If the data is imperfect why isn't the visual representation of that data imperfect as well?

Animated GIF by Jared Wilber

RoughJS is a Javascript library that generates charts that look hand-drawn. Nadieh Bremer used it in her Why Do Cats & Dogs ...?, and I just saw a tweet about it by Javor Jordacevic. It reminded me of Mona Chalabi's charts, Elijah Meeks's Semiotic, and this paper about sketchy rendering.

Given my interest in methods of depicting uncertainty, I've often wondered why this type of rough, fuzzy style isn't explored more often. I have the hunch that graphics based on conventional geometric shapes with sharp boundaries such as dots, bubbles, lines, or bars may sometimes give readers a misleading impression of complete precision, when the truth is that most point estimates we represent in our visualizations are surrounded by clouds of uncertainty. Perhaps charts that look imperfect may suggest that the data they're based on isn't perfect either.