My Three Tiny Startups: HackerEngine, Speaking for Hackers, and Vim University

"After enabling the survey on a popular post, I had almost 80 responses after just a day. 14% of users preferred screencasts, 41% preferred articles, and 45% thought both were good."

If this popular post was an article, it is not surprising that majority or the survey takers preferred articles since they were already reading an article. This is called sampling bias. How would the result look like if the survey was conducted on a page with a screencast. In that case people might favor screencasts.