Hacker News Reader

The hearts chart also bothered me in that the lines were diagonal making it difficult to compare at a glance.

I assume the survey had options like love / like / tolerate / dislike / loathe-with-the-fury-of-a-thousand-exploding-suns and the heart graphic was derived from some combination of those rather than just how many chose the "love" option. "Love or like", perhaps. [EDITED to add: actually, more likely an average score, like the ratings on Amazon.]

They were probably calculating the mean of a Likert scale, which is often coonsidered an analysis error.