There are 3 possible aspects contribute to the badness of a piece of data visualization discussed in the reading. They are aesthetic, substantive, and perceptual.
The aesthetic parts is mainly about finding the balance between data to ink ratio, meaning the relationship between data and design elements. Excessive and content-free decorations should be cleared off. The design/aesthetic should also take the detail complexity into account while presenting dense information.
The substantive aspect is the problem innate within the data itself. Problem may lies in the categorization of data, the sorting methods, and etc. In such case, no matter how clear the visual design is, the graph would inevitably lead to misunderstanding of the data.
The perceptual aspect mainly discusses how certain visuals reads differently according to human's eye and cognitive psychology. We tends to read more accurately and thus more willingly to read data on a fixed baseline. Our brain process length difference better than angles and areas. Different hues of color would mostly pops out more easily than different shapes. The gestalt rule explains how we read a group of different visuals: we group objects according to their similarity and proximity, and base on the connections we see continuity. All these affect the design choice within the data visualization, for all would help audience to read the graph easier or hinder their understandings.
I agree with most of the points discussed in the reading. I do wonder to what extent do the design elements become decorative and would be counted by the author as "decorative". Also, the purpose of a piece of data visualization should lie in its clarity to convey the data or its emotional effect on the audience.