The author emphases the importance of using intuitive colors when creating visualization projects, which I have found more difficult when representing qualitative or categorical data using different colors. It seems like our human eyes start to perceive different colors when the number reaches a certain ceiling. I once saw a project work using 48 colors for 48 countries and it caused so much confusion, as I constantly need refer to the legend and remind myself which color corresponds to which country. Besides, as the main accents of colors human can easily recognize and distinguish are the main seven colors (while the nearby colors remain a little difficult to distinct sometimes.), the 48 colors have many that are the same in hue and only different in saturation and/or value, resulting in a great difficulty to distinguish them.
I think colors are really helpful when used to display density or distribution. However, when it turns to represent the categories, it means different value for each of the color, and requires the audience to differ each color from the rest pairwisely. In this case, unless the number of categories is small, I think it is better to avoid the color representation of the categories.