Posted January 12, 2010 2:08 pm by with 10 comments

Tweet about this on TwitterShare on LinkedInShare on Google+Share on FacebookBuffer this page

Two weeks ago, it looked like Google might have been . . . well, fibbing. Back in June, the search engine announced that they would no longer redistribute “link juice” around nofollowed links, effectively deprecating the practice of PageRank sculpting. But late last month, it looked like the technique was still effective when SEOmoz published some research to that effect.

Not so much, it turns out. Today, Mozzer Danny Dover revisits the tests he used in the original post and shows that the results were, in fact, inconclusive.

The short reason is that the test didn’t include enough data to be statistically significant. He’d have to repeat the test with 168 domains (instead of the 20 he used) to assure that the results were meaningful and not merely a fluke.

Apparently, there’s been some backlash against Danny and SEOmoz for their innaccurate information. Danny didn’t realize his error until Darren Slatten corrected him in the comments two days after the post had gone live. Naturally, it was too late to unring the bell—and a number of people were very upset that SEOmoz would post something misleading or outright wrong.

In his correction post, Danny points out that this is just another example of the power of the Internet. I think it’s good that Danny did more than just update the old post, also publishing a new post to correct the misinformation (and apologize). Two weeks after the correction is fairly quick to reanalyze the data, though it’s possible he could have acted even faster. This is also a microstudy on reputation management and transparency, something we’re a little passionate about around here.

As a reminder, back in August, Matt Cutts told SEOmoz that it’d be okay to keep nofollow links in place, though when building new sites or redesignin old ones, PageRank sculpting shouldn’t be a consideration.

What do you think? Did everyone handle this situation well?