Site Meter

January 31, 2008

Rating Burner's Alexander Fedorov Explains Site, Hidden Features

Yesterday, we uncovered Rating Burner, a new site that ranks popular blogs by their total subscribers, as calculated by FeedBurner. The site, previously unknown to the blogosphere, offered little behind its story, and little clue to its owner, though we did find Alexander Fedorov as the registrant of ratingburner.com. Today, Alexander emerged, and sent me a few e-mails, which further explained the site's origin and purpose, uncovering an interesting approach to discovering new sites and RSS feeds, as well as Web advertising opportunities.

First off, the site is more than an Internet egotist's dream. Rating Burner has its own Web crawler, aimed to discover new FeedBurner feeds and subscriber counts. As Fedorov wrote today, "The rating is growing all by itself. As soon as someone links to your blog, sooner or later it will be found by Rating Burner's bot."

And the bot had to start somewhere. Fedorov kicked off the Web crawling using John Battelle's Searchblog at http://battellemedia.com/. You can see this by using the same type of URL hacking popularized with ReadBurner just yesterday. Try the URL http://www.ratingburner.com/?order=1&dir=a and find the exact order of the robot's work. Hovering your mouse over the "1" in the "N" column shows the site was indexed by January 18th at 11:18 a.m. 4 minutes later, it was off to the Black Hat SEO Diary. 3 minutes later, the robot found TechCrunch, and it was off to the races. The robot eventually found louisgray.com on January 20th after noon, making me #216 in the list, which now has 368 individual sites.

Even more interesting, Rating Burner shows a detailed map of how each of these blogs link to one another. (Download the detailed graphic) The sites with the most links their way have the most arrows, and those sites with the most FeedBurner subscribers have the widest arrows to and from their site. As the below graphic shows, his robot correctly noted I link out to Mashable and WinExtra links to me. The map was generated by GraphViz, an open source tool.


My position in Rating Burner's map.

The robot spawned by Rating Burner has specialized intelligence to discover ad networks, and enables site owners to display going rates. As Alexander explains, the "crawler scans all blogs for signatures of some well known sponsors/advertisers and lists this information. It is possible to publicize prices for banner ads using special html comment . Crawler will find this info and publish price in the rating."

There are also a number of hidden tricks in the database.
To find out the total number of comments in the most-recent post on any included blog, you can sort with the below URL:
http://www.ratingburner.com/?order=17

To sort included sites by the most recent post, it's a different command:
http://www.ratingburner.com/?order=16

... and by PageRank:
http://www.ratingburner.com/?order=12
If you recall, Rating Burner also hinted at further categorization, including tags for SEO and gadgets. Alexander promises there are many more to come, automatically generated, again by the robot and the intelligent database:
He writes, "...categorization is completely automated. I have several samples of texts on special subjects like "gadgets" and then all blogs (5 last posts) are tested againts this texts using bayesian filter and shingles, and if the similarity is close enough, this blog will be listed under this category. I just need to add more categories and collect more sample texts."
Rating Burner is still in its infancy, even after yesterday's debut, without question. But there are some interesting tricks lurking in the database. Want to find recent posts with lots of comments that might be "hot"? Try this link: http://www.ratingburner.com/?order=4&filter=category%3ATOP.

And the blogosphere has noticed. In addition to my post, which hit TechMeme, and Mashable's coverage this morning, the site's debut was included in Danny Sullivan's Search Engine Land wrap-up for the day, as well as Marketing Pilgrim.