Holy shit, I am totally guilty of this.
For those who haven’t yet read the article, the idea is that people interpret “80% of people prefer Pepsi Max to Coke” as “Pepsi Max is 80% yummier than Coke”, when in reality most of that 80% only slightly prefers Pepsi.
Basically a strong difference in proportion of people who prefer one option to another does not necessarily imply a strong difference in the average opinion between the two.