Sphere's Blog Search

Sphere launched a month ago, but i only just got to really check it out last week when i sat down with tony conrad, ceo and founder (Tony was previously involved in Oddpost and is an advisor for Automattic). The technical team is the crew that brought us Waypath.com – an early blog search engine.

Sphere is an impressive blog search engine and one that is sure to rise in traffic. In a very short time it has already reached feedster’s traffic levels and surpassed Pubsub (they have a while to go before reaching Technorati).


As they build their index they are focusing on avoiding splogs and pulling in quality (reminds me of techmeme.com’s approach). Their index allows for a search to run over a 4 month period and they have a very useful UI element that allows you to see the post distribution for your query. You can use this tool to focus your search on a custom date range (the default for a search is a week).


Their most impressive feature is SphereIt. Quite often I’ll read something and be curious what people are saying about it, but no one has commented on that specific article. SphereIt does realtime textual analysis to find bloggers discussing the same topics. (Add it to your browser from here.) The value of this is becoming apparent to other companies, and just this week Time implemented a SphereIt button on their Analysis & Opinion pieces. You can see the difference in results when looking at a piece like Time’s Are Stepparents Real Parents?. On Sphere I get 87 results (and I doubt this piece will join them, because even though it links to the piece, it doesn’t discuss its content), while on Technorati I get only 3. (I’m assuming that this post will increment the results – and it should based on the link-method of searching, but as someone looking for more information or opinion on the original article I wouldn’t find it to be an appropriate search result.)


They also have featured blogs. To determine them they seem to analyze post content, posting frequency, the link love, and the post length, and they use this to calculate the blogs that are most likely to provide quality posts on that topic. Running this through the ropes with china, web 2.0, knitting, and Seattle bring up good blog selections. More advanced queries do not succeed and there are definitely areas where there are gaps. This feature is labeled beta and will undoubtedly improve as they grow their index.

Sphere has really moved blog search forward with a greater focus on the blog’s content than on its linklove. I especially think that there’s a lot of future in SphereIt being used to create browse experiences.