The chart should be somewhat descriptive in the correlation, and usually processing in intensity. The single death count having two different levels of funny isn't really explained by the chart, which isn't a real statistical analysis which might have odd data points like that, but instead is an extra joke.
As a joke, it doesn't portray itself well if the axies aren't swapped. When swapped, the joke is pretty obvious and understandable.