better than the other blog searches, but it’s not pure. I am still getting spam on some of my queries that I tried and it still doesn’t make you confident that you really are finding the “big fish” in the scrapbooking community.

2) Link search. It’s not intuitive, but one way we find blogs is if we find one that we like, it’s usually good to see who else is linking to it (and who that one blog is linking to). Sphere doesn’t do this kind of search, but most of the others do.

3) Time-based RSS query. This is where you want to see every blog that writes a certain word or phrase. For instance, I search on a number of engines for the word “Microsoft.” If you write “Microsoft” on your blog, I’ll probably see it because PubSub, Technorati, Feedster, IceRocket will watch for that, and will spit it into an RSS feed that I subscribe to. I couldn’t test the quality of this on Sphere because the pre-release version they gave me access to didn’t yet generate RSS feeds (Tony said it will by the time it’s released). What I did like is that I can do the search for “Microsoft” on Sphere two ways: once ranked by time (latest post at the top) and other ranked by relevance (what they sense is the most important post is at top). Oh, I just saw that Feedster now does the same thing. Hmm, did they just add that?

Anyway, the race is definitely on for who can do the best blog search engine. Sphere is definitely a step in the right direction. But the path ahead of us is long and we’re only partway there.