Technorati recently introduced a “filter by authority” feature to their blog search. The idea, from CEO Dave Sifry’s blog, is to “[make] it easy to refine a search and look for either a wider array of thoughts and opinions, or to narrow the search to only bloggers that have lots of other people linking to them. This gives you the power to decide how much filtering you want.”
Scoble, Steve Reubel, and TechCrunch all offer opinions. It’s worth reading the comments for each entry.
I don’t like the new feature. The problem is that Technorati’s algorithm for measuring authority is too simple–they look only at inbound links (the more links, the more “authoritative” a blog is). Just because a blog has many inbound links doesn’t mean it necessarily has significant influence or control over others. That said, it isn’t immediately obvious how to do a better job? What additional factors should be considered? Can these factors be evaluated consistently? Will the process of evaluation scale to cover the entire internet?
I’m reminded of an article in this month’s Atlantic magazine about online dating (sub required, unfortunately). If you think determining a blog’s authority is difficult, imagine how hard it must be for a computer program to accurately assess compatibility between two people. Difficulty notwithstanding, eHarmony (among others), seems to be having success: according to a Harris Interactive poll, between September 2004 and September 2005 eHarmony “facilitated the marriages of more than 33,000 members–an average of more than 46 per day.” Another study suggested that people who met via eHarmony “report more marital satisfaction than those met by other means.”
How does eHarmony assess compatibility? It isn’t easy: the company has developed a 436-question personality survey (it takes a while to complete). They plan to hire 20 to 25 “top relationship researchers” away from acadamia to develop additional tools.
Clearly, this isn’t a completely fair comparision. eHarmony can successfully ask clients to complete a long survey, and then eHarmony can take its time to analyze data before making a recommendation. Technorati has no such luxury: like other search engines, users demand immediate, accurate results. That said, I think Technorati’s challenge–enabling a machine to correctly assess a qualitative concept like authority–is essential. The quantity of available content continues to increase at a seemingly exponential rate. It just isn’t possible for any single person to digest all–or even most–reports on any topic (after performing a Google search, how many times do you click past even the second page of results?). Better machine intelligence–especially related to filtering and processing qualitative, unstructured information–will likely be a very valuable tool for just about everyone. Perhaps Technorati can take a lesson from eHarmony and invest in developing a more sophisticated solution.
