in your face with sphere beta

A new blogging search engine is in town, but unfortunately this was the first time I got my invite… six days after good ol’ TDavid got his over at MakesYouGoHmm. I was definitely curious as to if things were fixed since I read the “First Look” and if things were going better.
Unfortunately for me, they weren’t… and it’s a long ways before it will boggle my mind, but definitely a new approach to blog-searching.

Of course LUX was nowhere to be found… or so I thought. After multiple entries into the engine, an entry from November 3rd made it into the fray. Titled “GimpShop 2.2.8 for Windows”. But what’s amusing is that there is nary a link that was scraped from that which resulted in LUX’s post. You would think that PlasticBugs or Make wrote those words that were scraped.
The LUX stats were off also just as TDavid had mentioned:
Average posts/week: 1
Average words/post: 473
Average links/post: 6
LUX posts on average 3-10 a day, at the minimum one post that is a scheduled tip or trick. Average words are probably close although I have never paid attention to it. Average links per post on average range around four although I suppose it could be six.
As a designer, I liked the color choices and the bold interface. It was simple but elegant design. The only complaint I had was with profiles where sometimes it would turn the words into a transparent block. Then you would get a mixing of the characters from the foreground with the background. Ick. Nasty bugs.
It was interesting that searching for the top bloggers turned up very few postings. Scripting News, Makes You Go Hmm, and even Scobleizer had very little when I would assume they would have had hundreds if not thousands of entries. It seems that the algorithm for the search cannot decipher substrings and account for substrings to be a lesser point value than an actual string match. Currently:
LUX.ET.UMBRA would have returned LUX.ET.UMBRA
A search for the words lux or et or umbra, would not give you LUX.ET.UMBRA though, or even if you wrote out “lux et umbra” without matching the periods.
Features I wish it had…
What I wouldn’t mind seeing is sphere being able to scrape a site and determine from the content if there is a likelihood that it is a splog versus a real blog. If they had a way to make that determination, they would definitely become a bigger player in the crowded blog search arena.
Definitely a unique style of grabbing the information (via scrapes) but there is no single “wow” effect that makes me want to give up what I use now for sphere. Perhaps when some other things come out of this closed beta, there will be some better and precise data to be received from this. Until then, they’ll have some growing up to do as the infant of the group.