The data is misleading. Obviously there’ll be more requests for domains with low TTL because those domains aren’t cached. The overall conclusion and points of the blog post may be correct, but the presented data doesn’t say anything.