DJ Grothe, in the wake of a bit of pattern recognition by Stephanie Zvan that made him look bad enough to evidently cause him to forget that he has a JREF communications director, recently made a claim in his reply (link to Greta’s sur-reply with blockquotes) to Greta’s two questions that certain blogs are under orders to write contrarian and controversialist blog posts in order to drum up hits. He hasn’t yet substantiated that claim with some actual specifics, so while we’re waiting, I thought I’d look at some statistics for the other major component of that claim, that controversial topics drive hit counts.
I have one very salient data point to get out of the way first. Recently, on my post about whether a blog owner has the right to “silence free speech” by moderating their blog, where I’ve been accused of squelching dissent and defending some sort of 1984-like dystopian thoughtcrime witchhuntery, someone who’s commented a few times in the past offered this comment:
When I started reading a few atheist blogs, they seemed to be about skeptical, atheist, freethinking ideas. I was attracted to the lousy canuck blog, because it offered one Canadian’s perspective on these ideas.
I’m dismayed to say this, but I think I’m going to leave many atheist blogs behind for a while, including some former favourites. I’d rather be reading more about philosophy, science, charity, and federal and provincial politics than about specific people, their fights, netiquette, etc. Maybe I’ll check back in 2013, after this has had more time to sort itself out.
I do hope you’ll have decided to tolerate people metaphorically pissing on your rug. That’s inherent in free speech, regardless of the availability of other venues. This venue is the only one guaranteed to have your readers.
If freely expressing my thoughts about how people are treating one another in the community is actively driving certain people to concern troll rather than actually, you know, skeptically examine how they’re treating one another (to the point where he literally ignores everything in the original post to make his last point, showing me an absurd level of disrespect and point-missery), that’s a huge strike against controvery = hits.
But we’re science-lovers, right? Greg Laden built a scatter plot of some of his statistics, and Stephanie Zvan built something similar. Feeling like I’d be letting down my strongest allies here at Freethought Blogs if I didn’t weigh in, I did likewise.
I created two graphs from my highest hitcount and highest view count posts, grabbing the top 15 hit count since I joined Freethought Blogs (excluding one huge outlier — the Santorum abortion post has something like a thousand times more views than the next one down, but only about 200 comments since it went viral). I then went down the list of top comments, and added the next five posts that were not already included in the top 15 viewcounts. Many of them were interspersed, and they certainly didn’t rank identically. I put all these posts in an X/Y scatter plot for views vs comments.
Notice the highest hitcount is a very low comment count thread. That outlier on the right is the three year old girl who figured out gender marketing. It was Pharyngulated, and went a little viral itself, but it seems I was riding on the back of a trend where that video was already going viral by the time I posted it.
My lesson from that: happy-making and troll-proof = hits.
Stephanie pointed out something very true — commenting causes two pageviews, one to view the thread and one for the refresh after your comment was posted. So, with that in mind, I decided to make another graph — a “reception” metric, calculated by subtracting the comments from the views (to get a sense for “unique impressions” that’s probably still wrong, but better in that you’re not doubling the views for each commenter), then dividing that number by the number of comments. Thankfully there were no zero-comment posts in the mix. Because this caused a logarithmic increase for any post that was well viewed but not controversial enough to merit comments — e.g. the the arrow to the knee epidemic video I posted, which is an explanation to a viral meme about Skyrim and is fairly well-indexed by Google — I then square-rooted the number to give a better impression of reception. That way outliers are less outlier-y.
The higher the graph, the more hits-per-comment the post generated. The reception metric is ordered from left being smallest hitcount in my sample to right being largest. You might notice there isn’t really a trend here. I would, personally, expect a gradated slope, probably downward, if controversy drove hits. The fact that this graph is all over the place tells me that there is no such correlation.