Posted on 10 Comments

Technorati Suffers Major Data Accuracy Loss

My link count was 262 last night, this morning it’s 220?

This is just plain wrong as I did not have a major linking event 180 days ago. I’m emailing this to David Sifry and Adam Herst and expect a quick resolution of this issue and a restoration of my ranking along with an explanation as to how it occurred.

Technorati is rapidly becoming unstable. It’s losing data or not reading certain blogs while double or triple counting the links of other blogs.

UPDATE: This link was not counted this morning.

UPDATE #2: Adam Herst has contacted me to say they were doing a test of a new feature they are testing and did not know it was going live.

10 thoughts on “Technorati Suffers Major Data Accuracy Loss

  1. I noticed this yesterday morning too Dave. I expected a few bugs with all of T’s rollouts, however, this is over the top.

  2. So they’re testing features on the live system?!

    What, in your opinion, should be the volatility percentage for links?

    mp/m

  3. Google does this all the time. They shouldn’t do it either, but I find it hard for this to be anywhere near the top of the list of things Technorati needs to fix most urgently.

  4. Dave, thanks for pointing out the problem, and sorry it hit your authority so hard. And thanks for the phone call this morning — it was great talking with you about all sorts of issues.

    For everyone else, here’s what happened. This is not really a new feature that we’re testing — rather, we noticed a data quality problem and are addressing it.

    Recently we noticed that several blogs were pinging us using permalinks, rather than the URL of the blog as the ping protocol intends. Our spider was fooled into interpreting these pages as blogs rather than posts. This created duplicate blogs in our database, as well as duplicate links in our indexes. Thus, any blog linked to from one of these posts had its authority improperly inflated.

    We have two things to do to address this — first, stop processing pings like this, and second, cleaning out the duplicate links. We’ve done the first, and have started to do the second.

    Our very rough estimate of the impact of this issue is that 1 in 500 of the new blogs we’ve added is actually a post. That means that, on average, the impact of the cleanup on people’s authorities will be modest.

    Unfortunately for Dave, a disproportionate number of the recent links to his blog fell into this category. So the cleanup wound up lowering his blog’s authority more significantly.

    Again, thanks to Dave for his patience and problem reports.

  5. Your comments do not correlate well with our conversation. Please call me to continue it.

    You should start with all the overcounted blogs in the Technorati 100 before you pick on the little guy next time. Start with – gigaom.com?reactions

    You still owe me at least 10 links over the past three months that have never been counted. I expect action on those.

  6. I seem to recall having reported a similar issue to Technorati support on March 14, 2006, and a followup email to David Sifry on March 27, 2006, after only receiving an automated response. I never got an answer. Hopefully, you’ve figured it out.

    Next problem I’ll just blog about instead of sending an email to support.

  7. I’m pretty sure they don’t count links from blogs that ask Technorati to stop scraping their content, too. So some link loss is due to blogs boycotting Technorati:
    http://www.elainevigneault.com/2007/05/27/blog-tools-how-to-boycott-technorati.html

  8. I notice my authority number has been jumpy too:

    http://techfold.com/2007/05/25/technorati-authority-tweaks/

    Fortunately, mine spiked up realtively speaking (without a linking event), so I’m less inclined to complain… good luck getting that sorted, and good job on running down the details…

  9. […] Godin Joseph Jaffe Darren Rowse Mitch Joel Robert Scoble Elaine Vigneault B.L. Ochman David Dalka Andy Beal Valeria Maltoni Mack Collier Todd Andrlik Technorati Tags: bloggers, Scoble, Seth Godin, […]

  10. […] May David Dalka reports an unusual loss in […]

Leave a Reply

Your email address will not be published. Required fields are marked *