Tufte is Dead; Long Live Tufte
Review of The Visual Display of Quantitative Information
Ok. Stick with me here. I’m going to review a book from the King: Edward R. Tufte. The man who made one of the earliest and most beautiful arguments for why we have to design better visualizations. In this review I am going to take a critical look at this book and try to to set aside the fact that this book should be on every UXer’s bookshelf. I’m going to try to ignore that Tufte printed these books himself because he was unsatisfied with regular pubishers. And, I’m going to ignore the fact that he is a legend within the field. Instead, I am going to present a critical argument regarding the quality of the material presented in the book.
Of all of Tufte’s books, The Visual Dispaly of Quantitative Information is my favorite. This is mostly because I make a whole lot of visualizations based on very large data sets for my job. The visualizations that I design are incredibly powerful and what wake me up in the morning excited to go to work. But, at the same time, getting them wrong is also incredibly frustrating. This book was the first one I read that talked about data visualizations as an art and really spoke about bridging the gulf of a user’s understanding. Here is a great video where Tufte and a few others argue for importance of data visualizations:
If you haven’t read this book, the visualizations are more than tantalizing. The book is also beautifully designed. The pages are cream and the paper has a density that makes you want to touch them. The written content takes up about 2/3 of the page, following the golden ratio for beautiful composition. There are references anchored in the remaining space that are small but still engaging. And, best of all, most pages have a visualization example to solidify the argument being made. The visualizations are printed in color and sometimes take up the whole page so that you can see all of the small details (even in the visualizations that Tufte argues are poor).
The book is divided into two sections: Part 1 that focuses on the history of quantitative visual design and why it matters; and, Part 2 that focuses on why graphs and charts really are an art but that there are some guidelines we should all be following. The sections flow easily from one to another and the conversational style of writing makes the book one that you can read in an afternoon but also one that you can easily return to multiple times.
Overall, the book is pretty much a masterpeice.
Now, lets put that aside for a second and talk about what Tufte is trying to say here. He is arguing that really good quantitative visualizations provide insight that you cannot get from just looking at the data. The classic example is shown below of Napoleon’s March through Russia and how the march had a severe impact on the amount of troops he has. Sure, you can look at the numbers. You can also look at Napoleon’s route. But, it isn’t until you put those two together than you really understand the magnitude that this march had on the army. The visualization is what enables that deeper level of understanding. Tufte’s argument on this topic is solid and has been proven time and time again.
However, in this book, Tufte also coins the term “chart junk.” Chart junk is all of the stuff that people add to visualizations to jazz them up. It is all the flare that gets added to a visualization to make them more interesting. Tufts writes:
When a graphic is taken over by decorative forms or computer debris, when the data measures and structures become Deisgn Elements, when the overall design purveys Graphical Style rather than quantitative information, then that graphic may be called a duck in honor of the duck-form store, “Big Duck.”
How funny is that? That building is a duck. And, it is a great metaphore. It really quacks me up.
Once the reader hits the Big Duck stage of the book, the argument rolls downhill into something that would make the Bauhaus Art School look like they are full jazz and whimsy. Tufte makes the case that everything that isn’t about presenting the data should be removed. No more bars in a bar chart. You only get lines that reach a certain height! No more axis. You get empty space! Everything must be stripped down to the bare essentials so as to only represent the data and only the data.
I like to say that a Tufte visualization is like soylent of visualizations: all function and little form.
Here is the problem with Visualizations… most of them are so stupid boring. They are so stupid boring that no one wants to pay attention to what you are trying to tell them. People want the data to sing. The problem is that sometimes the data is just humming its story rather than singing like they are Bey. All data cannot be Bey.
There is an argument to be made for chart junk. Maybe if you add a little chart junk users are going to pay attention to what you are trying to say. Maybe a little chart junk helps turn up the beat a little. Now I’m not saying that everyone should make their visualizations like this ass hat in the video below making a PowerPoint presentation. But I am saying that a little color doesn’t hurt. Axis can be your friend. Two dimensions can help fill out the story you are trying to tell.
The last thing I want to say about this book is that all of the visualizations in this book are static. The users cannot play with the data and that is a real shame. Because, playing with the data in a dynamic set of visualizations that allow for on-click filtering is the future. As a field, we really need to spend some time thinking more about how to make the data tell a story rather than assuming the user can click to get the insight they need. So far, I haven’t found the Tufte of dynamic visualizations, even though I’m a big fan of Stephen Few’s work.
Ultimately, quantitative data visualizations are powerful. Tufte is still king. But, his work is a bit dated and draconian to be practical for mundane datasets.
Book Club Questions
- How do you decide what is chart junk and what isn’t?
- Death by PowerPoint isn’t a kind way to kill anyone. How can you use the principles to make your presentations and visualizations stronger?
- What is your most egregious use chart junk?
- How can we translate Tufte’s argument into something useful for interactive quantitative visualizations?
