Tufte Talk: Chapter 8

We’re down to the last two chapters now. I’ll likely finish both of these this week, as tomorrow I’m starting a new position and won’t have much time to be slowly perusing the pages of esoteric data theory, as I’ll likely be neck deep in more technical aspects of my new career.

Chapter eight is kind of a speed bump chapter, if you ask me. I’m not saying I feel ungrateful for the topics covered in this chapter, it just feels short and strangely vague and super specific at the same time. The first half is about convincing you that Humans really can get a lot out of a dense visualization, or out of a sparse but tiny visualization. Then it goes in depth about Small Multiples explicitly, giving you a nice little matrix for doing those, specifically.

I mostly trust that Tufte is right when he gives his formula for data density in a graphic. I don’t really feel like there’s much to debate there, save, maybe, how useful such an empirical approach to weighing the “within reason” part of his maxim: “Maximize data density and the size of the data matrix, within reason.”

So let’s assume that, yes, it serves you well to increase the information density of your data graphic. Let’s assume it’s also an easy principle to say, “Hey, I can shrink this down and people can still read it and get it.”

My questions - and I’m seriously asking this; like, feel free to populate the comments section with dense data - Can’t this principle also serve the purposes of data malfeasance?

I can easily see how excluding data that communicates context is important. But what if someone takes a simple set of data points with a clear insight and clutters them with a dense rendering of additional information? Can’t you use data density to obscure insight? Or, more simply, to make the effort of deciphering what you’re trying to say greater than it’s worth?

And when it comes to shrinking the graphic - What results would this have on the reader? I want more context! I mean, sure, shrinking a data graphic for a pocket-sized handbook used by a technician or a medical worker, perhaps, is a great way to put faith in your viewer. But if you shrink down the graphic that tells an interesting point in an article, doesn’t that imply more that the information visualized isn’t as important as the words telling you what to infer? Shrinking a graphic could be a technique for making it easier to casually Not Actually Look At The Damn Thing.

I’m not being snarky when I ask these questions - please, tell me if I’m missing the point here?

As to Tufte’s axioms on Small Multiples - I can dig this. It’s clear, concise, and useful. My own spin is as follows:

• A good Small Multiple makes it instantly easy to compare the iterations
• A good Small Multiple organizes all the variables in an easy hierarchy
• A good Small Multiple is small enough to see what you need to, and dense enough to learn what you should from it
• A good Small Multiple uses enough data to be rich and insightful
• A good Small Multiple doesn’t waste a lot of space
• A good Small Multiple makes the comparison you’re doing efficient
• A good Small Multiple relates a story quickly and easily

Then T.S.Tufte hits us with a nice lyric:

“For non-data-ink, less is more.

For data-ink, less is a bore.”

You can use that the next time you have to prove a data graphic has value in a rap battle. Drop the mic.

The last chapter is about Aesthetics and Technique, and that has me more curious than any other chapter. I hope I have something fun to say before this is all over. I’m glad I finally got through this book, even if I had to navigate all the goodness in my spare moments.

But, seriously. Answer my questions from above.