Why Blogs Matter
I had hoped to write up some of the insights below for the Weblogging Ecosystem Workshop at WWW04. Instead, here's the quick version with an eye towards implementation.Using the Google API, I've been tracking the position of Blogs vs Traditional Media on a range of different queries for a number of months. The visualizations below, crafted with DHTML, portray a set of related queries.
For "google print" and "google print beta", the blogdom rules search results. In the image, green squares are blogs and traditional media outlets are red with yellow squares in between (often community bbs style). The 1st result is at center and subsquent results are layed out both radially and by radius with the 10th result at 12:00 after the first concentric color transition. This image is just of the top 40, the picture for the top 200 looks similar.

The blogdom has a notable but lesser presence in data for Patriot Act (sic) related queries. Traditional media dominate the top results:

Perhaps it's the case that widespread blogging of a top has a synergistic effect on the page rank of top blogs? My investigations suggest this is the case, though for Lord of the Rings, e-commerce sites tended to outperform blogs, garnering more top 20 positions.
So, to wrap up from the title, blogs matter because users encounter them in high ranking positions in web searches. This is good.
Finally, a note on the implementation. I know color coding shouldn't be used exclusively but getting scaling on non rectangular gif shapes is troublesome. SVG or Flash would solve the problem, but DHTML is so much quicker to develop (for me, and not that you can't do both in Mozilla). An interesting aspect to this visualization is a tooltip along with a opacity filter that features two levels: massive fade for un-related query results, and a partial fade for results from the same query. Browsers handle walking 100s of result datapoints in the DOM and applying the opacity filter with a minimum of latency!
|||
Posted at 1:26, Published in: Blogging