I liked Tufte’s intro to information design. I thought he synthesized the most important aspects data presentation (bountiful details with the capability for broad overview). Even though this book was published in 1990, the points were still relevant to today. Most of the examples of “good” information design I could have expected to see in the New York Times or some other big publication.

I liked that he brought up exploring multi dimensional data on computer screens. This is a point that feels especially relevant to the field of machine learning and model explainability. I had never seen a pentagonal graph and thought it was super cool. I honestly still don’t really know what it was plotting but had never seen a graph like this.

![Screen Shot 2022-11-06 at 5.53.17 PM.png](/assets/Screen Shot 2022-11-06 at

I was hoping that Tufte would dive deeper into meaningful approaches to simplifying multivariate data. However, he moved onto 3D physical information models and the stereo illustrations of paired images. The 3D models would only make sense for data rooted in 3D physicality (ie solar system coordinates, but not latent variables from machine learning model embeddings). The stereo illustrations felt novel, but I couldn’t even understand the example given. I wasn’t sure how stereo illustrations could be applied to tabular data.

Tufte’s continuation of flattening complex information felt weak addressing the more complex problem of visualizing interconnected high dimensional data. I hope future chapters present more approaches for dealing with abstract multivariate data.