The flaw in Technorati's popularity calculations

This is slightly off-topic, but if you look right now at Technorati's list of popular books you'll see that Make Easy Money with Google is first on the list, even beating John Battelle's The Search, an excellent book about the origins of Google and the way search engines have redefined the Internet.

Normally you'd think I'd be thrilled with this, but all it's done is proved to me that Technorati's way of calculating a book's popularity is all wrong. I mean, look at the Amazon rankings for those two books: Battelle's consistently stays below the 500 mark (meaning it's one of the top 500 sellers on Amazon) and mine hovers anywhere between 5,000 and 100,000. (For the curious, see this discussion about what Amazon rankings mean.) Battelle's book is a much broader topic than mine that will interest all kinds of people. Mine's about how to build websites and blogs and make money from them using AdSense. (And, apparently, the title turns some people off because they think it sounds like a scam… but if you read this blog you know it's not…) His book should always be above mine except perhaps in narrowly-focused lists.

The problem is that Technorati is counting the actual links to books on Amazon.com within blog entries as indications of that book's popularity. But the problem is that they count each entry in a blog, not whether a blog mentions the book. So if you have 20 blog entries in a blog that reference the book, that counts as 20 links in Technorati's popularity measurement scale.

The reason my book ranks so highly right now is that last night I was playing around with my blog software (blojsom), doing a little programming to automatically build archives of all my blog postings, which as of today will number 152 since the start of this site back in May. The first archive, sorted by date in reverse chronological order, is now up. To do this I briefly had to turn off the limiting feature of the blog software that limits the number of postings shown on the main blog page to 20 entries. So all 151 entries (this was last night, remember) were showing up. Unfortunately, Technorati's crawler showed up at that time and found all those entries and thought they were all new. And many of them, especially the most recent ones, have a link to the book.

So, thanks to this unfortunate bit of timing, Technorati now thinks that there are over 50 new links to my book within the last 48 hours. Don't get me wrong, I don't mind the free publicity, but it's convinced me that it's not a representative list at all if it can be so easily manipulated.

What Technorati should be doing is counting the number of blogs that mention a book, not the number of blog entries that mention it. This would be a more accurate measure. They already do this to some degree. If you look at the Technorati rankings for individual blogs, you'll see that they list both the number of blog entries linking to a particular blog as well as the number of sites linking to that blog, and the rankings are apparently based on some combination of both numbers. They could do something similar when counting the links to Amazon.

OK, tomorrow we get back on topic and continue with the AdSense case study I'm doing. Also, tomorrow's the last day to subscribe to my announcement list to get your chance at winning a free signed copy of my book when I draw a name from the subscriber list on November 1. If you join after tomorrow, you'll have to wait another month for you next chance.

Eric Giguere is the author of Make Easy Money with Google, a real (printed!) introductory AdSense book for non-technical people, available at all fine bookstores. Be sure to download the free sample chapter for more information about the book.

Comments