Wednesday, February 2, 2011

Follower Number is Poor Predictor of Blog Popularity

I have sometimes wondered about whether older blogs with large, established reader bases gain new readers faster than newer blogs. This is an important question because it gets to the heart of the issue about whether current post quality is more important than the momentum provided by historical post quality. Ideally, a good blog will attract readers based simply on the quantity and quality of its latest posts.

To make a long story short, the good news is that initial follower number is a poor predictor of how many followers will be gained over a subsequent period of time. This implies that a blog's current popularity is not determined by how many readers it has amassed in the past; readers will be gained based primarily on how good a blog is, not how big a blog is.

I came to this conclusion by looking at a dataset of 216 old school roleplaying game blogs. I plotted follower numbers as of 30 Sep 2010 against gain-of-follower numbers over the period between 30 Sep 2010 and 31 Jan 2011. While a simple linear regression shows a weak trend for larger blogs to gain followers faster, the r-squared value for the trend is very low (0.3575). If we remove the juggernaut blogs GROGNARDIA and Playing D&D With Porn Stars the r-squared value drops down to 0.2087. An r-squared value of 1 is a strong association; the low values we see here suggest a very very weak association. This is good news! Keep on blogging!

Note that if you don't look at the (weak) trendline, the plot is basically a cloud...