Commonalities: The followup

A couple of days ago, I posted “Commonalities and generalizations“–offering three lists of liblogs and asking what each list had in common.

I love the responses (see the comments on that post)–and one of them (Steve Lawson), either through deep memory or, ahem, doing a little research, had the right answer: Each list consisted of blogs using one blogging platform (combining MovableType and TypePad into a single software platform).

The first list is blogs using Blogger, most (but not all) hosted at The second list is blogs using MovableType or hosted at The third list is blogs using WordPress, some of them hosted at

Generalizations? Just this: For at least the first and third list, it wouldn’t be all that improbable to do a reasonably broad sampling of liblogs, enough to be statistically valid, and conclude that the software in question was used by the vast majority of liblogs. (It would be a little tougher with TypePad/MovableType, but certainly not impossible.)

And it would be wrong.

Here are the actual figures for 519 of the 521 blogs in the project I’m currently working on–all of them blogs by library people, all with blogs still visible in September 2009 and having had at least three posts in either March-May 2007, March-May 2008, or March-May 2009, all primarily in English and, a serious limiting factor deliberately added to reduce the size of the universe, all of them having a Google Page Rank of 4 or higher in either September 2008 or sometime in Spring 2009. (I was hoping that set of limits would yield around 400 blogs, making the research a lot easier. In practice, it yielded 521…which is still better than the 650-700 I might have ended up with using last year’s rules.)

Program Blogs Percentage
WordPress 245 47.2%
Blogger 190 36.6%
TypePad/MovableType 48 9.2%
Other 24 4.6%
Drupal 7 1.3%
LiveJournal 5 1.0%

Looking back at the 2007-2008 study, which included 607 blogs (but a different sample–with 127 removed and 41 added), the percentage using WordPress has jumped from 37.9% to 47.2% while the percentage using TypePad/MovableType has slightly increased from 8.8% to 9.2%–and the Blogger percentage is unchanged, at 36.6%. This may mean that a lot of “other” blogs changed platforms, it may mean I did a better job of identifying WordPress blogs (I don’t remember looking at source last year), or it may mean nothing at all.

At this point, if this set of 521 blogs (two disappeared between early September and September 30, when I did this scan) is in fact representative of more visible, more active liblogs as a whole–an assertion I am not willing to make–then:

  • You’d be wrong to say that any program is used by “most” libloggers.
  • You’d be awfully close in the case of WordPress, and it’s certainly a strong plurality.

Angel’s comment on the other post intrigued me, so I looked at the subsets of these blogs with Google Page Rank of 5, 6, and 7. (There’s only one with GPR 8 and none with GPR 9). There’s no real correlation–in fact, the percentage of higher-profile blogs using WordPress is slightly lower (44%).

But I can understand how he could arrive at this assumption–because I grabbed each list in descending GPR order. Since there are more WordPress blogs than either Blogger or TypePad/MovableType, the sample for WordPress is skewed toward higher-profile blogs.

Anyway, there’s one table from But Still They Blog: The Liblog Landscape 2007-2009, a work very much in progress. Incidentally, while I use GPR4 as a cutoff, I don’t mention the GPR for any blog within the book; this time, I won’t even have an overall chart of blogs-by-GPR. I think a couple of the very high GPRs are flukes, and I know that moving a blog can wipe out your GPR for some time… It’s not a great metric, but right now it’s the only one available that’s easy enough to use.

2 Responses to “Commonalities: The followup”

  1. Blake says:

    ” I know that moving a blog can wipe out your GPR for some time”

    Moving? Moving URL? Moving to a new server? Moving to a new CMS?

  2. walt says:

    Blake: Changing URLs can (and, I think, usually does) wipe out the GPR for a site, not surprisingly. As far as I know, nothing else has a drastic effect, but I’m not a GPR expert, just an observer.